Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The long-term history of plant communities on southeastern Vancouver Island based on vegetation resurveys… McCune, Jenny Lyn 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_spring_mccune_jenny.pdf [ 7.42MB ]
Metadata
JSON: 24-1.0165697.json
JSON-LD: 24-1.0165697-ld.json
RDF/XML (Pretty): 24-1.0165697-rdf.xml
RDF/JSON: 24-1.0165697-rdf.json
Turtle: 24-1.0165697-turtle.txt
N-Triples: 24-1.0165697-rdf-ntriples.txt
Original Record: 24-1.0165697-source.json
Full Text
24-1.0165697-fulltext.txt
Citation
24-1.0165697.ris

Full Text

THE LONG-TERM HISTORY OF PLANT COMMUNITIES ON SOUTHEASTERN VANCOUVER ISLAND BASED ON VEGETATION RESURVEYS AND PHYTOLITH ANALYSIS  by Jenny Lyn McCune  MSc, The University of Kent, 2004 BSc, The University of Guelph, 2000  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Botany)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2013  ? Jenny Lyn McCune, 2013   iiAbstract Terrestrial plant communities are complex systems, and major storehouses of global biodiversity.  The composition of a plant community today is contingent on conditions that have occurred throughout its history.   Therefore, an understanding of current plant community structure requires an understanding of its origin and its variability over time.  In this thesis, I investigate the history of plant communities on southeastern Vancouver Island, Canada, at two time scales.  These communities are now highly fragmented and threatened by human disturbance, but they also have a long history of management by indigenous peoples.    To quantify changes due to recent urbanization, I resurveyed 184 vegetation plots originally surveyed in 1968.  I documented striking increases in plot-level and total species richness, but a decline in the variation in plant community composition between plots, a phenomenon called ?biotic homogenization?.  Exotic species were more likely than natives to increase over time, but exotic colonizations were not correlated with biotic homogenization or native declines. Plant life history traits predicted colonizations based on landscape context within 500m of a plot, but extirpations were rare and much less predictable, suggesting time lags in plant community response to landscape disturbance and fragmentation, and a potential extinction debt.  I used plant microfossils called phytoliths extracted from soil to investigate changes in plant communities prior to European settlement.  I established that the ratio of asterosclereid phytoliths produced in the needles of Douglas-fir (Pseudotsuga menziesii) to the rondel phytoliths produced by most grasses can accurately distinguish between Douglas-fir dominated forests and Garry oak (Quercus garryana) savannah habitats today.  I then examined changes in this ratio with depth at   iiiseven local sites, finding that infilling by Douglas-fir forest first began at different times, depending on the site.  However, some savannah sites have supported grassy vegetation for at least two thousand years.  Active management to maintain open conditions will be necessary to preserve rare species that evolved in these conditions.    These investigations demonstrate that examining the history of plant communities can reveal surprises and challenge assumptions about how they respond to disturbance.  This knowledge can improve ecological theory, and inform management and conservation strategies.                   ivPreface The design, execution and analysis of this work were led by the author with advice and assistance from members of my supervisory committee.  Versions of three chapters have been published.    Chapter 2 is a literature review.  McCune, J.L., M.G. Pellatt, and M. Vellend. 2013. Multidisciplinary synthesis of long-term human-ecosystem interactions:  a perspective from the Garry oak ecosystem of British Columbia.  Biological Conservation 166:293-300.  I wrote the manuscript with editorial advice from my co-authors.  Helpful comments were also provided by Dr. A.R.E. Sinclair.  Chapter 3 has been published.  McCune, J.L., and M. Vellend. 2013. Gains in native species promote biotic homogenization over four decades in a human-dominated landscape.  Journal of Ecology 101:1542-1551.  I collected the 2009 data, and Dr. Mark Vellend assisted with data analysis and writing.  The 1968 data were collected by Dr. Hans Roemer.  The measurements of specific leaf area were conducted by Dr. Will Cornwell and kindly shared for this analysis.  Chapter 5 has been published.  McCune, J.L. and Pellatt, M.G. 2013. Phytoliths of southeastern Vancouver Island, Canada, and their potential use to reconstruct shifting boundaries between Douglas-fir forest and oak savannah.  Palaeogeography, Palaeoclimatology, Palaeoecology 383-384:59-71.  I collected most of the data, with assistance from Dr. Pellatt in obtaining soil cores.  I carried out the analyses.  Dr. Pellatt provided editorial advice on writing the paper.   vTable of Contents  Abstract.......................................................................................................................................... ii Preface........................................................................................................................................... iv Table of Contents ...........................................................................................................................v List of Tables ................................................................................................................................ xi List of Figures............................................................................................................................. xiii Acknowledgements ................................................................................................................... xvii Chapter  1: Introduction ...............................................................................................................1 1.1 Thesis chapters................................................................................................................ 9 Chapter  2: The long-term history of human-ecosystem interactions in the Garry oak ecosystem of British Columbia ...................................................................................................13 2.1 Synopsis ........................................................................................................................ 13 2.2 Introduction................................................................................................................... 13 2.3 The Garry oak ecosystem.............................................................................................. 16 2.4 Ecological research reveals the hidden causes of exotic invasions .............................. 19 2.5 Historical ecology:  fire on the landscape..................................................................... 21 2.5.1 Written historical records...................................................................................... 21 2.5.2 Land survey records.............................................................................................. 23 2.5.3 The tale of the tree-rings ....................................................................................... 24 2.6 Digging deeper:  paleoecology and archaeology.......................................................... 25 2.7 Implications for this and other threatened systems....................................................... 29 2.8 Conclusion .................................................................................................................... 32   viChapter  3: The role of native species in promoting biotic homogenization over four decades on the Saanich peninsula...............................................................................................33 3.1 Synopsis ........................................................................................................................ 33 3.2 Introduction................................................................................................................... 33 3.3 Materials and methods .................................................................................................. 37 3.3.1 The study area ....................................................................................................... 37 3.3.2 Resurvey methods................................................................................................. 39 3.3.3 Statistical analyses ................................................................................................ 40 3.4 Results........................................................................................................................... 44 3.5 Discussion..................................................................................................................... 48 3.6 Conclusion .................................................................................................................... 55 Chapter  4: Using plant traits to predict the sensitivity of colonizations and extirpations to landscape context .........................................................................................................................56 4.1 Synopsis ........................................................................................................................ 56 4.2 Introduction................................................................................................................... 57 4.3 Materials and methods .................................................................................................. 60 4.3.1 The study area ....................................................................................................... 60 4.3.2 Resurvey methods................................................................................................. 62 4.3.3 Landscape data...................................................................................................... 63 4.3.4 Species and environmental data............................................................................ 65 4.3.5 Trait data ............................................................................................................... 66 Origin and disturbance tolerance ..................................................................................... 67 Longevity and form ........................................................................................................... 68   viiShade tolerance................................................................................................................. 68 Nutrient regime preference ............................................................................................... 69 Dispersal ability................................................................................................................ 69 Specific leaf area............................................................................................................... 70 Trait correlations .............................................................................................................. 71 4.3.6 Statistical analyses ................................................................................................ 71 4.4 Results........................................................................................................................... 76 4.4.1 Overall changes in diversity.................................................................................. 76 4.4.2 Landscape changes................................................................................................ 76 4.4.3 Relative influence of local environment and surrounding landscape on community composition........................................................................................................................... 77 4.4.4 Relating traits to presence, colonizations, and extirpations.................................. 79 4.5 Discussion..................................................................................................................... 81 4.5.1 The changing landscape........................................................................................ 81 4.5.2 Predictability of colonizations and extirpations.................................................... 88 4.6 Conclusion .................................................................................................................... 91 Chapter  5: Phytoliths of southeastern Vancouver Island .......................................................93 5.1 Synopsis ........................................................................................................................ 93 5.2 Introduction................................................................................................................... 94 5.3 Materials and methods .................................................................................................. 97 5.3.1 Reference collection.............................................................................................. 97 5.3.2 Soil sampling across an ecotone ......................................................................... 106 Soil collection.................................................................................................................. 106   viiiPhytolith extraction......................................................................................................... 107 Phytolith counts .............................................................................................................. 108 5.4 Results......................................................................................................................... 110 5.4.1 Phytoliths found in grasses ................................................................................. 110 5.4.2 Phytoliths found in non-grasses .......................................................................... 113 5.4.3 Soil core analyses................................................................................................ 117 5.5 Discussion................................................................................................................... 120 5.5.1 Documenting changes in savannah understorey composition ............................ 120 5.5.2 Asterosclereids as indicators of the presence of Douglas-fir.............................. 122 5.6 Conclusion .................................................................................................................. 123 Chapter  6: Combining phytolith analysis with historical ecology to better understand the long-term dynamics of an endangered ecosystem...................................................................125 6.1 Synopsis ...................................................................................................................... 125 6.2 Introduction................................................................................................................. 126 6.3 Materials and methods ................................................................................................ 132 6.3.1 Site selection ....................................................................................................... 132 6.3.2 Phytolith extraction............................................................................................. 136 6.3.3 Phytolith counts .................................................................................................. 137 6.3.4 Analyses.............................................................................................................. 139 6.3.5 Radiocarbon dating ............................................................................................. 140 6.4 Results......................................................................................................................... 140 6.4.1 Surface calibration .............................................................................................. 140 6.4.2 Full core analysis ................................................................................................ 144   ixCurrent savannah sites ................................................................................................... 145 SOM2 transition site ....................................................................................................... 145 SOM3 and COW14 ......................................................................................................... 147 Long-term forested sites.................................................................................................. 148 Burnt phytoliths and morphotype proportions with depth .............................................. 156 6.5 Discussion................................................................................................................... 158 6.5.1 The landscape before 1859 ................................................................................. 158 6.5.2 Evidence for changing abundance of native bunchgrasses................................. 162 6.5.3 Phytoliths and fire on the landscape ................................................................... 163 6.5.4 The phytolith record in soil as a paleoenvironmental indicator.......................... 164 6.5.5 Outliers in the surface calibration ....................................................................... 168 6.5.6 A possible index for dating the Douglas-fir infilling.......................................... 169 6.6 Conclusion .................................................................................................................. 172 Chapter  7: Conclusion..............................................................................................................174 7.1 Chapters 3 and 4:  the invisible present ...................................................................... 176 7.2 Chapters 5 and 6:  vegetation before European settlement......................................... 178 7.3 General conclusions .................................................................................................... 181 7.4 Study limitations and future research ......................................................................... 183 References...................................................................................................................................188 Appendices..................................................................................................................................222 Appendix A Supplementary data for resurveys (Chapters 3 and 4) ....................................... 222 A.1 Compilation of trait data (Chapter 3)...................................................................... 222 A.2 Indicator species analysis results (Chapter 3)......................................................... 226   xA.3 Comparison of phylogenetically corrected versus uncorrected tests (Chapter 4)... 231 Appendix B Supplementary phytolith data (Chapters 5 and 6) .............................................. 234 B.1 Collection information and incineration data for plant tissue samples................... 234 B.2 Detailed phytolith extraction protocol .................................................................... 245 B.3 Quantification of phytolith loss during extraction .................................................. 252 B.4 Some unknown phytolith morphotypes observed during phytolith counts ............ 254    xiList of Tables Table 3.1: Results of a PERMANOVA with year of survey and vegetation type as fixed variables and plot identity as a random variable .............................................................................. 46 Table 3.2: Results of pairwise comparisons between plots of the same vegetation type in different years in terms of multivariate dispersions using PERMDISP ........................... 48 Table 3.3: Results of cumulative logit model comparisons for the association between indicator status (loser, no change, winner) and traits, and for paired tests of the change in abundance-weighted average trait value per plot.............................................................. 49 Table 4.1: Pearson?s correlation coefficients between landscape context variables measured within 500m of each plot in 1964 and 2005. .................................................................... 65 Table 4.2: Pearson?s correlation coefficients between plant traits for the colonization dataset ... 72 Table 4.3:  The results of distance-based linear models relating environmental variables and landscape context to plant community composition. ........................................................ 78 Table 4.4:  Predicted versus observed relationships between colonizations over 40 years, landscape structure within 500m of the relev?, and species traits. ................................... 83 Table 4.5:  Predicted versus observed relationships between extirpations over 40 years, landscape structure within 500m of the relev?, and species traits..................................................... 84 Table 4.6:  Observed relationships between presence of species in each year, landscape structure within 500m of the relev?, and species traits.................................................................... 85 Table 5.1: Phytolith types and type frequencies found in grass species of southeastern Vancouver Island................................................................................................................................. 99 Table 5.2: Phytolith types and type frequencies in non-grass species of southeastern Vancouver Island............................................................................................................................... 104   xiiTable 6.1:  Location and characteristics of the 25 sites used in the study .................................. 131 Table 6.2:  AMS (accelerator mass spectrometry) radiocarbon and calibrated calendar ages of charcoal or wood samples from soil cores...................................................................... 140 Table A.1: Hypothesized drivers of plant community change, expected shifts in plant traits, and the percentage of species for which data was available for each trait ............................ 225 Table A.2:  Results of indicator species analysis for all 101 understory species (or taxa) present in at least 5% of all plot-year combinations.................................................................... 226 Table A.3:  Comparison between phylogenetically corrected and uncorrected models testing the relationship between traits and the positive or negative association of (a) colonizations, (b) extirpations, (c) presence/absence in 1968, and (d) presence/absence in 2009 with landscape context variables............................................................................................. 231 Table B.1:  Collection information and incineration data for plant tissue samples.................... 234 Table B.2:  Results of quantification of phytoliths found in the supernatant from various stages in the phytolith extraction protocol..................................................................................... 253          xiiiList of Figures Figure 1.1: The location of the study region................................................................................... 3 Figure 1.2:  Schematic representation of the contribution by research in various disciplines to the understanding of long-term vegetation change on southeastern Vancouver Island over different scales of space and time.. ..................................................................................... 6 Figure 2.1: The range of Garry oak in North America and on Vancouver Island is shown in red.  Black triangles are archaeological sites within this range compiled from the Canadian Archeological Radiocarbon Database............................................................................... 17 Figure 2.2: (a) Garry oak savannahs consist of an open canopy of oak trees with an understory of grasses and wildflowers; (b) Here the understory is dominated by exotic annual grasses and the exotic shrub Scotch Broom (Cytisus scoparius); (c) An aerial photo of Summit Park, in the city of Victoria, which contains a reservoir and a fragment of Garry Oak savannah completely surrounded by urban development.. ............................................... 18 Figure 2.3:  Changing pollen abundance for Western red-cedar and Garry oak over the past 10,000 years, as a percentage of the maximum relative abundance for each species, from the Saanich Inlet sediment core.  Bars show the age frequency of radiocarbon-dated materials from 71 archaeological sites within the range of Garry oak ecosystems on Vancouver Island.. ............................................................................................................ 26 Figure 3.1:  The study area.  Relev? plot locations are indicated by black dots........................... 36 Figure 3.2: Plots of (a) total understorey species richness, (b) total native understorey species richness, and (c) total exotic understorey species richness in 2009 versus 1968. (d) shows the number of species gained and lost per plot by origin. ................................................ 45   xivFigure 3.3: Non-metric multidimensional scaling ordination of all plots in 1968 and 2009 in understorey species space ................................................................................................. 47 Figure 3.4: The relationship between the change in distance from the overall multivariate species centroid over time and (a) change in number of exotic species present, and (b) change in number of native species present from 1968 to 2009 ....................................................... 50 Figure 3.5: The percent of species in each category for the four traits that were found to be strongly associated with indicator status........................................................................... 52 Figure 4.1:  Inset: location of the Saanich Peninsula on southeastern Vancouver Island.  Centre: The Saanich Peninsula extends north from the city of Victoria.  Black dots indicate elev? plot locations.  Right: Aerial photographs from 1964 and 2005 show landscape context within 500m of a relev? plot. ............................................................................................ 61 Figure 4.2:  Change over 40 years in (a) total length of roads within 500m of each plot, (b) percent of area naturally vegetated within 500m of each plot, and (c) shape index......... 77 Figure 4.3:  The percentage of species in different trait categories with a positive association between colonizations and shape index in 1964.. ............................................................. 86 Figure 4.4:  The percentage of species in different trait categories with a positive association between colonizations and shape index in 2005............................................................... 87 Figure 5.1: Main map: the West coast of North America showing Vancouver Island, with the range of Garry oak indicated in grey. Inset: Southeastern Vancouver Island and the Gulf Islands, showing Tumbo Island, the location of the soil samples. Lower right: the boundary between oak savannah and Douglas-fir forest at the soil sampling location ... 95 Figure 5.2: Phytolith morphotypes found in grass species ......................................................... 112 Figure 5.3: Phytolith morphotypes found in non-grass species.................................................. 117   xvFigure 5.4: Asterosclereid phytolith extracted from the soil core located within the Douglas-fir-dominated forest on Tumbo Island. ................................................................................ 118 Figure 5.5: Estimated total number (in thousands of phytoliths) of (a) asterosclereids, (b) elongate phytoliths, (c) rondels and (d) bilobates per gram of dry soil plotted against depth in the soil core ....................................................................................................... 119 Figure 5.6: Percentage of total estimated number of phytoliths per slide represented by (a) asterosclereids, (b) elongate phytoliths, (c) rondels and (d) bilobates plotted against depth in the soil core................................................................................................................. 120 Figure 6.1:  Map of southeastern Vancouver Island showing the location of soil samples (red dots), and the estimated historical range of Garry oak ecosystems ................................ 132 Figure 6.2:  Location of soil cores taken from the Somenos Garry Oak Preserve showing the landscape in 1859 and today. .......................................................................................... 135 Figure 6.3: Examples of the five phytolith morphotypes counted in the study .......................... 138 Figure 6.4:  Non-metric multidimensional scaling ordination of all soil sample sites based on current vascular plant species composition within a 20x20m plot. ................................ 141 Figure 6.5:  Differences in total phytoliths per gram of surface soil estimated for each phytolith morphotype by vegetation type....................................................................................... 142 Figure 6.6:  Differences in the percentage of all phytoliths represented by each morphotype in surface samples by vegetation type................................................................................. 143 Figure 6.7:  (a) The log ratio of asterosclereid to rondel phytoliths in surface samples by vegetation type,  (b) The log ratio of asterosclereid to rondel phytoliths in surface soil samples plotted against the difference in current percent cover of Douglas-fir and all grasses ............................................................................................................................. 146   xviFigure 6.8:  The SOM1 core. ...................................................................................................... 149 Figure 6.9:  The NCC2 core........................................................................................................ 150 Figure 6.10:  The SOM2 core ..................................................................................................... 151 Figure 6.11:  The SOM3 core. .................................................................................................... 152 Figure 6.12:  The COW14 core................................................................................................... 153 Figure 6.13:  The COW1 core..................................................................................................... 154 Figure 6.14:  The COW2 core..................................................................................................... 155 Figure 6.15:  Changes in the proportion of all phytoliths represented by (a) elongate phytoliths, (b) rondels, (c) asterosclereids, and (d) bilobates with depth in the first 25cm from the surface of all cores .......................................................................................................... 157 Figure 6.16:  Potential relationship between the concentration of asterosclereids in surface soil phytolith assemblages, and the date of encroachment by Douglas-fir............................ 171 Figure B.1: Final phytolith extraction protocol used to extract phytoliths from soil samples.... 251 Figure B.2:  Photos of ?unknown? phytoliths or potential phytoliths. ....................................... 256    xviiAcknowledgements I would like to thank my advisor, Mark Vellend, for trusting me to direct my own research while providing prompt and thoughtful responses when I needed advice, and for many interesting conversations about ecology and conservation.  Marlow Pellatt signed on as my paleoecology mentor, and generously provided laboratory space, equipment, advice, and pep talks during my adventures with phytoliths.  This more than made up for his tendency to miss (or nearly miss) ferries during fieldwork.  The other members of my multidisciplinary committee provided insight from their respective fields.  Mike Blake helped me delve into the archaeological literature, advised me on radiocarbon date calibration, and introduced me to archaeologists interested in the historical interaction between indigenous peoples and terrestrial ecosystems on Vancouver Island.  Gary Bradfield stepped in to help with supervisory duties after Mark moved away, and gave me the opportunity to help teach his fourth year plant ecology class.  Tony Sinclair shared his expertise in historical ecology.  I would like to thank my past academic and professional mentors, who helped make me a better researcher, ecologist, and natural historian:  John Klironomos at the University of Guelph, Suzanne Koptur and Chris Borg in Florida, Mark Nesbitt at Kew Gardens, Denise Knapp and Lisa Stratton on Catalina Island, and Kathy Fleming and Jonathan McKnight at the Maryland Department of Natural Resources.  This research would not have been possible without the assistance of Dr. Hans Roemer.  Hans provided access to the original notes and data from his 1968-69 study of the vegetation of the Saanich peninsula, which allowed me to resurvey a number of his plots.  I am grateful to Hans for many hours of assistance at his kitchen table and in the field, for helping me to identify numerous plants in person and via email, and for challenging me to think about what the results mean in terms of specific species.  Hans also provided several plant samples for my phytolith reference collection.  My fellow graduate students were a source of support, advice, and amusement throughout my five years at UBC.  Thanks especially to my labmates Tanis Gieselman, Emily Drummond, Anne Bjorkman and Heather Kharouba.  Anne shared the data she compiled from the 1859 land survey of the Cowichan Valley, which was crucial for my phytolith research and its interpretation.  Heather was my go-to statistics guru and my PhD big sister, completing each step just ahead of me and letting me benefit from her experience and advice.   I thank the following for permission to access land:  BC Parks, District of Saanich, District of Central Saanich, District of North Saanich, Capital Regional District, City of Victoria, Royal Roads University, Herzberg Institute of Astrophysics, Royal Oak Burial Park, Victoria International Airport, University of Victoria, the Pauquachin Nation, Nature Conservancy of Canada, Cowichan Tribes, Cowichan Valley Regional District, Providence Farm, and several private landowners.   xviii Eric McLay introduced me to the Somenos archaeological site near Duncan, gave me a personal tour of the site, and provided his expertise on the archaeology of terrestrial resource use in the region.  He also provided camas samples for my phytolith reference collection.  Eric, Nicole Smith, Iain McKechnie, and Dana Lepofsky provided advice on the archaeology section of Chapter 2. Thanks to Andrew MacDougall for his advice on where to sample soil at the Cowichan Garry Oak Preserve.  Judith Harpel helped me identify moss samples.  The staff at the Geographic Information Centre at UBC helped me obtain 1964 aerial photos of the Saanich peninsula.  Dr. Will Cornwell kindly shared his SLA database for plants of the region. Todd Golumbia and Rob Walker of the Gulf Islands National Park Reserve arranged boat transportation to Tumbo Island.  Carly Ziter and Rachel Germain helped collect plant samples for my phytolith reference collection.  Rand Evett directed me to important publications on asterosclereid abundance, Anka Marsh gave advice on the identification of phytoliths found in Agrostis and Digitaria, and Heloisa Coe provided helpful information about phytolith movement in soil.  Simon Smart provided comments on an earlier version of Chapter 3.  I could not possibly have collected all the data I did without the help of my three undergraduate research assistants in the lab and the field:  Anthony Ho, Alice Cang, and Alejandra Canela.    I thank my mentors in teaching at UBC, Lacey Samuels and Joanne Fox, who inspired me with their skill and enthusiasm for teaching and learning.  Funding for this research was provided by an NSERC Canada Graduate Scholarship, an NSERC BRITE Graduate Fellowship through the Biodiversity Research Centre at UBC, a UBC Four Year Fellowship, an NSERC Discovery Grant to Mark Vellend, and a grant from the Pacific Institute for Climate Solutions to Marlow Pellatt.  Parks Canada provided laboratory space, equipment, and boat transportation.  Finally, I would like to thank my family ? especially Gracia, Wayne, and Nancy McCune ? for getting me through the hard parts.      1Chapter  1: Introduction A plant society is not a product of present conditions alone, but the past is involved as well.     ~ H.C. Cowles, 1901  The fundamental question that plant ecology seeks to answer is:  what factors determine plant community composition, diversity, and change?   Most ecologists study the role of current conditions including soil type, climate, and interactions with competitors, herbivores, and mutualists; and changes in these factors over the short-term (i.e. years).  However, changes in these factors over a much longer timeframe (i.e. decades to millennia) also can influence the structure and function of plant communities we observe in the present.  Human societies, as major consumers of plant products and agents of disturbance, may affect these conditions both in the present and in the past.  The importance of history in the composition of plant communities was well recognized by early plant ecologists (e.g. Cowles, 1901).  More recently the notion of taking a long-term view in order to understand ecological change has been reinvigorated by the establishment of the Long Term Ecological Research (LTER) program in the U.S. in 1980 (Hobbie et al., 2003), and by the influence of paleoecological research on the understanding of ecological problems (e.g. Davis, 1994).  Extending the timeframe of ecological understanding provides needed context, both in developing ecological theory and in attempts to restore plant communities heavily impacted by recent (e.g. decades to centuries) human activities (Egan and Howell, 2005; Jackson, 2006).  As managers of organisms and systems much longer lived than ourselves, we lack the ability to directly perceive changes that occurred centuries before we were born, or even changes that   2occur within our lifetimes but in such a gradual way as to be invisible to us (Magnuson, 1990).  The challenge is to integrate information collected from different disciplines on different timescales to create a coherent picture of plant community change over time.    A second challenge is to understand how human actions have interacted with other biotic as well as climatic factors to influence vegetation change.  While early ecologists acknowledged the ability of modern humans to disrupt or redirect pathways of plant succession following natural or human-caused disturbance (e.g. Cowles, 1911), pre-modern human societies outside of Europe were deemed too primitive to have much impact on their environment (Kirch, 2005).  This has turned out to be wrong in many cases, with evidence from paleoecology and archaeology that human populations outside of Europe had measurable and sometimes very large impacts on local and even regional vegetation long before Europeans arrived (e.g. Athens and Ward, 1993; Bayliss-Smith et al., 2003; Delcourt and Delcourt, 2004; Lentfer and Torrence, 2007; Pyne, 1995; Redman, 1999; Rival, 2006; Willis et al., 2004).  The history of pre-modern human impacts on ecosystems may provide important lessons to us in our modern understanding of how human actions influence the environment.  However, past human impacts are difficult to disentangle from climatic effects (Deevey, 1969; Lepofsky and Lertzman, 2008).  Doing so requires a good understanding of the society in question, its technology, settlement patterns, and resource use, as well as a grasp of climatic changes that have occurred over various temporal scales.  This thesis aims to fill in the gaps in current understanding of the long-term history of the plant communities of southeastern Vancouver Island and the adjacent Gulf Islands in southwestern   3British Columbia, Canada (Fig. 1.1).  This region is found within the Coastal Douglas-fir Biogeoclimatic Zone of British Columbia, which represents less than 0.5% of the area of the province (Meidinger and Pojar, 1991).  It is located in the rainshadow of the Olympic and Vancouver Island mountains, which results in drier conditions than those found anywhere else along the coast.  The climate is described as sub-Mediterranean, with mild winters and long, dry summers (Meidinger and Pojar, 1991).  This unique climate in turn supports unique vegetation types, including the Coastal Douglas-fir forests for which the zone is named, and Garry oak ecosystems (Egan, 1999).  Figure 1.1: The location of the study region, southeastern Vancouver Island and the Gulf Islands, including the city of Victoria.   4Coastal Douglas-fir forests are dominated by Douglas-fir (Pseudotsuga menziesii), with components of western red-cedar (Thuja plicata), grand fir (Abies grandis), red alder (Alnus rubra) and bigleaf maple (Acer macrophyllum).  In the driest forests, Pacific madrone (Arbutus menziesii) grows interspersed with Douglas-fir (Egan, 1999; Flynn, 1999).  Coastal Douglas-fir forests have been heavily logged over the past 150 years, and only 0.5% of the original 220,000 hectares of coastal Douglas-fir forest remains as relatively undisturbed old forest (Flynn, 1999).  The Garry oak ecosystem includes a range of vegetation types from open, nearly treeless prairie to oak woodland (Fuchs, 2001).  The archetype of this ecosystem is the Garry oak savannah, which typically consists of an open canopy of oak (Quercus garryana) with an understory dominated by native wildflowers and grasses.  It is estimated that over 90% of Garry oak savannah has been lost to agriculture, development, and infilling by conifers since European settlement of the region in the mid-1800s (Lea, 2006).  The Garry oak ecosystem is now recognized as one of Canada?s most at-risk terrestrial ecosystems, with over 100 associated threatened species (Fuchs, 2001).    Both Garry oak savannahs and Douglas-fir forests are facing continued threats from human disturbance.  The human population of the region has more than doubled in the past four decades.  Urbanization, in particular surrounding the city of Victoria, is threatening the remaining natural habitat (Fuchs, 2001).  Therefore, these vegetation types are of high conservation concern on a provincial and national level, and there are ongoing efforts to protect and restore them (e.g. GOERT, 2011; Parks Canada Agency, 2006a, 2006b).  However, they are   5also a product of centuries and millennia of climatic changes, and human influence.  This region has a long history of human occupation, with higher pre-European human densities than anywhere else in Canada (Ames and Maschner, 1999; Duff, 1997; Matson and Coupland, 1995).  An understanding of this long-term history is important in order to understand the conditions under which these now-threatened vegetation types originated and their historical range of variation.    Research from multiple disciplines has contributed to the current understanding of the history of these systems over different scales of time and space (Fig. 1.2).  Many ecological studies (on timescales of 1-10 years and small spatial scales) have studied the impact of introduced exotic species on plant communities (e.g. Gonzales and Arcese, 2008; Lilley and Vellend, 2009; MacDougall and Turkington, 2005).  These studies show that the root causes of exotic spread are habitat fragmentation and changes in herbivory and disturbance regimes, rather than competitive dominance by exotic species.  Historical ecological studies have used land survey records, the written accounts of early European explorers and settlers, aerial photographs, and tree rings to reconstruct changes over the past few centuries (Bjorkman and Vellend, 2010; Gedalof et al., 2006; Lea, 2006; MacDougall et al. 2004; Smith, 2007).  For example, comparison of current forest conditions with the first land survey records has shown that the density of trees on the landscape has more than doubled since European settlement (Bjorkman, 2008).  Long-term (10,000 years and more before present) palynological studies have documented changes in plant communities on a broad spatial scale since the last glacial maximum (Brown et al., 2008; Hebda, 1995; Pellatt et al., 2001; Whitlock, 1992).   These studies show that, across the region as a   6whole, the climatic conditions and vegetation formations characteristic of the Coastal Douglas-fir Zone today were in place by approximately 3,800 years before present (Pellatt et al., 2001).    Figure 1.2:  Schematic representation of the contribution by research in various disciplines to the understanding of long-term vegetation change on southeastern Vancouver Island over different scales of space and time.  The areas outlined with dashed lines represent the gaps addressed in Part 1 (Chapters 3 and 4) and Part 2 (Chapters 5 and 6) of this thesis.   Two major gaps remain in the understanding of long-term vegetation history in this region (Fig. 1.2).  First, there is little quantitative data on changes in plant communities on a local scale over the past few decades.  The perception of scientists and many others is that human impacts in this region over recent decades have caused a dramatic increase in exotic species and resulted in   7plant communities that are more homogeneous across the landscape (e.g. Fuchs, 2001).  But the extent of this proliferation of exotic species has been difficult to quantify.  In Part 1 of this thesis, I examine changes in plant communities over the past 40 years.  I use detailed plant community data from the late 1960s and the present day to quantify changes in diversity and composition of forest and savannah understoreys with accelerated urbanization.  Part 1 addresses the following broad question:  What have been the consequences of recent urban expansion on plant community diversity and composition, and are these consequences predictable based on plant traits and the landscape configuration surrounding individual sites?  Second, we do not know what plant communities looked like or how variable they were on smaller spatial scales before Europeans arrived.  Historical ecological research has shown that open oak savannahs were much more prevalent at the time of the first land surveys, and that tree density is more than double in the remaining forest (Bjorkman and Vellend, 2010; Lea, 2006).  However, it is not known whether this snapshot of conditions immediately prior to European settlement represents a relatively stable state, or if the extent of more open vegetation types on the landscape varied with climatic and/or human influences prior to the first land surveys.  In Part 2 of this thesis, I test the use of phytolith analysis to determine the range of variability in vegetation on a local scale over the last few millennia prior to European settlement.  Part 2 addresses the following questions:  Can phytoliths from surface soils distinguish between current vegetation types?  If so, can they be used to learn more about the historical range of variation in forest density at local sites prior to the first land surveys?  I address these questions by investigating the utility of the phytolith record in soil profiles to reflect known changes in vegetation at local sites since European settlement.  My goal was to extend our understanding of   8what the landscape looked like before European colonization from a static, pre-settlement snapshot to a long-term trajectory, providing context in terms of targets for ecological restoration.  The motivation for my research is two-fold.  First, I aim to contribute to answering questions directly relevant to the conservation of Douglas-fir forests and Garry oak savannahs on southeastern Vancouver Island.  For example, in Part 1 I aim to determine which species have been most vulnerable to recent changes on the landscape, and which have been most successful.  This information is necessary in order to prioritize conservation efforts, and also to predict the changes in plant communities that are likely to occur with further increases in urbanization, fragmentation, and human disturbance.  In Part 2, I aim to increase our knowledge of the variability of open savannah versus closed canopy conditions at particular sites prior to the first written records.  This information does not dictate which vegetation state should be the target for restoration, but provides context by showing what past vegetation states have existed on a site, and what climatic or human-influenced factors may have contributed to the maintenance of a vegetation state, or a shift to another state.  Secondly, I aim to contribute towards answering fundamental questions in plant ecology.  For example, in Part 1 I test the prevailing theory that the spread of exotic species across a landscape leads to homogenization of plant communities.  I also compare the ability of plant traits to predict colonizations versus extirpations based on surrounding landscape conditions.  Long-term data is necessary in order to answer these questions because plant community dynamics are often   9?slow?, requiring decades or longer for the consequences of exotic colonization or landscape disturbance to be manifested in plant community diversity or composition.  1.1 Thesis chapters Chapter 2 highlights the unique Garry oak ecosystems of southeastern Vancouver Island and the Gulf Islands.  I provide a thorough review of the research from diverse disciplines including ecology, paleoecology, ethnography and archaeology that has contributed towards the current understanding of the long-term dynamics of this system.  This chapter highlights the surprises revealed by each discipline that challenged former ideas of how the ecosystem functions.  I argue that the Garry oak ecosystem is just one example that demonstrates how bringing together research from multiple disciplines is necessary in order to understand the long-term history of the plant communities that occupy a particular landscape.  I also highlight the two main temporal gaps in our understanding of vegetation history in this region, which my research sought to fill.  Chapter 3 introduces Part 1 of my research, a resurvey of 184 vegetation plots on the Saanich peninsula originally surveyed in the late 1960s by Dr. Hans Roemer.  I use a comparison of these plots in 1968 and 2009 to test for changes in plant understorey diversity, both in terms of plot-level species richness and total species richness across all 184 plots.  I found striking increases in plot-level richness of both exotic and native species, as well as an increase in the total number of species across all plots.  I hypothesized that the spread of exotic species would lead to a reduction in the variation in understorey plant species composition across plots between the two years, a phenomenon known as biotic homogenization.  However, the results show that the spread of exotic species across the landscape is not driving this reduction in variability across   10plots.  I also test which plant traits are most common amongst the species that were the most successful in terms of increased frequency and abundance over the four decade period.   Chapter 4 continues the analysis of the data compiled in Chapter 3 by testing for links between changes in plant community diversity, the apparent colonization or extirpation of species with particular traits, and the amount of disturbance of the landscape immediately surrounding each plot.  The ability to use species traits to predict which species will thrive in conditions of disturbance and fragmentation and which species are vulnerable to extirpation is critical in order to successfully conserve plant diversity.  The trait-based response of plant species to direct disturbance is well established, but links between plant traits and the landscape context in which a plant community is located are more tenuous.  I use aerial photographs of the Saanich peninsula from 1964 and 2005 to measure the changes in landscape context within 500m of each relev? plot, including total length of roads, total area of natural vegetation, and the shape index of the patch (an index of area:perimeter ratio).  I then use these data to test the hypothesis that landscape context has become a more important determinant of plant community composition now than forty years ago.  I also test whether plant life-history traits can predict the association of colonization and extirpation with landscape context.  I predicted that plants having traits associated with tolerance for disturbed, fragmented conditions (e.g. disturbance tolerance, shade intolerance, high specific leaf area) would be more likely to colonize, and less likely to be extirpated from, plots with high levels of surrounding road density, low levels of naturally vegetated area, and high edge:area ratio.    11Chapter 5 introduces Part 2 of my research, in which I explore the potential use of phytoliths extracted from soils to indicate shifts in vegetation in the centuries prior to the settlement of Vancouver Island by European peoples.  Phytoliths are plant microfossils formed when hydrated silicon dioxide (SiO2?nH20) is deposited within and between the cells of a living plant (Pearsall, 2000; Prychid et al., 2003).  The resulting cell-shaped silica casts are released upon decomposition of the plant, and can survive in sediments for thousands of years (e.g. Blinnikov et al., 2002).  Not all plant species produce phytoliths, and not all phytoliths are diagnostic to species or even to genus or family.  Therefore, the first step in using soil phytolith assemblages to reconstruct past changes in vegetation is to create a reference collection to show which plants of a region produce phytoliths, and which phytolith shapes are diagnostic of particular taxa.  In Chapter 5, I catalogue the phytolith morphotypes found in 72 of the most common plant species in the study region, and extract phytoliths from two soil cores on either side of the boundary between a Douglas-fir dominated forest and an oak savannah.  Chapter 5 demonstrates that soil phytoliths can reliably distinguish between Douglas-fir forest and Garry oak savannah habitats within 20m of the vegetation boundary, and therefore the phytolith record is a potential tool to document local shifts between Douglas-fir forest and savannah prior to European settlement.    Chapter 6 presents the results of phytolith analysis of 25 surface soil samples and seven complete soil cores taken across the historical range of Garry oak ecosystems on Vancouver Island.  I collected surface soil samples from a broad range of vegetation types, including open Garry oak savannah, former savannah sites in various stages of encroachment by Douglas-fir, and closed-canopy Douglas-fir forests.  I use the phytolith assemblages from these samples to define an index which can distinguish different vegetation types based on the ratio of asterosclereid   12phytoliths (produced by Douglas-fir) to rondel phytoliths (produced by grasses only).  I then examine the changes in this index with depth in the soil cores.  I also test the utility of discoloured phytoliths as an indicator of past fire frequency, and ask whether native bunchgrasses (which are now very rare) were more common on the landscape prior to the introduction of exotic agronomic grasses.   In Chapter 7, I summarize my findings in terms of what I learned about the unique history of the vegetation of southeastern Vancouver Island, as well as the implications for ecological theory and conservation.  I highlight the strengths and limitations of my work, and suggest potential future research.   13Chapter  2: The long-term history of human-ecosystem interactions in the Garry oak ecosystem of British Columbia  2.1 Synopsis Many ecosystems of conservation concern owe their unique characteristics to long-term management by indigenous peoples.  However, there are serious debates concerning the degree and extent of this influence.  In this chapter, I review research from multiple disciplines regarding the long-term history of the Garry oak ecosystem of southern British Columbia.  I use this to demonstrate how delving into the long-term history of a system helps ecologists to understand the key drivers of ecosystem structure and dynamics, including the role of humans.  Considering an extended timeline can also reveal surprises that challenge conceptions of the way an ecosystem functions.   I review the findings of ecological experiments, historical ecology, ethnography, archaeology and paleoecology, and highlight surprises that came out of this research.  I also highlight two gaps in our understanding of the history of this ecosystem, which this thesis aims to fill.  I argue that insights from multidiscliplinary research contribute both to improved ecological theory and better restoration strategies, and show that ecosystems created via long-term human management are equally valid targets for conservation as ecosystems that have experienced less human influence.  2.2 Introduction Ecologists used to believe that the ecosystems of the Americas were ?pristine? prior to European arrival, assuming the indigenous peoples had little or no discernible impact (McCann, 1999).  More and more we realize this belief was false:   indigenous peoples of North and South   14America affected ecosystems significantly (e.g. Delcourt and Delcourt, 2004; Denevan, 1992; Willis et al., 2004).  This realization has required ecologists and conservationists to revise their understanding of the origins and functioning of these ecosystems, and also to question the meaning and goals of ecological restoration (e.g. Allison, 2004).    The challenge is to understand and manage ecosystems that are now greatly fragmented and increasingly dominated by human influence, but that at some point were created and/or maintained by cultural activities.  Ecologists and conservationists are faced with key questions regarding the interrelationship between humans and their environment over the long term (see Box 1).  The challenges in answering these questions are great, but the potential benefits are great too.  A greater understanding of the origins of an ecosystem, its variability over time, and its response to human actions can help us to improve ecological theory, and hence develop more effective management and restoration strategies (Landres et al., 1999).   In this chapter, I present a multidisciplinary perspective on these questions for the threatened Garry oak savannah ecosystem of southeastern Vancouver Island.  Habitat loss and degradation resulting from a rapidly rising human population over the past several decades is currently the dominant threat to this ecosystem.  Yet, an understanding of the history of the ecosystem going back to its beginnings over 8,000 years ago is necessary to fully understand its current structure and function, and therefore how best to achieve conservation goals.  The construction of this long-term history required a synthesis of information from multiple disciplines, and led to some surprises that fundamentally changed our current understanding of this ecosystem.      15  The call for increased consideration by ecologists and conservationists of long-term ecosystem dynamics and incorporation of research from disciplines outside ecology is not new (e.g. Briggs et al., 2006; Dearing et al., 2006; Foster et al., 1990; Foster et al., 2003; Rick and Lockwood, 2013; Smith and Boyer, 2012; Swetnam, 1999; Szabo, 2010; Willis and Birks, 2006; Willis et al., 2007).  However, examples that explicitly weave studies of both historical and contemporary dynamics to yield novel insights are still rare in the ecological literature, in particular examples where the timeline is extended from thousands of years ago all the way to ecological studies on the scale of only a few years.  The Garry oak ecosystem is an excellent case study that demonstrates that consideration of such an extended timeline can (1) challenge assumptions about the state of an ecosystem in the past, (2) suggest or refute hypotheses concerning the causes of current ecological patterns, and (3) help identify gaps in understanding at particular Box 1:  Key questions for understanding human-ecosystem interrelationships (1) How pervasive and necessary was/is human management in the creation and/or maintenance of a particular ecosystem state?  (2) What exactly were/are the human actions involved, their purpose, frequency, and duration? (3) What was/is the relative impact of human versus climatic or other non-human-caused change, and how do they interact?  (4) What was the historical range of variability of the ecosystem?  (5) Can aboriginal land use practices that were important to the maintenance of ecosystem characteristics we value be modified and used as restoration tools today?   16temporal and/or spatial resolutions. This example can serve as a model to encourage ecologists and conservationists to delve into the long-term history of other study systems via multidisciplinary synthesis or collaboration.  2.3 The Garry oak ecosystem Garry oak (Quercus garryana), also called Oregon white oak, ranges from British Columbia to California (Fig. 2.1).  In Canada, it is found only on the southeastern tip of Vancouver Island, some of the nearby Gulf Islands, and in two isolated populations in the lower Fraser Valley of mainland British Columbia (Fig. 2.1; Fuchs, 2001).  It is the only native oak in British Columbia.  The term ?Garry oak ecosystem? refers to a range of vegetation types along with their associated animal species (Fuchs, 2001), and includes open habitats without significant tree cover.   However, the archetype of these ecosystems is the Garry oak savannah, which consists of an open canopy of oak with an understory dominated by native wildflowers and grasses (Fig. 2.2a).  Garry oak ecosystems in British Columbia are restricted to the rainshadow of the Olympic and Vancouver Island Insular Mountains where the climate is drier than the rest of the British Columbia coast (Meidinger and Pojar, 1991).    These open landscapes were attractive to early European settlers, who founded the city of Victoria in the heart of Vancouver Island?s savannahs.  Captain George Vancouver himself wrote of ?? extensive spaces that wore the appearance of having been cleared by art? (Turner, 1999).   Since the establishment of the city in 1843, it is estimated that over 90% of Garry oak savannah has been lost to agriculture, development, and forest infilling (Lea, 2006; Fig. 2.2c).  Calls to    17 Figure 2.1: The range of Garry oak in North America and on Vancouver Island is shown in red.  Black triangles are archaeological sites within this range compiled from the Canadian Archeological Radiocarbon Database (Morlan, 2005) where radiocarbon-dated materials indicate human presence prior to European settlement.  Range maps based on Erickson (2008).  preserve the Garry oak itself have been ongoing from the early 1900s, and more recently the tree has become the flagship species in a campaign to restore and conserve the remaining savannahs (Cavers, 2009; GOERT, 2011).  The Garry oak ecosystem is now recognized as one of Canada?s most at-risk terrestrial ecosystems, with over 100 associated threatened species (Fuchs, 2001).   Ecologists recognize two broad types of Garry oak vegetation (Roemer, 1993).  On flatter areas with deeper soils, a Garry oak ?parkland? habitat supports large, stately trees.  These ecosystems are generally less moisture-limited and are prone to invasion by conifers like Douglas-fir (Pseudotsuga menziesi; Fuchs, 2001).  In steeper areas with shallow, drier soils, smaller Garry   18oak trees present a ?scrub oak? type habitat that is not conducive to conifer invasion.   Besides habitat destruction, fragmentation and conifer encroachment, the main threat to native biodiversity in Garry oak ecosystems is thought to be the spread of invasive, non-native species (GOERT, 2011; Parks Canada Agency, 2006a,b).   Figure 2.2: (a) Garry oak savannahs consist of an open canopy of oak trees with an understory of grasses and wildflowers; (b) Here the understory is dominated by exotic annual grasses and the exotic shrub Scotch Broom (Cytisus scoparius); (c) An aerial photo of Summit Park, in the city of Victoria, which contains a reservoir and a fragment of Garry Oak savannah completely surrounded by urban development. The photo in (a) was taken in this park.   192.4 Ecological research reveals the hidden causes of exotic invasions Contemporary ecological research, by which I mean ecological research on the scale of a few years to a decade, has mainly focused on the impact of exotic plant species.  Many of the rare species are native wildflowers that thrive in sunny, open savannah conditions, but seem to be easily outcompeted by exotic grasses and shrubs (Fig. 2.2b).   As of the late 1960s, introduced plants were already so firmly established in many Garry oak sites as to be part of the core of species defining the community (Roemer, 1972).   Although there have been few regional extirpations of native species, natives are generally declining while exotic species continue to spread (MacDougall, 2004).  The proportion of species that are exotic in Garry oak savannah patches on Vancouver Island currently ranges from 20 to 70% (Lilley and Vellend, 2009; MacDougall and Turkington, 2006).    However, the success of exotic plants and the concurrent decline of natives turns out not to be a result of straightforward competitive superiority.  In small plots with experimentally manipulated levels of herbivory and competition, herbivory had a much greater negative influence on the establishment, growth and reproduction of native species than the presence of competitors (Gonzales and Arcese, 2008).  An increase in deer populations in the region since the 1940s is thought to be related to forest fragmentation, extirpation of predators, and reduced hunting levels.  The dominance of non-native species is at least in part an indirect result of increased herbivory, which limits the performance of native plants (Gonzales and Arcese, 2008).     The success of exotic plants is also caused by differing responses of exotics and natives to the pervasive human disturbance since European settlement.  In 43 remnant savannah fragments in   20the Victoria area, the richness of exotic plants was found to be positively associated with the density of roads found within 500m of the savannah fragment, whereas native richness declined with increasing road density (Lilley and Vellend, 2009).  High road density surrounding a fragment means increased human presence, more trampling and more frequent introduction of exotic plants and their seeds.  Rather than outcompeting natives, exotics appear to be benefiting from increased human disturbance, which facilitates the dispersal of the seeds of exotics well-adapted to high human disturbance levels, while at the same time causing declines in native species not able to withstand this type of disturbance.  On small islands with little recent human impact, native species still dominate Garry oak savannahs (Bennett, 2012; Smith, 2007).  There is also evidence that the current rarity of native species - and the resulting scarcity of available seeds - slows their recovery when conditions improve.  In one Garry oak savannah, the two dominant exotic perennial grasses, orchard grass (Dactylis glomerata) and Kentucky bluegrass (Poa pratensis), declined under annual mowing while some subordinate species increased in abundance (MacDougall and Turkington, 2005).  However, native perennial grasses also declined significantly with mowing, and almost half of the subordinate species could not increase without seed addition.  Subsequent restoration experiments have shown that at many sites, seed addition is necessary to increase native plant cover (Stanley et al., 2011). Thus, even when the disturbance regime is designed to favour natives, seed limitation of rare natives may prolong exotic dominance.  In sum, ecological studies have shown that despite the appearance of superior competitive ability, the root causes of exotic spread are habitat fragmentation and changes in herbivory and   21disturbance regimes.  These factors have facilitated exotic success while isolating native populations to such an extent that seed limitation now slows their resurgence even where exotic species are removed or disturbance regimes are modified.     2.5 Historical ecology:  fire on the landscape The condition of Garry oak ecosystems today has clearly been heavily influenced by European settlement and landscape modifications beginning in the mid-1800s.  Research in historical ecology has helped to provide a clearer picture of the dynamics of the ecosystem prior to these sweeping changes.   So far, researchers have examined the written records left by early land surveyors, explorers and ethnographers, as well as the biological record preserved in tree rings.  2.5.1 Written historical records In addition to descriptions of the beautiful park-like landscapes, early explorers and settlers often mentioned the prevalence of fire, which they usually attributed to the First Nations peoples.  For example, in 1857 one prominent settler wrote:  ?The natives all along the [Victoria area] coast have a custom of setting fire to the woods in summer? (Turner, 1999).  Most European settlers viewed these fires as dangerous and destructive, but a few acknowledged that the fires had a purpose:  ?? to clear away the thick fern and underwood in order that the roots and fruits on which they [the indigenous peoples] in a great measure subsist may grow the more freely and be the more easily dug up? (Turner, 1999).   These historical descriptions indicate that the open character of the landscape at the time of European settlement, in particular on deep soil sites, was likely a result of frequent, low-intensity fires set by the indigenous inhabitants (MacDougall et al., 2004; Turner, 1999).     22 Ethnographic research supports the purposeful and regular use of fire by the Coast Salish peoples of this region.  Before the introduction of the potato (Solanum tuberosum), the most important source of starch in the diet of the Coast Salish was the bulb of the camas plant (Camassia spp.), which grows in Garry oak savannahs (Beckwith, 2004; Turner, 1999; Turner and Bell, 1971; Turner and Kuhnlein, 1983).  The most productive camas meadows were owned by particular families or individuals (Suttles, 1951).  Women used digging sticks to harvest the camas bulbs in summer after the plants finished flowering, and then the area was burned (Turner, 1999).   The bulbs were steamed in underground pits for several hours, then dried and stored for use throughout the fall and winter (Beckwith, 2004; Suttles, 1951; Turner and Bell, 1971).    It is estimated that at least 10 000 camas bulbs were harvested each year per family (Beckwith, 2004; Deur and Turner, 2005).  Therefore, very large areas of productive camas meadows would have been required to sustain the Coast Salish peoples on Southern Vancouver Island prior to European settlement, even with conservative population estimates (Beckwith, 2004).   But the extensive deep-soil savannahs close to Victoria were the first to be appropriated by Europeans for agriculture and by the early 1900s the Bush Fire Act forbidding fires was more strictly enforced (Beckwith, 2004; Turner, 1999).   Some anthropologists have suggested that smallpox epidemics several decades before European settlement may have decimated the population of indigenous peoples on the British Columbia coast by as much as 90% (e.g. Harris, 1994).  If so, the Garry oak savannahs as seen by the first settlers may have already been in decline due to a reduction in human management via prescribed burns, resulting in infilling by conifers on deep soil sites.   23 Ethnographic and historical research continues to challenge the longstanding assumption that the indigenous people of this region did not cultivate or manage plant-based foods due to bountiful marine resources (Deur and Turner, 2005).  The overwhelming ethnographic evidence of the importance of plant foods and materials in Coast Salish culture has forced ecologists to realize that people were influencing Garry oak ecosystems long before the first Europeans arrived.  2.5.2 Land survey records The maps and notes produced during the early land surveys just prior to European settlement in the mid-1800s, along with historical photographs and remnant oaks, have been used to quantify the decline of Garry oak ecosystems (Lea, 2006).   Prior to European settlement, Garry oak ecosystems were found across a wide range of topographic, soil and climate conditions, whereas today they are concentrated in areas of higher elevation, steeper slopes, cooler temperatures and shallower soils - where conditions did not favour agriculture.   As a result, the savannahs that remain, mostly in protected parks or preserves, represent a highly biased subset of the conditions once occupied by Garry oak ecosystems (Vellend et al., 2008).  The first European land surveyors measured and identified as many as ten ?bearing trees? closest to each stake they placed at the intersection of new property boundaries, in case the stake was lost.  By returning to the same intersections and assessing the present-day tree community, researchers can quantify changes in forest density and composition over time (Whitney and Decant, 2005).  In this region, forest density has approximately doubled, fire-sensitive species like Western red cedar (Thuja plicata) have increased significantly in frequency, and the size   24distribution of trees has shifted to a much higher proportion of small trees (Bjorkman and Vellend, 2010).  This evidence supports the idea that frequent fire prior to European settlement and fire suppression afterwards profoundly changed the landscape.  However, the abundance of Garry oak itself relative to other tree species has not declined significantly, although the frequency of open habitats has declined tremendously.  The descriptions provided by the surveyors indicate that most of the open habitats described as ?prairie? or ?plains? were sprinkled with sparse Douglas-fir trees, not oaks.  The surveyors called this vegetation ?pine plains?, but it is extremely rare in the region today.  The Garry oak has become the flagship species for these ecosystems because it is the dominant tree species of open habitats on the current landscape.  However, this historical research revealed the surprise that in the past many of the open habitats were actually associated with Douglas-fir or no trees at all (Bjorkman and Vellend, 2010).    2.5.3 The tale of the tree-rings Dendroecologists have reconstructed the history of Garry oak stands using tree cores and analyses of present-day forest size and age structure.  The oldest of the Garry oaks they measured were over 300 years old, but these ancients are few.   At each site, a major spike in oak recruitment occurred in the mid-1800s, right around the time of European settlement (Dunwiddie et al., 2011; Gedalof et al., 2006; Pellatt et al., 2007; Smith, 2007).  An increase in Douglas-fir recruitment followed about 10-20 years later.  Douglas-fir recruitment has continued since then, but oak recruitment stopped almost entirely by the mid-1900s.   Although some sites have abundant oak seedlings, saplings are rare, revealing a failure of seedlings to make it to the   25sapling stage and eventually into the adult population.  The result is the increasing dominance of closed-canopy Douglas-fir forests on former savannah sites.  The surprising implication of these results is that many of today?s oaks on deep soil sites appear to be a result of fire suppression at or just after European settlement.  Surface fires set frequently by the Coast Salish people would have severely reduced survival of oak seedlings, but the end of this fire regime seems to have stimulated a pulse of oak recruitment in the sunny, open conditions.  With continued lack of disturbance, conifers like Douglas-fir have been able to establish under the oak canopy, filling in savannahs and creating conditions that are too shady for oak seedlings to survive to sapling age.  Prior to European settlement, open habitats were dotted with large mature oaks or Douglas-firs, or trees were largely absent (Gedalof et al., 2006).  Following fire suppression, coniferous forests shifted in composition towards fire-sensitive species, while open habitats saw a spike in recruitment of oaks, followed by Douglas-firs and eventually leading to closed-canopy conditions.    2.6 Digging deeper:  paleoecology and archaeology Historical and ethnographic evidence suggests that the extensive savannahs at the time of European settlement were a product of human management, but how far back did this relationship extend prior to that point, and when did Garry oak ecosystems arise?    The history of vegetation on southern Vancouver Island since the retreat of the glaciers approximately 12,000 years ago has been documented in multiple studies of pollen and charcoal stored in sediments of the region?s bogs, lakes and the Saanich inlet (e.g. Brown and Hebda,   262002; Hebda, 1995; Pellatt et al., 2001; Fig. 2.1).   From approximately 11,450 to 8,300 years ago, the landscape was dominated by sparse Douglas-fir and high levels of grass and bracken fern (Pellatt et al., 2001).  A relatively dry climate was typical throughout the coast of British Columbia at the time.  Garry oak pollen is not found in significant amounts in the area until approximately 8,300 years ago, and began increasing rapidly, reaching its peak between 8,000 and 6,000 years ago (Heusser, 1983; Pellatt et al.,. 2001; Fig. 2.3).  Figure 2.3:  Changing pollen abundance for Western red-cedar and Garry oak over the past 10,000 years, as a percentage of the maximum relative abundance for each species, from the Saanich Inlet sediment core (Pellatt et al., 2001).  Dots are raw data; lines are locally weighted (LOESS) regression curves.  Data for the most recent 1,000 years were not available from this core.  Bars show the age frequency of radiocarbon-dated materials from 71 archaeological sites within the range of Garry oak ecosystems on Vancouver Island (see Fig. 2.1 for site locations).  I compiled archaeological data from the Canadian Archaeological Radiocarbon Database (Morlan, 2005).  To avoid overrepresentation of sites with many radiocarbon determinations, I used only one radiocarbon date from each site per 200-year interval, resulting in a total of 185 individual dates.    27After this point, the regional climate became cooler and wetter, and moisture-loving trees including Western red cedar increased in abundance (Hebda, 1995; Pellatt et al., 2001).  Garry oak pollen declined at this time, yet by 3,800 years ago Garry oak persisted and even began to increase, despite climatic conditions favouring moisture-loving, shade-tolerant species like cedar (Fig. 2.3).  Paleoecologists suggest that this maintenance of oak savannah habitat could be attributed to human management using fire (Pellatt et al., 2001).  A study of lake sediment charcoal showed an increase in fire frequency on southern Vancouver Island approximately 2,000 years ago, although weather conditions remained cool and moist at that time - the cause is again attributed to humans (Brown and Hebda, 2002).    Archaeological evidence provides support for the idea that humans could have been managing these systems thousands of years ago.  The earliest known archaeological sites from within the range of Garry oak ecosystems in British Columbia date to just over 5,000 years ago, and people almost certainly lived in the area even earlier, but rising sea-levels have left the evidence underwater (Grier et al., 2009).  The time period between 2,400 and 1,000 years ago marks the appearance in the archaeological record of the large, rectangular plank houses, multi-house settlements, and other aspects of technology and art that were characteristic of Coast Salish cultures at the time of European contact (Matson and Coupland, 1995; Lepofsky et al., 2005).  While these characteristics are not required (or sufficient) to indicate human management of the landscape, they provide evidence that practices observed in the ethnographic and historic records extended back in time thousands of years.  A compilation of calibrated radiocarbon dates I obtained from the Canadian Archaeological Radiocarbon Database (CARD; Morlan, 2005) shows the frequencies of dated archaeological materials from 71 sites within the current range of   28Garry oak ecosystems in British Columbia (Fig. 2.1; Fig. 2.3).  The increase in dated sites beginning around 3,400 years ago and again after 2,000 years suggests an increasing human presence on the landscape, although more dates and more sites are needed to make a stronger case for changes in indigenous population levels.  It is credible that the indigenous peoples of southern Vancouver Island were managing savannahs using fire by at least 2,000 years ago and maybe even earlier.  But there are still open questions regarding the details of this management, the spatial extent of human influence, and the relative importance of human-caused versus climatic vegetation changes.  For example, two studies of charcoal fluctuations in sediments from three lakes in the region yielded mean fire return intervals of 27-41 years and 88 years, respectively (Lucas and Lacourse, 2013; McCoy, 2006; Pellatt et al., 2007).  If the Coast Salish were burning savannahs yearly or every few years, this is not evident in the charcoal record at these sites.   It could be that the low-fuel, low-intensity fires characterizing savannah management did not produce sufficient charcoal to be detected above the natural variability reflected in the background charcoal signal.  The archaeological record of southern Vancouver Island also lacks definitive evidence of the pit ovens used to cook camas bulbs (McLay, 2010), although the analysis of ancient camas ovens in the United States has been used to document the intensification of camas use on the lower Columbia River region between 3,500 and 2,000 years ago (Thoms, 1989).   More archaeological investigations into ancient terrestrial resource use on southeastern Vancouver Island are needed to solidify our understanding of the history of human management of these ecosystems.    292.7 Implications for this and other threatened systems Studies from ecology, historical ecology, ethnography, paleoecology and archaeology have uncovered several surprises concerning the structure and dynamics of Garry oak ecosystems.  More importantly, the combination of disciplines has been necessary to begin to answer the questions posed in Box 1:  (1) Landscape burning by the Coast Salish people was common practice prior to European settlement, and necessary to maintain savannah on deep soil sites; (2) These fires were likely frequent, low-intensity surface fires designed to maintain the open conditions favouring preferred food plants;  (3) Climate has been the main driver of broad regional vegetation change over the past 10,000 years.  The cooling, moistening trend that began 6,000 years ago has favoured coniferous forests overall.  However, in some areas human management via fire has facilitated the maintenance of savannah systems; (4) Savannahs were much more open and widespread just prior to European settlement, and many were associated with Douglas-fir trees rather than Garry oaks.  Their variability in space and time before that is not known; and  (5)  Given the extreme fragmentation of the landscape today, and the high density of human settlements, implementing widespread landscape fires is not feasible (MacDougall, 2004).  Research is underway to explore the potential use of fire or fire surrogates (e.g. MacDougall and Turkington, 2007; Stanley et al., 2011) and reinstating traditional camas harvesting practices (Proctor, 2010) to restore Garry oak savannah remnants.  An especially rich combination of data sources is available to elucidate the processes underlying long-term dynamics of the Garry oak ecosystem, but it is clear that delving into the long-term history of any ecosystem will often lead to illuminating results.  For example, a consideration of historical records of the ivory trade revealed that the abundance of trees on the Serengeti plains   30in the mid-1900s was a direct result of the decimation of elephant populations for ivory several decades before; the increase in elephant numbers in the 1960s (resulting in the decline of trees due to elephant herbivory) was simply due to the population recovering (Sinclair, 2012).  Similarly, paleoecology has revealed that some tree range expansions in the American southwest, previously assumed to be invasions resulting from post-European fire suppression, actually represent recovery from overharvest by ancient societies or ongoing expansion following climatic changes thousands of years ago (Swetnam, 1999).  Delcourt and Delcourt (2004) combined archaeology, paleoecology, geomorphology and witness tree data to reconstruct the abandonment of the Cahokia metropolis of the Mississippi Valley over 500 years ago due to deforestation and subsequent erosion and flooding.  In Sweden, researchers have used century-old landscape maps to show that current grassland plant species diversity is more closely related to landscape connectivity 100 years ago than it is to connectivity today, revealing that further losses of species can be expected even if the current amount of grassland is maintained (Lindborg and Eriksson, 2004).  The list could go on, and I argue that this kind of approach is applicable to any ecosystem.  While large well-planned and coordinated multidisciplinary collaborations are certainly the ideal (e.g. Berglund, 1991), it is worthwhile piecing together the information that is already available as a result of independent research, as I have done here.  Compiling the long-term history of an ecosystem also reveals gaps in the timeline.  For the Garry oak ecosystem, historical documents, ethnography, land-survey data and dendroecology have shown that open, savannah-like conditions were much more extensive at the time of European arrival than they are today.  However, the degree of variation in the openness of this landscape at particular locations in the centuries prior to that is not known.  In Chapters 5 and 6, I use the   31phytolith record in soil cores to extend this record back in time with the goal of determining when Douglas-fir began increasing at particular sites on the landscape, and whether this was related to climatic and/or cultural changes.  Further archaeological investigations into ancient terrestrial resource use will also be crucial to fill in this gap.  In addition, there is little quantitative data for ecosystem changes over the most recent few decades.  Caught between typical ecological studies of less than five years and historical studies on the scale of a century or more, this temporal scale has been called ?the invisible present? (Magnuson, 1990). This is a knowledge gap in the history of many ecosystems.  In Chapters 3 and 4, I resurvey 184 ?legacy? vegetation plots (i.e. those surveyed by ecologists decades in the past) on the Saanich peninsula to quantify changes that have occurred in the understory of Garry oak and associated plant communities since the late 1960s (Roemer, 1972; Vellend et al., 2013).    Delving into the long-term history of an ecosystem can raise the question of whether we should be trying to conserve an ecosystem state that was created by human activities (Allison, 2004).   Given that deep-soil Garry oak savannahs will ?naturally? be invaded by Douglas-fir without fire or other disturbances, should we be worried about losing the many threatened species of open Garry oak habitats?  In my opinion, the answer is ?yes?.  The fact that past human actions played a role in creating habitat for now-threatened species makes these species no less worthy to conserve.  We need to revise our understanding of ?natural? ecosystems to include those created or maintained by cultural activities.  These ?cultural landscapes? are already valued and protected in Europe (e.g. Malmer, 1991; Rackham, 1998; Bignal and McCracken, 2000).  North American conservationists have been slower to recognize cultural landscapes, but it is starting to happen.  For example, the recent decline of grassland birds and other specialists of open habitat   32in the northeastern US has been linked to the abandonment of agricultural practices and subsequent invasion of shrubs and trees (Norment, 2002; Dunwiddie, 2005). By gaining a deeper understanding of the role of human activities in ecosystems in the past, we can better determine how much ongoing management will be required to restore them, and the limitations to restoration due to land-use history (Foster et al., 2003).     2.8 Conclusion It is widely accepted that most of the ecosystems of the Americas were influenced by indigenous peoples (e.g. McCann, 1999), but the degree of influence was not the same for all regions (Vale, 2000).  Work is needed to determine the extent, location, and duration of human influence for different ecosystems and different regions.  Information from other disciplines can suggest hypotheses and/or corroborate explanations for ecological patterns.  For example, the historical and ethnographic understanding of the use of fire prior to European settlement suggested that fire suppression following settlement may have changed the Garry oak ecosystem in a particular and profound way, which could then be tested and confirmed using information from historical land survey records and contemporary vegetation surveys.  In addition, by using information from multiple disciplines, ecologists can gain a better understanding of the long-term history of the human-ecosystem relationship.  The goal is not merely to be able to choose a restoration baseline from among former ecosystem states, but to better understand the potential effects of human management in the present, and hopefully be more successful in our efforts, whatever our restoration goals may be.      33Chapter  3: The role of native species in promoting biotic homogenization over four decades on the Saanich peninsula  3.1 Synopsis A long-term perspective is needed in order to understand how disturbance is affecting plant communities in human-dominated landscapes.  In recent decades, increased human disturbance has often resulted in declining local native species richness, gains in exotic species, and a decline in beta diversity.   However, it is far from certain whether a general decline in plant diversity is occurring across all disturbed landscapes, and knowledge gaps remain concerning how the spread of exotic species influences beta diversity over long time scales.  In this chapter, I examine changes in total diversity, local diversity, and beta diversity over four decades on the Saanich peninsula using data from a resurvey of 184 vegetation plots that were originally surveyed in the late 1960s.  I also compile information on the traits of each species, and test for correlations between traits and plant species success over four decades.  The results show striking increases in local and total plant species richness driven by both native and exotic species. The most successful species tended to be exotic, disturbance tolerant, shade tolerant and shrubs.  Biotic homogenization occurred, but not as a result of exotic species colonization, instead being significantly correlated with gains in native species.  The loss in beta diversity has resulted in a shrinking of the gradient of vegetation types, blurring the distinction between them.     3.2 Introduction Habitat loss, disturbance and the introduction of exotic species to new areas are causing declines in biodiversity on a global scale (e.g. Butchart et al., 2010; McKinney and Lockwood, 1999).    34Changes in diversity at the local scale are more difficult to quantify because of the scarcity of data on species presences in small-scale plots over time periods longer than a decade (Sax and Gaines, 2003).  Recently, plant ecologists have been addressing this gap by using so-called ?legacy? data (Vellend et al., 2013) to quantify long-term changes in plant community diversity at the local scale (e.g. Damschen et al., 2010; Keith et al., 2009; Rogers et al., 2008; Smart et al., 2006).  Most have found declines not only in plot-level diversity (alpha diversity) and regional diversity (gamma diversity), but also in beta diversity:  that is, a reduction in variability in community composition across plots.  However, these studies are still relatively few, and it is far from certain whether a general decline in plant diversity is occurring across all landscapes.  Some legacy studies from Europe have found increases in local and regional diversity (Aggemyr and Cousins, 2012; Van Calster et al., 2007).    Loss in beta diversity occurs via biotic homogenization, whereby widespread generalist species colonize plots across a region while more specialist species are lost, reducing the variability in community composition (McKinney and Lockwood, 1999).   Loss of diversity at this level can go unnoticed because beta diversity can decline even though alpha and gamma diversity remain unchanged or even increase (Keith et al., 2009; Olden and Poff, 2003; Van Calster et al., 2007).   Biotic homogenization as originally defined is ?? the replacement of local biotas with nonindigenous species, usually introduced by humans? (McKinney and Lockwood, 1999).  But studies using native and exotic species lists to calculate beta diversity between localities, watersheds, or states have found the relationship between exotic species richness and beta diversity to vary depending on the taxonomic group under study and the spatial scale (McKinney, 2004).  Among the handful of legacy studies documenting biotic homogenization   35over time, only one found a significant linkage between exotic gains and biotic homogenization (Rooney et al., 2004).  Another found the two uncorrelated, despite substantial gains in exotics at the plot level (Rogers et al., 2008).  Clearly knowledge gaps remain concerning how exotic species influence beta diversity over long time scales, and how their potential effect may vary with changing disturbance levels and/or the characteristics of the ecological community in question.  Increasing human dominance of the landscape results in multiple potential drivers of change in ecological communities, including climate change, eutrophication, and habitat fragmentation (Jackson and Sax, 2010). The challenge is to understand which of these drivers are most important in determining which species are ?winners? and ?losers? in a particular region, thus mediating changes in alpha and beta diversity (Olden and Poff, 2003).  One way to do this is to use traits of species.  For example, if eutrophication via nitrogen deposition is an important driver of change, species associated with high nutrient conditions should be more likely than others to increase in frequency and/or abundance over time (e.g. Smart et al., 2005).  In this chapter, I investigate changes in local plant community diversity over a period of four decades on the highly developed Saanich peninsula (Fig. 3.1).  In Canada, the highest levels of human population density and biological diversity coincide in the south (Kerr and Cihlar, 2004), and therefore many areas of southern Canada are priorities for conservation.  I resurveyed 184 vegetation plots originally surveyed in the late 1960s.   At that time, exotic species were already present in 63% of plots.  Since the first survey, the human population in the region has doubled   36(Statistics Canada, 1969, 2011), the landscape has become increasingly fragmented, and new invasive exotic species have emerged.  Therefore, I predicted (i) increases in exotic species richness and abundance, declines in native species, and a decline in alpha diversity since 1968; (ii) biotic homogenization, driven by the expansion of exotic species across the landscape, and  Figure 3.1:  The study area.  Relev? plot locations are indicated by black dots.  Light grey shading indicates residential, commercial, industrial or institutional land use.  Dark grey shading indicates agricultural land use.  Unshaded regions of the peninsula are parks or other green spaces, military land, or First Nations reserve land.    37 (iii) a shift in community composition favouring disturbance tolerant species, primarily exotic species.  Contrary to my expectations, I found that plot-level and total diversity have both increased over forty years, with increases in both exotic and native species, although beta diversity has declined.  Despite the spread of exotic species across the landscape, I did not find a correlation between exotic invasion and biotic homogenization, finding instead that colonization by native species was a stronger predictor of losses in beta diversity.  3.3 Materials and methods 3.3.1 The study area The Saanich Peninsula is an area of approximately 330 km2 including the city of Victoria and partially separated from the bulk of Southeastern Vancouver Island by Saanich Inlet (Fig. 3.1).  It became one of the first areas in British Columbia to be settled by Europeans when Victoria was established in 1843.  Since then steady growth in agriculture, logging, and suburban development has resulted in a highly fragmented landscape.  At the time of the first survey, about one third of the peninsula outside the city of Victoria still supported forest or savannah vegetation (Roemer 1972).  Much less than that remains today (Fig. 3.1).     Most of the study area is underlain by granitic or volcanic bedrock and covered with a layer of unconsolidated glacial till (Day et al., 1959).  Soils are mainly brunisols and gleysols which have developed on morainal or marine deposits (Jungen, 1985).  The Saanich Peninsula is located in the rainshadow of both the Olympic Mountains and the Vancouver Island Mountains, resulting in an unusually warm and dry climate compared to the rest of coastal British Columbia.   The climate is described as ?cool Mediterranean?, with warm, dry summers and mild, wet winters.   38Total annual precipitation averages about 87cm, with only about 10cm of this falling in the summer months (June to September).    The Saanich peninsula lies within the Coastal Douglas Fir (CDF) biogeoclimatic zone (Meidinger and Pojar, 1991).  Douglas-fir forests are the dominant vegetation, characterized by a canopy of Douglas-fir (Pseudotsuga menziesii), sometimes mixed with Western red cedar (Thuja plicata) or grand fir (Abies grandis), and an understorey often dominated by salal (Gaultheria shallon), Oregon-grape (Mahonia aquifolium or M. nervosa) and sword fern (Polystichum munitum). In the driest forests, Pacific madrone (Arbutus menziesii) grows interspersed with Douglas-fir (Flynn, 1999).  Also within the CDF zone are oak woodlands and savannahs characterized by open grassland and/or shrubs and a sparse canopy of Garry oak trees (Quercus garryana).  Garry oak communities are home to many species that are imperiled provincially or nationally (Fuchs, 2001; Parks Canada, 2006a,b).  Both Douglas-fir forests and Garry oak communities have been reduced in area by more than 90% since European settlement (Flynn, 1999; Lea, 2006).  Due to the fact that savannahs were usually the first areas to be converted to agriculture, current oak community vegetation is now largely confined to higher elevation, steeper areas with shallow soils (Vellend et al., 2008).  The remaining forests of southeastern Vancouver Island also have a much higher density of trees than at the time of European settlement, a consequence of more than a century of fire suppression (Bjorkman and Vellend, 2010).    393.3.2 Resurvey methods The original survey was carried out in 1968-1969 (hereafter 1968; Roemer, 1972).  The abundance of all vascular and non-vascular plant species in 409 relev? plots of 400m2 area was recorded using the Braun-Blanquet cover-abundance scale (Roemer, 1972).  Plot locations were subjectively chosen based on homogeneity of vegetation and avoiding highly disturbed areas and young second-growth forests.  However, given the widespread logging that occurred on the peninsula prior to the original survey, some relatively early successional areas were sampled by necessity.  Plots were not permanently marked, however a description of their location, slope and aspect and a grid reference on an aerial photo (1:10,000 scale) were recorded for each plot.  In 2009, I relocated and resurveyed 184 of the original relev?s.  After the original survey, each relev? was classified into one of 28 plant communities in 7 associations (Roemer, 1972). I limited my resurvey efforts to four associations that cover the portion of the moisture gradient that includes the dominant vegetation type and those of greatest conservation concern:  two Garry oak dominated associations (which I combined and refer to as the ?Garry oak? vegetation type), the Douglas-fir dominated association (?Douglas-fir? vegetation type), and the Arbutus-Pseudotsuga association (?Arbutus? vegetation type), which represents the transition between Garry oak and Douglas-fir (Roemer, 1972).     All relev?s were resurveyed between May and August, 2009.  Relev?s range in elevation from sea level to approximately 325 m above sea level.  I resurveyed only where sites were still vegetated and where I could obtain permission from municipal or provincial park authorities or private landowners.  The aerial photo grid reference system in conjunction with GPS and advice   40from the original surveyor (H.L. Roemer) allowed me to navigate to the approximate location of the plot.  I then used the original notes on the slope, aspect, proximity to roads and/or other landscape features to select the most likely site of the original plot.  As such, the plots cover the same range of environmental conditions (slope, aspect, elevation, soil type, etc.) as found in the initial survey.  I also maximized stand homogeneity within the relocated plot in order to avoid the potential problem of increased heterogeneity in plots initially established with a preference for homogenous vegetation, given spatial variation in disturbance (Palmer, 1993; Ross et al., 2010).  I recorded the abundance of all vascular and non-vascular plant species to the nearest one percent cover, with species covering less than one percent denoted as 0.5 and very rare species covering much less than one percent (i.e. only one or two small individuals) denoted as 0.1.    3.3.3 Statistical analyses I transformed all cover values for the 1968 and 2009 data to the percent cover midpoints of the Braun-Blanquet cover scale, except for species of abundances 0.5 and 0.1 which were maintained as the lowest two classes of the scale.  I standardized plant nomenclature and combined taxa at the level of genus where the accuracy of species identifications was in doubt.  Plant nomenclature follows Douglas et al. (1998).  I separated vascular plant species into tree species and understorey species, which include grasses, forbs, ferns, and shrubs.  Understorey species collectively comprise ~80% of the full set of species observed, and I focused the analysis on this subset of species.  I used linear mixed-effects models to test the effect of year, vegetation type (as classified in 1968), and their interaction on plot-level understorey species richness.  I then used paired t-tests   41to test for a significant change in total species richness, native species richness and exotic species richness between years.    In order to test the hypothesis of a shift in community composition over time, I used permutational MANOVA (PERMANOVA; Anderson, 2001a; Anderson et al., 2008) with year of survey and vegetation type as fixed variables and plot identity as a random variable.  I tested for changes in beta diversity between the two years across all plots, and within each vegetation type separately, using a distance-based test for homogeneity of multivariate dispersions (PERMDISP; Anderson et al., 2006). To visualize any shifts in composition or changes in dispersion, I used non-metric multidimensional scaling (NMDS) to ordinate all plots in both years. I used the Bray-Curtis dissimilarity measure because it takes species abundances into account, and is recommended for tests of biotic homogenization (McCune and Grace, 2002; Olden and Rooney, 2006).    To test for predictors of changes in beta diversity, I first computed the ecological distance from each plot in species space to the centroid of all plots in both years (n=368).  I then calculated the change in this distance by subtracting the value for 2009 from the value for 1968 for each plot.  Positive values indicate that a plot shifted towards the average in terms of plant community composition, thereby contributing to biotic homogenization.  I then used linear regression to test whether the net change in exotic species per plot between years or the net change in native species per plot between years were correlated with the degree of movement towards the centroid.  I also tested for any correlation between the change in number of exotic species and the change in number of native species using Spearman?s rank correlation test.   42 To explore possible drivers of temporal community change I gathered information on plant traits from various published and unpublished sources.  I define ?traits? as both qualitative and quantitative characteristics of plant species and/or their preferred habitats.  I chose traits based on hypothesized potential drivers of change in plant communities in the region, including disturbance, fragmentation, climatic warming, increased nitrogen deposition and plant community succession.  The traits include disturbance tolerance, origin (exotic or native), soil nutrient regime preference, broad geographic range (ranging primarily to the south of the study area vs. study area central in the species? range), seed weight, seed dispersal mechanism, palatability to deer, shade tolerance, form (herbaceous vs. shrub) and specific leaf area (SLA).  For details see Appendix A.1.   In order to determine which species were significantly more frequent and/or abundant in each of the two years, I used an indicator species analysis (Dufr?ne and Legendre 1997) with 1,000 randomizations to calculate P values.  Significant indicators of 1968 were deemed ?losers?, significant indicators of 2009 were deemed ?winners?, and species that were not significant indicators of either year were deemed ?no change? species (P< 0.05).  I then tested for associations between ?winner?, ?no change?, or ?loser? status and each individual trait value using cumulative logit models.  These models preserve the information contained in the ordered nature of the categorical response variable without assuming the distance between categories to be constant (Agresti, 2002).  To assess whether a particular trait was significantly associated with the indicator status of species, I compared a cumulative logit model with the trait value as an independent variable to a model with intercept only.  I limited my analyses to species which   43occurred in at least 18 plot-year combinations (5%), and considered a trait to be related to plant indicator status when the difference in AIC (AIC intercept only minus AICincluding trait) was greater than 5, which corresponds to strength of evidence at least 12 times stronger than the alternative model (Anderson, 2008; Burnham et al., 2011).    I did not attempt to build a model using all traits because the subset of species for which all trait data was available was very small (45 species) and this limited my ability to detect any differences.   I also carried out indicator species analysis on each of the three vegetation types separately, to see whether the same species were ?winners? in different vegetation types.  In addition to testing the association of indicator status with individual traits, I calculated the abundance-weighted average trait value for each plot in each year and then used paired tests to determine whether there have been significant shifts over time.  The data often did not meet assumptions of normality, so I used the Information-Theoretic equivalent to the paired t-test (Burnham et al., 2011), and deemed there to have been a shift in plot-level trait value where the difference in AIC was greater than 5.  I limited comparisons to plots in which at least 80% of the total cover in both years was represented by species for which trait values were available.  I used PRIMER version 6 with PERMANOVA+ (Clarke and Gorley, 2006) to carry out permutational MANOVA and tests for homogeneity of multivariate dispersions.  All other analyses were carried out in R, using the ?vegan? package for NMDS ordination, the ?labdsv? package for indicator species analysis, and the ?ordinal? package for cumulative logit models (Christensen, 2013; Oksanen et al., 2009; R Core Development Team, 2012; Roberts, 2010).    443.4 Results The total number of understorey species recorded increased from 179 in 1968 to 237 in 2009.  Eighty-seven species were new to this set of plots, while 29 of the original species were not recorded in 2009.   The total number of native species increased from 146 to 149 and the total number of exotic species increased from 33 to 82.   With the exception of a few very recently naturalized exotic species, all the species new to my sample of plots were present somewhere on the Saanich peninsula in 1968 (H. Roemer pers. comm.).  The best linear mixed-effects model for species richness included year, vegetation type, and their interaction.  Understorey species richness increased by an average of 8 species, from 21 to 29 species on average per plot (paired t= 12.5946, df = 183, P < 2.2e-16; Fig. 3.2a).  Native species also increased, from an average of 19 to 22 native species per plot (paired t = 6.0091, df = 183, P = 4.935e-09; Fig. 3.2b), and exotic species increased from an average of 2 to 7 exotic species per plot (paired t = 17.2808, df=183, P=2.2c-16; Fig. 3.2c).  On average each plot gained 15 species (9 native and 6 exotic), and lost 7 species (just over 6 native and less than one exotic; Fig. 3.2d).  There was a shift in plant community composition and a decline in beta diversity between 1968 and 2009 (Fig. 3.3).   PERMANOVA confirmed that plots in different years and different habitats are significantly different in terms of understorey plant species composition (Table 3.1).  PERMDISP showed a significant decline in heterogeneity between 1968 and 2009 for all plots (F1,366=9.4394, P=0.006, mean Bray-Curtis distance to centroid(1968)=59.744, mean Bray-Curtis distance to centroid(2009)=56.763).  However, when analyzing vegetation types separately, only Douglas-fir dominated plots showed a significant reduction in dispersion over time, while   45 Figure 3.2: Plots of (a) total understorey species richness, (b) total native understorey species richness, and (c) total exotic understorey species richness in 2009 versus 1968.   Circles = Douglas-fir, squares = Arbutus, triangles = Garry oak.  The line is the 1:1 line; (d) shows the number of species gained and lost per plot by origin.  Arbutus plots showed a non-significant increase in multivariate dispersion and Garry oak plots showed a non-significant decline in multivariate dispersion (Table 3.2).  The average pairwise Bray-Curtis similarity between vegetation types increased from 12.54 in 1968 to 15.10 in 2009.    4671% of plots moved towards the overall centroid over time.  The compositional shift of each plot towards the average was uncorrelated with the change in number of exotic species per plot (F=2.303, P=0.1309, adjusted R2=0.007; Fig. 3.4a) but positively correlated with the change in number of native species per plot (F=12.53, P<0.001, adjusted R2=0.05929; Fig. 3.4b).  The change in number of native species was positively correlated with the change in exotic species (Spearman?s rank-correlation coefficient: ?=0.30977, P<0.0001).    Table 3.1: Results of a PERMANOVA (Anderson, 2001a) with year of survey and vegetation type as fixed variables and plot identity as a random variable.  The analysis was performed using Bray-Curtis dissimilarities on untransformed, non-relativized abundance data, with sequential sums of squares and 999 permutations FACTOR Degrees of Freedom SS MS Pseudo-F P(perm) # of unique permutations year 1 27070 27070 15.4 0.001 999 vegetation type 2 221210 110600 27.6 0.001 998 plot identity 181 725660 4009.2 2.3 0.001 994 year x vegetation type 2 17223 8611.3 4.9 0.001 998 Residual 181 317250 1752.7    Total 367 1308400       There were 101 understorey species with a frequency of at least 5% across all plot-year combinations.  29 species were identified as overall ?winners?, 63 showed no significant change in frequency and/or abundance, and 9 were deemed ?losers? (see Appendix A.2).  Cumulative logit models showed that ?winners? were more likely to be exotic, disturbance tolerant, shade tolerant, and shrubs (Fig. 3.5; and Table 3.3).  The paired tests for change in abundance-weighted trait values between survey periods also found significant increases in abundance-weighted exoticness and disturbance tolerance, but no significant change in abundance-weighted shade tolerance or shrub cover over time (Table 3.3).     47  Figure 3.3: Non-metric multidimensional scaling ordination of all plots in 1968 and 2009 in understorey species space.  The Bray-Curtis dissimilarity measure was used with double Wisconsin standardization and square root transformation.  To avoid a maximum dissimilarity threshold, flexible shortest paths were found between plots with no species in common (Oksanen et al., 2009).  The number of dimensions was set to 3.  The ordination has been centred, the axes scaled to half-change units, and the configuration rotated so that the variation among plots is maximized on axis NMDS1.  Stress in three dimensions is 12.37.      48Table 3.2: Results of pairwise comparisons between plots of the same vegetation type in different years in terms of multivariate dispersions using PERMDISP and the Bray-Curtis similarity measure.  P-values were calculated based on 999 permutations VEGETATION TYPE mean distance to centroid (1968) mean distance to centroid (2009) t P Douglas-fir (n=115) 54.5 50.7 3.19 0.007* Arbutus (n=27) 40.6 44.2 1.15 0.323 Garry oak (n=43) 58.1 55.3 1.89 0.072  3.5 Discussion This study shows a striking increase in taxonomic diversity of plant communities on the Saanich Peninsula over the last 4 decades, with gamma diversity increasing by 32% and alpha diversity increasing by 38%.  However, these increases were accompanied by a significant decline in beta diversity over time.  This is the first North American study I know of to find this pattern, although it has been observed in Europe (Van Calster et al., 2007).   The total number of exotic species more than doubled and each plot gained six exotic species on average, but these dramatic gains do not seem to be the cause of biotic homogenization.  In fact, the shift of plots towards a more similar plant community composition was correlated with gaining native species (although the proportion of variation explained was low: Fig. 3.4b).  The tendency of a colonizing species to cause biotic homogenization depends on the frequency of occurrence of the species elsewhere.  If a plot gains a species that is found in very few other plots, this can lead to biotic differentiation (Olden & Poff 2003; McKinney 2004).  For this reason, the role of exotic species in homogenization has usually been found to be greater at larger scales, where native floras of distant territories have very few species in common and thus are made more similar by the addition of a few widespread exotic species (McKinney 2008).  Of the 101 species included in the indicator species analysis, native species were found in an  49Table 3.3: Results of cumulative logit model comparisons for the association between indicator status (loser, no change, winner) and traits, and for paired tests of the change in abundance-weighted average trait value per plot.  Asterisks indicate strong evidence (?AIC >5) that a trait is associated with indicator species status or that abundance-weighted mean plot-level values for that trait have shifted from 1968 to 2009  association with indicator value (cumulative logit model) change in abundance-weighted average plot value (paired tests) TRAIT explained deviance (pseudo R2) ?AIC (intercept alone versus model with trait)? n  (# of spp.) mean of the paired differences (2009-1968) ?AICc n (# of plots) origin  (exotic = 0, native = 1) 12.54% 19.99* 101 -0.097 36.05 * 184 disturbance tolerance  (not tolerant = 0, tolerant = 1) 11.27% 17.76* 101 0.074 14.27 * 184 form  (0=herb, 1=shrub) 5.28% 7.26* 101 0.035 0.77 184 shade tolerance  (1=very tolerant ? 5=very intolerant) 5.61% 5.08* 72 -0.036 2.94 152 seed weight  (g per 1,000 seeds) 3.23% 2.58 83 1.295 2.54 86 specific leaf area (SLA)  (mm2/g) 0.80% -0.88 86 0.571 -1.43 177 range  (1=southerly, 2=central) 0.07% -1.87 101 0.036 0.70 184 seed dispersal mechanism  (0=none, 1=wind, 2=vertebrate) 0.30% -1.79 37 0.022 -1.83 34 deer palatability  (0=not eaten ? 3=highly palatable) 3.46% 0.93 50 0.034 -1.26 101 nutrient regime  (1=oligotrophic ? 6=hyper eutrophic) 0.11% -1.85 82 0.0415491 1.75 150   50  Figure 3.4: The relationship between the change in distance from the overall multivariate species centroid over time and (a) change in number of exotic species present, and (b) change in number of native species present from 1968 to 2009.  The line in (b) is the best fit line by linear regression; there was no significant correlation in (a). Positive change in the distance to centroid indicates a shift towards increased similarity with other plots.  Circles = Douglas-fir plots, squares = Arbutus/Douglas-fir plots, triangles = Garry oak plots.  average of 45 plots in 1968, while exotic species had an average frequency of only 14 plots.  Therefore a gain in a native species was more likely to cause homogenization.  The importance of gains in native species is supported by an examination of the indicator species of Douglas-fir plots alone, which was the only vegetation type to show biotic homogenization when analyzed alone.  The Douglas-fir vegetation type was the only type that showed an increase in total number of native species across all plots (from 69 to 94), and the majority of the ?winners? in Douglas-fir plots are native (see Appendix A.2).  Local gains in native species have been shown to drive biotic homogenization where exotics are rare or absent (e.g. Naaf and Wulf, 2010; Van   51Calster et al., 2007), however even with substantial gains in exotics at the local and regional scale, it appears that the spread of native species on the Saanich peninsula has promoted the reduction in beta diversity.   The overall ?winners? were more likely to be exotic, disturbance tolerant, shade tolerant and shrubs.  For example, the most successful exotic species include Ilex aquifolium, Hedera helix, and Daphne laureola, all shade-tolerant shrubs, each of which has colonized more than 70 plots since the first survey (see Appendix A.2).  Shade tolerant exotic ?winners? also include the intriguing orchid Epipactis helleborine, which was not found in any plot in 1968 but has since colonized 80 plots (Appendix A.2).  Cytisus scoparius has been an overwhelming ?winner? in Garry oak vegetation where conditions are less shaded.  Native ?winners? include the disturbance tolerant shrubs Rubus ursinus subsp. macropetalus, Oemleria cerasiformis, and Amelanchier alnifolia, and herbs Nemophila parviflora and Galium triflorum.  ?Losers? include herbs that are not disturbance tolerant, such as Trillium ovatum, Erythronium oreganum, and Tiarella trifoliata, and the shade intolerant species Carex inops and Bromus carinatus.  These results suggest that direct human disturbance has been the most important driver of plant community change in this region in the past 40 years, with some role for increased shading and   52 Figure 3.5: The percent of species in each category for the four traits that were found to be strongly associated with indicator status.  The number of species for which trait values were available is listed below each bar (total n=101).  infilling by shrubs.  I did not find evidence that plant communities are shifting in terms of deer palatability, geographic range characteristics, soil nutrient regime preferences, seed dispersal mechanisms or specific leaf area.  However, I realize that my ability to detect shifts in these traits   53could be limited when more than one mechanism is operating that favour different extremes of a trait.  For example, the winners include many heavy-seeded bird-dispersed shrubs, such as A. alnifolia, I. aquifolium, and O. cerasiformis, which may be favoured with forest fragmentation.  However, tiny-seeded species associated with mature forests, such as Monotropa uniflora, and Corallorhiza maculata are also on the winning list, probably as a result of forest maturation over the past four decades, in particular in plots that were at an earlier seral stage when first surveyed.  I also recognize that other, unmeasured characteristics correlated with my measured traits may influence the relative success of a species over the 4 decade period.   My finding that ?winners? were more likely to be shade tolerant suggests a potential role for succession in driving the changes observed in plant community composition.  However, the majority of the plots were already considered mature in 1968.  In addition, the enormous increase in understorey richness in Douglas-fir plots in particular (Fig. 3.2), the significant tendency for disturbance tolerant species to be winners (Fig. 3.5b), and the significant shift in abundance-weighted plot-level disturbance tolerance are trends opposite to what would be expected if succession were the dominant driver.  The tendency for ?winners? to be shade tolerant was also not strong enough to produce a significant increase in plot-level abundance-weighted shade tolerance (Table 3.3).  Thus, although succession may be playing a role in some cases, the results implicate disturbance as the dominant driver of change.  Exotic species were much more likely than natives to be ?winners?:  58.6% of exotic species were ?winners?, while only 16.2% of native species were.  This may be explained if relatively recently introduced exotics had simply not yet filled in their potential ranges in 1968 (an   54?invasion debt?; Bennett et al., 2013).  Also note that the trait ?exotic? is highly correlated with disturbance tolerance.  In the entire dataset, 93% of the exotic species are disturbance tolerant, but only 23% of native species are disturbance tolerant.  Although many exotic species have been highly successful over the time frame of the survey, they are not yet the driving force behind biotic homogenization, nor do they appear responsible for extirpations of native species, given the positive correlation of temporal change in exotic and native species richness.  The indicator species analyses identified many more ?winners? than ?losers?.  The ratio of extirpated to colonizing species (the ?extinction pressure?, sensu Olden and Poff, 2003) is low for this dataset (0.33 for all plots pooled, 0.46 on a plot-level basis), indicating that at this time colonizations are not causing high levels of extinction.  Of course, this could change in the future, especially if there are time lags between colonization and resulting extinctions, resulting in an extinction debt (Jackson and Sax, 2010; Vellend et al., 2006).     Finally, I would like to emphasize the importance of the breadth of the environmental gradient to my finding of biotic homogenization.   Had I considered only Garry oak plots, for example, I would have found a shift in plant species composition, but no significant biotic homogenization.   The changes occurring on this landscape are resulting in a blurring of the distinction between plant community types.  The total length of the dominant vegetation gradient has declined from 3.41 to 3.11 half-change units (Fig. 3.3), and the average Bray-Curtis similarity between vegetation types has increased.  This is a trend that has also recently been observed over 50 years in upland plant communities in Scotland (Ross et al., 2012).  The reason why the increase in alpha diversity led to significant biotic homogenization among Douglas-fir plots but not the other vegetation types may be due to different starting points on the gradient from less to more   55disturbed (e.g. Smart et al., 2006), a result of differences in land use history.  Most Garry oak plots were already highly invaded by exotics and isolated in small fragments at the time of the first survey, whereas Douglas-fir plots were still found in more intact, contiguous forest areas where exotic species and disturbance-associated natives were extremely rare.  As a result, Douglas-fir plots saw a much more dramatic relative increase in species, and significant biotic homogenization with respect to each other. The ability to detect biotic homogenization not only depends on the spatial and temporal scale at which it is tested (e.g. Baiser et al., 2012; Olden and Poff, 2003), but also the length of the plant community gradient over which it is examined.  3.6 Conclusion A long-term perspective is crucial in order to understand how increasing human dominance is changing plant communities worldwide.  The majority of long-term studies to date have shown a decline in plant diversity over time.  This study shows that human-mediated disturbance is an important driver of plant community changes in recent decades on the Saanich Peninsula, but the net result has actually been an increase in richness for both natives and exotics.  However, these increases were accompanied by a decline in beta diversity across the landscape, promoted by the colonization of common, disturbance-tolerant native species.  This biotic homogenization is equivalent to a shrinking of the vegetation gradient, and may not have been detected if the full range of vegetation types had not been included.       56Chapter  4: Using plant traits to predict the sensitivity of colonizations and extirpations to landscape context  4.1 Synopsis In Chapter 3, I documented a large increase in alpha diversity on the Saanich peninsula over four decades, with species richness going up by 8 species per plot on average.  The ?winners? tended to be exotic and disturbance tolerant, suggesting that human disturbance of the landscape is likely the largest driver of these changes; but the landscape is not homogeneous.  In order to conserve plant community diversity in highly fragmented regions, we would like to be able to predict which species are most vulnerable and where.  In this chapter, I test the ability of plant life history traits to predict colonizations and extirpations based on surrounding landscape context.  I relate apparent colonizations and extirpations to road density, total forested area, and patch shape within 500m of each plot in 1964 and 2005, and then ask whether the direction of these relationships is predictable based on nine traits.  I also test for a change in the relative influence of local environmental factors and surrounding landscape context on plant community composition.  Plant community composition in the present is still correlated most strongly with environmental variables, with a slightly increased role for landscape context.  Apparent colonizations were predictable based on some traits, with exotic, annual, herbaceous, shade intolerant species and those with higher specific leaf area more likely than others to colonize areas with higher surrounding disturbance and fragmentation.  However, extirpations were rarely predictable based on traits, suggesting a potential time lag of more than four decades for extirpations in response to landscape change.     574.2 Introduction Ecologists want to understand how and why plant communities change over time in order to improve ecological theory and conservation of plant diversity.   Terrestrial plant communities are being affected more and more by the accelerating human dominance of landscapes.  Deforestation and land conversion to agriculture have resulted in the fragmentation of formerly large, contiguous areas of natural vegetation into small ?islands? (Fahrig, 2003; Lindenmayer and Fischer, 2006; Wilcove et al., 1986).  Within these ?islands?, plant community composition is altered due to the configuration of the remaining habitat and human disturbances in the region, such as changing agricultural intensity, urbanization, pollution, introduction of new species, and changes in fire regimes.  The study of plant populations in discrete patches of ?natural? habitat in a matrix of agricultural and urban land use seems to lend itself well to island biogeography or metapopulation theory, in which a network of populations more or less isolated in a matrix of unsuitable areas are connected via dispersal.   Colonizations and extirpations within each patch depend on its size and distance from other patches (Hanski, 1999; MacArthur and Wilson, 1967).  But applying this framework to plant communities has been difficult due, in part, to the difficulty of measuring plant dispersal events, the long-lived nature of many plants, and their ability to survive long periods of unsuitable conditions by lying dormant in the seed bank (Alexander et al., 2012; de Blois et al., 2002; Debinksi and Holt, 2000; Freckleton and Watkinson, 2002).  These characteristics can lead to long time lags between landscape fragmentation and the achievement of any sort of equilibrium in terms of plant community composition (Eriksson, 1996; Helm et al., 2006; Jackson and Sax, 2010; Kuussaari et al., 2009; Lindborg and Eriksson, 2004; Vellend et   58al., 2006).  Nevertheless, colonizations and extirpations are the key to understanding plant community dynamics, and hence there is a need for long-term studies of plant communities in fragmented landscapes (Alexander et al., 2012; Dupr? and Ehrl?n, 2002; Foster et al., 1998; Lunt and Spooner, 2005).  In particular, we would like to understand what traits of plant species allow them to colonize (or prevent them from being extirpated) in the face of landscape changes, because this will enable us to predict which species are most vulnerable.       For plant communities in fragmented landscapes, the surrounding landscape context may be at least as important as soil and climatic factors in determining plant community richness and composition (e.g. de Blois et al., 2001; Gergel et al., 2002; Williams et al., 2005).  Landscape context includes both the spatial configuration of vegetation patches and the degree of human disturbance ? as determined by the type of land use - in the surrounding matrix (de Blois et al., 2001).  Spatial configuration, including the area of the patch, its shape, and its isolation from other patches can influence the ease with which important processes like pollination and seed dispersal can occur (e.g. Kolb and Diekmann, 2005).  Land use determines the type and frequency of human disturbance to which the patches are subjected, particularly near the edges.  These can include relatively low-disturbance activities like extensive grazing, or more intensive disturbances like urban expansion.  The role of plant life-history traits in determining plant community changes in response to changing landscape context is not well understood (Aggemyr and Cousins, 2012; Lindborg, 2007).   Many studies have examined the effects of current landscape context on the presence or species richness of plant species (e.g. Cousins and Aggemyr, 2008; Digiovinazzo et al., 2010; Dupr? and   59Ehrl?n, 2002; Heegaard et al., 2007; Kolb and Diekmann, 2005; Marini et al., 2012; Petit et al., 2004; Tremlova and Munzbergova, 2007).  These studies have found that relationships between plant species richness and area, connectivity, or shape of the patch vary for different plant community types, different regions, and plants with different traits.   However, as Dupr? and Ehrl?n (2002) point out, a lack of correlation between current landscape context and the presence or absence of a particular species may simply reflect non-equilibrium conditions resulting from time lags in colonization and/or extirpation.  To test for such lags, some researchers have used historical photographs or maps to reconstruct landscape configuration at some point in the past, predicting stronger correlation strength between presence or richness of species with landscape conditions in the past compared to the present (e.g. Aggemyr and Cousins, 2012; Bagaria et al., 2012; Cousins et al., 2007; Helm et al., 2006; Lindborg, 2007; Lindborg and Eriksson, 2004; Piqueray et al., 2011a).  In addition, historical information has given researchers the ability to classify habitat patches by their age (if some were created on formerly disturbed sites), and to determine differences in plant community composition based on the length of time a patch has existed, or former land use (e.g. Jacquemyn et al., 2001, 2003; Piqueray et al., 2011b; Vellend et al., 2007; Verheyen et al., 2003).  Some, but not all, of these studies have found a stronger dependence of current plant species richness on historical compared to current landscape conditions (e.g. Helm et al., 2006; Lindborg and Eriksson, 2004; Bagaria, 2012; Cousins et al., 2007), and an increasing role of landscape context in determining plant community composition in fragmented areas (e.g. Foster et al., 1998; Rogers et al., 2009; Vellend et al., 2007).  However, relating the current presence of a species to landscape conditions in the past or present remains problematic because presence as a response may reflect any one of (a) resilience to land cover change, i.e. the long-term ability of the species to persist despite landscape changes, (b) a time-  60lag between changes on the landscape and the resulting extirpation of the species, or (c) its suitability to landscape conditions in the present only.  Some estimate of colonizations and extirpations along with presence/absence data could provide a stronger indication of plant community responses to landscape-level disturbances, and which life-history traits may mediate these responses, but such data are rarely available (Fischer and Stocklin, 1997; Rogers et al., 2009; Sutton and Morgan, 2009; Williams et al., 2005).  In Chapter 3, I reported changes in plant community diversity and composition on the Saanich peninsula determined by resurveying 184 vegetation plots first surveyed in the late 1960s.  This landscape has seen striking increases in local and regional plant species diversity, with the most successful species tending to be tolerant of disturbance (Chapter 3; McCune and Vellend, 2013).  In this chapter, I test the ability of landscape context ? including amount of naturally vegetated area, shape of the patch, and surrounding road density - to predict colonizations and extirpations.  I use the data to ask three broad questions:  (1) How has the landscape changed quantitatively over the past four decades? (2) Is landscape context now more important relative to environmental factors in determining plant community composition than it was in the past? and (3) Are the colonizations and extirpations that occurred over four decades predictable based on plant traits and landscape context?  4.3 Materials and methods 4.3.1 The study area The Saanich Peninsula is located on the southeastern tip of Vancouver Island (Fig. 4.1).  Since settlement by Europeans in the mid-1800s steady growth in agriculture, logging, and suburban    61 Figure 4.1:  Inset: location of the Saanich Peninsula on southeastern Vancouver Island.  Centre: The Saanich Peninsula extends north from the city of Victoria.  Black dots indicate relev? plot locations.  Light grey shading indicates residential, commercial, industrial or institutional land use.  Dark grey shading indicates agricultural land use.  Unshaded regions of the peninsula are parks or other green spaces, military land, or First Nations reserve land. Right: Aerial photographs from 1964 and 2005 show landscape context within 500m of a relev? plot.  Road density increased from 1.4km to 5.8km, total naturally vegetated area declined from 86% to 40%, and shape index declined from 0.68 to 0.34.   development has resulted in a highly fragmented landscape.  Douglas-fir forests are the dominant vegetation, characterized by a canopy of Douglas-fir (Pseudotsuga menziesii), sometimes mixed with Western red cedar (Thuja plicata) or grand fir (Abies grandis).  In the driest forests, Pacific   62madrone (Arbutus menziesii) grows interspersed with Douglas-fir (Flynn, 1999; Meidinger and Pojar, 1991).  The vegetation also includes oak woodlands and savannahs characterized by open grassland and/or shrubs and a sparse canopy of Garry oak trees (Quercus garryana).  In the late 1960s, about one third of the peninsula outside the city of Victoria still supported forest or savannah vegetation (Roemer, 1972).  Much less than that remains today (Fig. 4.1).    Most of the Saanich peninsula consists of gently rolling land with brunisolic or gleysolic soils formed on unconsolidated glacial till (Day et al., 1959; Jungen, 1985), but the landscape is punctuated by a number of small, rocky hills.  The study plots range in elevation from sea level to 325m above sea level.  Climate correlates strongly with the elevation gradient:  the higher elevations are slightly colder and receive more precipitation (Lilley and Vellend, 2009).  Elevation also correlates quite strongly with land use, as agricultural and residential land use is primarily concentrated on the flatter areas at lower elevations (Vellend et al., 2008).    4.3.2 Resurvey methods As described in Chapter 3, in 1968 and 1969 (hereafter 1968) Hans Roemer recorded the abundance of all vascular and non-vascular plant species in 409 relev? plots of 400m2 area (Roemer, 1972).  The plots were not permanently marked, however for each plot a description of its location, slope and aspect and a grid reference on an aerial photo (1:10,000 scale) were recorded.  In 2009, I relocated and resurveyed 184 of the original relev?s.  I resurveyed only where sites were still vegetated and where permission could be obtained from municipal or provincial park   63authorities or private landowners, and I limited the resurvey to non-riparian vegetation types (i.e. the dominant vegetation types described above).  The aerial photo grid reference system, in conjunction with GPS, allowed me to navigate to within approximately 100m or less of the plot location.  I then selected the most likely location of the original plot using the original notes on slope, aspect, and nearby features, and advice on interpreting these notes from the original surveyor (H.L. Roemer). I recorded the abundance of all vascular and non-vascular plant species to the nearest one percent cover.     4.3.3 Landscape data I obtained aerial photographs from just before each survey period.  For the 2009 survey, I obtained digitized, orthorectified colour aerial photographs taken in 2005 from the regional government.  For the 1968 survey, I obtained copies of black and white aerial photos taken in 1964 (1:10,000), which were the air photos used by the original surveyor in 1968.  I scanned, imported and orthorectified the 1964 air photos into ArcGIS (ArcGIS 9.3; ESRI, Redlands, CA, USA).  I defined a buffer area of 500m radius around each relev? plot (Fig. 4.1).  An earlier study in the same region showed that the density of roads within 500m of a focal patch is a good predictor of exotic species richness, and that the density of roads within 100m, 500m and 1000m are strongly correlated (Lilley and Vellend, 2009).  Within each 500m buffer, I digitized all roads and naturally vegetated areas in both years (1964 and 2005).  Naturally vegetated areas include either forest or savannah, but exclude agricultural fields, playing fields, cemeteries and built up areas.  I then calculated three measures of landscape context:  the total length of roads within 500m, the   64total amount of naturally vegetated area within 500m, and the shape index of the naturally vegetated patch in which the plot was located.  I used the following index for patch shape:  shape = 2?(?A)/P, where A is the area of the patch and P is its perimeter (Digiovinazzo et al., 2010).  Therefore, a perfectly circular patch would have a shape index of 1, with values approaching 0 having decreasing area to perimeter ratio (i.e. increased ?edginess?).  Note that plots situated within large patches whose edges were beyond the 500m buffer would have a shape index value of 1, even if the distant edges were convoluted.  I also calculated the area of the patch each plot was in (which was necessary to calculate the shape index), but since patch area was highly correlated with total naturally vegetated area (Table 4.1), I used only the latter for the analyses.  The three landscape variables together represent both disturbance type and landscape configuration.  Road density represents the type of surrounding disturbance, with high road density indicating urban or suburban land use.  Total naturally vegetated area gives an indication of patch area and isolation from other naturally vegetated areas.  Shape index is an indication of the opportunity for edge effects to dominate.  The three landscape context variables are correlated (Table 4.1), with higher road density, lower naturally vegetated area, and lower shape index generally indicating higher levels of surrounding disturbance and fragmentation.  However, they are not completely linked; for example, plots with a low shape index can have high or low road density depending on the surrounding land use.  The maximum correlation was 0.72 (shape index vs. naturally vegetated area), indicating that almost half or more of the variance in any one variable across the landscape occurred independently of the other variables (r2 = 0.722 = 0.52).  For this reason, I examined the patterns across each variable separately, and looked for patterns across all three.   65 Table 4.1: Pearson?s correlation coefficients between landscape context variables measured within 500m of each plot in 1964 and 2005. 1964 2005  road density naturally vegetated area patch area  road density naturally vegetated area patch area naturally vegetated area -0.63   naturally vegetated area -0.67   patch area -0.56 0.91  patch area -0.64 0.96  shape index -0.48 0.72 0.65 shape index -0.33 0.59 0.62  4.3.4 Species and environmental data I standardized plant nomenclature and combined taxa at the level of genus where the accuracy of species identifications was in doubt.  Plant nomenclature follows Douglas et al. (1998).  I focused analyses on vascular understorey species only, including grasses, forbs, ferns, and shrubs, which represent ~80% of the full set of species observed.   I calculated two species by plot matrices:  one for colonizations and one for extirpations.  In the colonization matrix, ?0? represents absence in both years, and ?1? represents presence in 2009 only (colonization).  Species present in a plot in both years were coded ?NA?.  Similarly, in the extirpation matrix, ?0? represents presence in both years, and ?1? represents presence in 1968 only (extirpation).  Species absent in both years were coded ?NA?.  I limited the colonization and extirpation dataset to species that were present in at least 10% of all 368 plot-year combinations (77 species), and to species that colonized or were extirpated from at least 5 plots.  This resulted in a colonization dataset of 75 species and an extirpation dataset of 57 species out of the total 266 understorey species that were observed in both years pooled.  I also created a species by plot matrix for presence/absence in each year.  For presence/absence I limited the dataset for each   66year to species present in at least 10% of the 184 plots in that year.  This resulted in datasets of 63 species for 1968 and 79 species for 2009.  Because the plots were not permanently marked, some apparent colonizations or extirpations may result from imprecise relocation of the plot or from false absences in either year due to surveyor error (?pseudo-turnover?, sensu Fischer and Stocklin, 1997). I also acknowledge that apparent colonization could result from the germination of a species that was present only in the seed bank in 1968, and species appearing to be extirpated may still be present in the seed bank in 2009.  However, I am testing only for patterns across all 184 plots as a whole, rather than attempting to draw inferences about individual plots, and I have no reason to expect the incidence of ?pseudo-turnover? to be correlated with landscape context.  I measured the slope and aspect of each plot in the field, estimated the average soil depth as one of four categories, and estimated the percent cover of exposed rock.  I determined the elevation of each plot via GIS.  I determined a measure of soil type from a map of Canada Land Inventory soil classes, which map soils into one of seven broad categories based on potential for agricultural use (National Archives of Canada, 1999).  4.3.5 Trait data In order to test for the role of plant traits in determining colonizations/extirpations in relation to landscape context, I compiled trait data for each understorey species.  I define ?traits? as both qualitative and quantitative characteristics of plant species and/or their preferred habitats.  I chose nine traits hypothesized to be important in the ability of a plant species to maintain its   67population in highly disturbed and fragmented landscapes.  I compiled all trait data from the Electronic Atlas of the Flora of British Columbia (E-Flora; Klinkenberg, 2013), the USDA Plants database (USDA, NRCS, 2011), the Kew seed information database (Royal Botanic Gardens Kew, 2008), and two previous studies in the same region (Bennett et al., 2013; Lilley and Vellend, 2009), except for specific leaf area, which I obtained from an unpublished database of standardized measurements compiled by Dr. Will Cornwell.  Values for each trait were available for at least 77% of all species (see Results).  Origin and disturbance tolerance Previous studies have found that native species and introduced exotic species respond in opposite ways to human disturbance and climate, with exotic species generally favoured by human disturbance (Lilley and Vellend, 2009; McIntyre and Lavorel, 1994).  I coded each species as ?exotic? or ?native?.  I also assessed disturbance tolerance for each species from the ?Habitat/Range? description in E-flora.  If any of the following were encountered in the habitat description, the species was coded ?disturbance tolerant?:  burns, continuously disturbed sites, disturbed areas/sites, disturbed communities, ditches, fencelines, fencerows, fields, fire-disturbed sites, lawns, pastures, railways, roadsides, waste places, early-seral/young-seral/early successional forests, garden, or garden escape.  A few rare garden escapes not given habitat data in E-flora were coded as ?disturbance tolerant?.  All others were classified ?disturbance intolerant?.  I hypothesized that exotics and disturbance tolerant species would be more likely to colonize and less likely to be extirpated from plots with higher surrounding road density, lower total naturally vegetated area, and lower shape index (i.e. higher surrounding disturbance and fragmentation).   68 Longevity and form The ability to persist in isolated fragments is often higher in longer-lived species (Miller et al., 2012; Piqueray et al., 2011a; Purschke et al., 2012).  However, annual species are the most able to take advantage of the unstable, high nutrient conditions of highly disturbed areas (Knapp et al., 2008; Marini et al., 2012; McIntyre et al., 1995; Vallet et al., 2010). I classified each species according to its life form: herbs (including grasses, forbs, and ferns) and shrubs.  I also compiled a coarse categorical representation of lifespan: ?annual? (including annual and biennial species) and ?perennial?.  I predicted that annuals and herbaceous species would be more likely to colonize and less likely to be extirpated from plots with higher surrounding disturbance and fragmentation.   Shade tolerance   Increased fragmentation can lead to more high light, forest edge conditions, favouring shade intolerant species (e.g. Metzger, 2000).  Therefore, I put each species into one of two shade tolerance categories:  shade tolerant (very shade tolerant, shade tolerant, or tolerant/intolerant), or shade intolerant (shade intolerant or very shade intolerant) based on the description in E-Flora (Klinkenberg, 2013) or the USDA Plants database (USDA, NRCS, 2011).  I predicted that shade intolerant species would be more likely to colonize and less likely to be extirpated from plots with higher surrounding disturbance and fragmentation. .    69Nutrient regime preference The increase in roads and fossil-fuel powered vehicles causes eutrophication of terrestrial systems via nitrogen deposition (Trombulak and Frissell, 2000).  I assessed the preferred soil nutrient status of each species using the ?Ecology? section of E-Flora, which provides the ?modal nutrient regime class? for many plant species as calculated from data provided by the British Columbia Ministry of Forests and Range.  This value represents the average nutrient availability of all survey plots (out of over 30,000 across British Columbia) in which a species was present, with values ranging from ?1? (very poor, oligotrophic) to ?6? (saline, hypereutrophic).  I reduced this to two categories, ?low nutrient? (1 to 3) and ?high nutrient? (4 to 6) in order to avoid categories with very few species.  I predicted that high nutrient regime-associated species would be more likely to colonize and less likely to be extirpated from plots with higher surrounding disturbance and fragmentation.   Dispersal ability The ability for plant species to survive in fragmented habitats is also related to their ability to disperse propagules successfully between patches (Marini et al., 2012; Purschke et al., 2012).  Dispersal ability is determined in part by the mode of seed dispersal, so I classified each species into one of three seed dispersal categories:  (1) no seed dispersal mechanism, ballistic dispersal or dispersal by ants, (2) wind dispersal, and (3) dispersal by vertebrates.  Seed mass is also linked to colonization ability, with plants producing smaller seeds able to produce more of them (Eriksson and Jakobsson, 1998; Westoby et al., 2002).  I compiled values for seed weight from the Kew seed information database (Royal Botanic Gardens Kew, 2008), which reports the average weight of 1000 seeds in grams.  The role of dispersal in plant success in fragmented   70landscapes is not entirely clear.  Many studies have found extirpation more likely for wind dispersed species and less likely for animal dispersed species in urban and/or fragmented landscapes (e.g. Knapp et al., 2008; Marini et al., 2012; Piqueray et al., 2011a; Sutton and Morgan, 2009; Williams et al., 2005).  Long-distance dispersal by wind may become a liability when suitable habitat patches are few and far between (Saar et al., 2012; Sutton and Morgan, 2009).  However, other studies have noted negative effects of fragmentation on animal-dispersed species (e.g. Cordeiro and Howe, 2001; Santos et al., 1999).  Results of studies examining seed mass also vary, with some showing no relationship between landscape context and seed mass (e.g. Bagaria et al., 2012; Dupr? and Ehrl?n, 2002; Vallet et al., 2010; Lindborg, 2007; Piqueray et al., 2011a), while others find species with heavy seeds to be favoured in fragmented and/or urban environments (e.g. Duncan et al., 2011; Kolb and Diekmann, 2005; Saar et al., 2012).  This may be because seed mass is not only associated with potential dispersal distance, but also with establishment success and seed banking potential (Lindborg, 2007; Westoby et al., 2002).  I did not make predictions for the relationship between seed mass or dispersal mechanism and colonizations or extirpations of plots with varying landscape context.  Specific leaf area Finally, specific leaf area (SLA; leaf area per unit dry mass) has been found to correlate with gradients of nutrient and water availability, with high SLA being associated with fast-growing species in productive conditions, and low SLA being associated with more stress-tolerant species (Westoby, 1998).  Plants with higher SLA are favoured in fragmented, urban environments (Knapp et al., 2008; Vallet et al., 2010; but see Piqueray et al., 2011a). I predicted that species   71with higher SLA would be more likely to colonize and less likely to be extirpated from plots with higher surrounding disturbance and fragmentation.     Trait correlations There has been some debate about whether combining species into emergent groups based on traits is a better strategy than testing individual traits on their own (e.g. Vallet et al., 2010; Ockinger et al., 2010).   Some of my chosen traits are correlated.  The strongest correlations are between form and dispersal mechanism (shrubs often have vertebrate dispersed seeds, r=0.62) and origin and disturbance tolerance (exotics are almost always disturbance tolerant, r=0.60).  All other correlations between traits are less than 0.45 (Table 4.2).  Given that most of the variance in any one trait is independent of any one other trait (maximum r2 = 0.622 = 0.38), and that I wished to test hypotheses about individual traits, I decided to test individual traits rather than emergent groups.  However, I recognize that significant trait responses cannot be definitively attributed to one trait, if that trait is highly correlated with one or more other traits, measured or unmeasured.   4.3.6 Statistical analyses To quantify changes in the three landscape disturbance measures over time, I used paired Wilcoxon signed rank tests to check for significant increases in road density, decreases in total naturally vegetated area and decreases in shape index.        72Table 4.2: Pearson?s correlation coefficients between plant traits for the colonization dataset (n=75).  Correlations are similar for the extirpation and presence/absence datasets.  Disturbance tolerance Origin Form Lifespan Shade tolerance Nutrient Regime Dispersal mechanism Seed weight Origin  -0.60        Form  0.16 -0.01       Lifespan  -0.28 0.37 0.26      Shade tolerance 0.003 -0.17 -0.21 -0.10     Nutrient regime 0.40 -0.24 -0.016 -0.13 -0.26    Dispersal mechanism 0.24 -0.14 0.62 0.12 -0.15 -0.08   Seed weight  0.19 -0.15 0.35 0.12 -0.11 0.16 0.23  SLA  0.15 -0.21 -0.41 -0.41 -0.12 0.16 -0.32 -0.17  I used distance-based redundancy analysis (dbRDA; Anderson, 2001b; McArdle and Anderson, 2001) to quantify the relative strength of correlation of environmental variables and landscape context with plant community composition in order to test whether landscape context has become a stronger determinant of plant community composition since the initial survey.  dbRDA models the relationship between plant community composition, as represented by a dissimilarity matrix, and predictor variables using a multiple regression approach (Anderson et al., 2008).  I created a plot by species matrix for each year based on the recorded abundances of all plant species, and used the Bray-Curtis dissimilarity measure to compute dissimilarities between all possible pairs of plots.  Then I used the DISTLM function in PRIMER version 6 with PERMANOVA+ (Clarke and Gorley, 2006) to model plant species composition in each year based on environmental variables and landscape factors.  Candidate environmental explanatory variables included soil type, soil depth, slope, aspect and elevation.  I excluded rock cover from   73the analysis as it was highly correlated with soil depth.    Landscape variables included the three measures of landscape context within 500m of each plot.    In order to determine whether environmental or landscape variables were stronger predictors of plant community composition, I first grouped the explanatory variables by type, and used model selection with AICc (Akaike?s Information Criterion, corrected for finite sample size) to compare the strength of relationship (separately for each year) between plant community composition and all environmental variables versus all landscape variables.  I also compared the ability of all three landscape factors in 2005 with all three landscape factors in 1964 to explain plant community composition in 2009.  Then, I used the same method to compare a model with only environmental variables to a model with environmental and landscape variables to determine whether including landscape variables significantly improved the model.  Finally, I determined the best single environmental and the best single landscape variable for each year by comparing the AICc for models based on one variable at a time.  I considered a model to be substantially stronger than another model if the difference in AICc was 5 or greater, which corresponds to a strength of evidence at least 12 times stronger (Anderson, 2008; Burnham et al., 2011).    I used logistic regressions to relate three binary variables ? absence vs. presence (in each year separately), continued absence vs. colonization, or continued presence vs. extirpation ? to the three landscape context variables in each year.  For each species and each landscape variable (road density, naturally vegetated area, and patch shape) in each year I built logistic regression models as follows:   74 Binary variable              ~         elevation     +       landscape disturbance measure       I included elevation in the model to control for climatic determinants of species distributions, as elevation has previously been shown to be a strong correlate of climate in this region (Lilley and Vellend, 2009).    I checked for spatial autocorrelation in the residuals of these logistic regression models using Moran?s I.  For all three dependent variables (colonization, extirpation, and presence/absence) a significant proportion (73%, 25% and 95%, respectively) of species showed evidence of spatial autocorrelation in the model residuals.  I examined spline correlograms (Bjornstad and Falck, 2001) for a subset of species, and determined that significant positive spatial autocorrelation tended to occur at lag distances of 1km or less.  To account for this, I defined 22 groups of relev? plots for which there was overlap of their 500m buffers (e.g. they were within 1km or less of one or more other plots in the group).  I then built models for each species, as above, relating colonization, extirpation, or presence/absence to the landscape variables using generalized linear mixed effects models (GLMMs) with spatial group as a random factor.    After running the GLMMs, I extracted the regression coefficients for each species/disturbance measure combination.  Positive coefficients indicate a positive association between that species? presence (or colonization or extirpation) with the landscape variable (e.g. road density within 500m); negative coefficients indicate a negative association.  The calculation of coefficients for multiple species makes it likely that some significant correlations were obtained by chance (Dupr? and Ehrl?n, 2002).  However, my goal was not to analyze the response of individual   75species, but to find out whether the tendency of species to have colonizations or extirpations positively or negatively associated with landscape context was predictable based on traits.  Therefore, I included all species in the subsequent analysis.    Finally, I tested for significant relationships between a tendency for positive versus negative association of colonization, extirpation, or presence of species with landscape context and species? traits.  I reduced the logistic regression coefficients to a categorical variable with two categories:  negative or positive.  I then used Fisher?s exact tests (for categorical traits), or Wilcoxon?s rank sum tests (for the continuous traits seed weight and SLA) to test whether there were significant differences in traits of species having a positive versus a negative response.  I used one-sided tests for traits for which I had a clear prediction for the expected direction of the relationship, and two-sided tests for seed weight and dispersal mechanism.  I also tested whether the evolutionary relatedness of my pool of species influenced the results of the trait tests, as some studies have found contrasting results after controlling for phylogeny (e.g. Tremlova and Munzburgova, 2007).  I first built linear models for each test of trait versus logistic regression coefficient category.  These models are not ideal for ordinal variables, but they produced results practically identical to the Fisher?s exact tests and Wilcoxon?s rank sum tests, above.  Then I built phylogenetic generalized linear models (pgls), which fit a linear model that controls for the phylogenetic relatedness of the species (Orme et al., 2013).  I set the parameter lambda to its maximum likelihood estimate, as recommended by Freckleton et al. (2002).  To estimate phylogenetic relatedness, I used Phylomatic (Webb and Donoghue, 2005) to   76obtain a tree for each set of species (colonization, extirpation, and presence/absence) using the Phylomatic tree R20120829 for plants.  I set branch lengths to one.  All analyses except for the DISTLM models were carried out in R (R Core Development Team, 2012), using the ?ape? package for calculating Moran?s I, the ?ncf? package for producing correlograms, the ?glmmML? package for building GLMMs, and the ?ape? and ?caper? packages for phylogenetic analyses (Bjornstad, 2012; Brostr?m and Holmberg, 2011; Orme et al., 2013; Paradis et al., 2013).  4.4 Results 4.4.1 Overall changes in diversity There were 179 understorey species recorded in 1968, and 237 species in 2009.  Apparent colonizations were more frequent than extirpations.  Eighty-seven species were found only in 2009, while only 29 of the original species were not recorded in any plot in 2009.  On average each plot gained 15 species (9 native and 6 exotic), and lost 7 species (just over 6 native and less than one exotic; see Chapter 3).  4.4.2 Landscape changes Wilcoxon signed rank tests showed significant changes in all three landscape context measures between the 1964 and 2005 aerial photos (Fig. 4.2).  Most striking was the change in road density, which went up on average 62% from a mean of 2.2 km of roads in 1968 to a mean of 3.6 km of roads in 2009.  Very few plots saw a reduction in road density (Fig. 4.2a).  Total naturally vegetated area within 500m declined by an average of 12%, from a mean of 74% of the buffer   77around each plot down to a mean of 65%.  Most plots saw a decline in total naturally vegetated area, but a few plots saw gains in surrounding vegetation (Fig. 4.2b).  Similarly, the shape index of the vegetation patches in which plots were located declined from an average index value of 0.62 to a value of 0.44, indicating a decline in area:perimeter ratio.   Again, the majority of plots saw a decline in shape index, although a few showed an increase (Fig. 4.2c).     Figure 4.2:  Change over 40 years in (a) total length of roads within 500m of each plot, (b) percent of area naturally vegetated within 500m of each plot, and (c) shape index (closer to 1 means higher area:perimeter ratio).  In each case the dashed line is the 1:1 line.  Pearson correlation coefficients (r) are shown for each relationship.    4.4.3 Relative influence of local environment and surrounding landscape on community composition For both years the set of environmental variables was a stronger predictor of plant community composition than the set of landscape context variables (Table 4.3).  The best lone predictor of plant community composition in both years from among all environmental variables was soil depth, with all other potential predictors having ?AICc greater than 16.  The strongest predictor   78of plant community composition from amongst the three landscape variables alone was road density for 1968 and total naturally vegetated area for 2009, however the two other landscape variables alone had AICc values within 2 of the best model for both years.  When I compared models with only the five environmental variables to models with all eight variables (landscape and environmental), the landscape variables improved the model very slightly for 1968 data (?AICc=1.5, r2 improved from 14.8% to 18.5%) and more for 2009 data (?AICc=4.9, r2 improved from 18.4% to 23.3%).  Landscape context in 1964 was not a better predictor of present plant community composition than landscape context in 2005.  In fact, the model based on 2005 landscape variables had more support (Table 4.3).  Table 4.3:  The results of distance-based linear models relating environmental variables and landscape context to plant community composition. YEAR # of variables Explanatory variables r2 AICc   All environmental versus all landscape context   1968 5 soil type + soil depth + elevation + slope + aspect 0.148 1492.1 3 road density +  area naturally vegetated + shape index (1964) 0.043 1509.42009 5 soil type + soil depth+ elevation + slope + aspect 0.184 1466.6 3 road density + area naturally vegetated + shape index (2005) 0.069 1486.7 3 road density + area naturally vegetated + shape index (1964)  0.053 1489.8  Environmental variables only compared to all variables   1968 5 all environmental variables 0.148 1492.1 8 all environmental variables + all landscape variables (1964) 0.185 1490.62009 5 all environmental variables 0.184 1466.6 8 all environmental variables + all landscape variables (2005)  0.233 1461.7  Best single predictors, landscape   1968 1 road density 1964 (all others with ?AICc < 2) 0.023 1509.12009 1 area naturally vegetated 2005 (all others with ?AICc < 2)  0.031 1489.7  Best single predictors, environmental   1968 1 soil depth (all others with ?AICc > 16) 0.106 1492.62009 1 soil depth (all others with ?AICc > 18) 0.125 1470.9     794.4.4 Relating traits to presence, colonizations, and extirpations In the colonization dataset, 14 of 75 species had significant coefficients relating colonization to one or more landscape context variables in 1964.  Thirty-seven of 75 had significant coefficients in relation to 2005 landscape context.  In the extirpation dataset, 17 of 57 species had significant coefficients relating extirpation to one or more landscape context variables in 1964.  Only 11 of 57 had significant coefficients based on landscape variables in 2005.  For presence/absence, 20 of 63 species in the 1968 dataset had a significant relationship between presence and one or more landscape context variables in 1964, 31 of 79 species in 2009 had significant relationships with at least one 1964 landscape context variable, and 50 of 79 species in 2009 had a significant relationship of presence with at least one 2005 landscape context variable.   Fisher?s exact tests and Wilcoxon rank sum tests supported some of my predictions for the role of traits in mediating the relationship of colonizations and extirpations with landscape context (Table 4.4, 4.5).  For example, the colonization of herbs was more likely than shrubs to be positively related to a higher road density, while shrubs were more likely than herbs to colonize plots in areas with more naturally vegetated area and higher shape index (higher area:perimeter ratio).  Significantly more shade intolerant species than shade tolerant species had positive relationships between colonization and road density in 2005, whereas shade tolerant species were more likely to colonize plots with a higher shape index in 1964.  Natives and disturbance intolerant species were more likely to colonize plots with a higher shape index in 1964 (Table 4.4, Fig. 4.3).  Traits did not predict the nature of the relationship between colonizations over 40 years better based on landscape conditions in 1964 compared to 2005 except in the case of patch shape.  Six of nine traits predicted the relationship of colonization to patch shape in 1964 as   80expected, whereas only three traits predicted the relationship based on shape in 2005 as expected, and one (nutrient regime) was opposite to my prediction (Table 4.4, Fig. 4.3, 4.4).    Overall, plant traits were less useful in predicting extirpations than colonizations based on landscape context (Table 4.5).  Extirpations followed our predictions based on form, lifespan and SLA in a few cases, but most tests revealed no significant differences.  The relationship of extirpations to landscape context was never predictable based on disturbance tolerance, origin, shade tolerance, nutrient regime, dispersal mechanism or seed weight.  The relationship between species presence and landscape context in terms of plant traits showed similar patterns to colonization (Table 4.6).  The relationships of species? presence in 2009 to landscape conditions in both 1964 and 2005 were quite consistently predictable based on origin, form, lifespan, shade tolerance, and SLA.  However, as for colonizations, the relationship of presence to shape index in 2005 showed fewer predicted relationships based on traits than shape index in 1964 (Table 4.6).    The results of tests using phylogenetic generalized least squares models to control for evolutionary relatedness of species were very similar to the initial results with no phylogenetic correction.  Approximately 12 percent of trait tests yielded a different result (e.g. significant effect when previously not significant, or vice versa), but no relationship between trait category and model coefficient changed direction (see Appendix A.3).    814.5 Discussion 4.5.1 The changing landscape The landscape of the Saanich peninsula has undergone striking changes in the past four decades, with a decline in naturally vegetated area, an increase in edge length relative to the area of forest patches, and an enormous increase in roads.  Deforestation over the past four decades has almost always made way for suburban development, with some former agricultural land also being developed for housing.  Despite these substantial changes, environmental factors, in particular soil depth, are still greater determinants of plant community composition than landscape context.  Adding landscape variables to a model with all environmental variables improved the model more for 2009 compared to 1968 data (Table 4.3), which suggests an increase in importance of landscape context in 2009 compared to 1968.  But environmental variables are still a much stronger determinant of plant community composition.  Given the wide gradient of vegetation and edaphic conditions sampled by my plots, it is not surprising that soil depth is the most important determinant of community composition.   It may be that the influence of landscape context will continue to increase.  Perhaps change in plant community composition resulting from recent drastic landscape changes has not yet been fully realized, and landscape context will eventually become more important than local environmental conditions as a determinant of plant community composition.  Rogers et al. (2009) found that over a period of 55 years the composition of herbaceous understorey communities in fragmented Wisconsin forests became more strongly determined by landscape context than environmental conditions, but this was not so for shrub communities.  They speculated that shrubs respond more slowly due to slower   82population dynamics than herbaceous species.  However, when I repeated the dbRDA analysis with herbaceous species only, the addition of landscape context variables to a model with only the five environmental variables improved the model even less than it did with shrubs included (?AICc=0.9 for 1968 data, ?AICc=3.3 for 2009 data).  Therefore, the inclusion of shrub species does not explain why I did not see a dramatic increase in the importance of landscape context relative to environmental variables in determining plant community composition.   Unlike some other studies (e.g. Helm et al. 2006; Lindborg and Eriksson, 2004) that found current species richness was more strongly associated with past than current landscape context, I found no evidence that current plant community composition as a whole is more strongly determined by landscape conditions forty years ago than those in the present.  This may again reflect more recent landscape changes, as my study took place over only four decades compared to five to ten decades in those studies.  However, I did find some evidence for potential time lags in colonizations and extirpations, as described below.  83Table 4.4:  Predicted versus observed relationships between colonizations over 40 years, landscape structure within 500m of the relev?, and species traits.  Results are based on Fisher?s exact tests (categorical traits) or Wilcoxon rank sum tests (continuous traits) relating species traits to positive versus negative coefficients of generalised linear mixed models relating extirpation to landscape variables.  For non-significant (p >0.05) relationships, only the p-value is reported.  N=75 species for all traits except shade tolerance (N=60), nutrient regime (N=62), dispersal mechanism (N=70), seed weight (N=60) and SLA (N=64).  Results in bold indicate a relationship opposite to that predicted. PREDICTED TRAIT RELATIONSHIPS OBSERVED TRAIT RELATIONSHIPS Colonizations ~ road density Colonizations ~ road density 1964  Colonizations ~ road density 2005 Disturbance tolerant species +      Exotic species +                            Herbs +                                          Annual species +                         Shade intolerant species +            High nutrient regime species + Dispersal mechanism? Seed weight? High SLA species + p=0.2683 p=0.3108 Herbs + (p=0.0440) p=0.4210 p=0.1060 p=0.9601 p=0.3335 p=0.3079 p=0.0943 p=0.1910 p=0.1920 Herbs + (p=0.0000) Annual species + (p=0.0127) Shade intolerant species + (p=0.0195) p=0.5578 None/wind dispersed species + (p=0.0215) p=0.1322 High SLA species + (p=0.0215) Colonizations ~ area naturally vegetated Colonizations ~ area nat. veg. 1964 Colonizations ~ area nat. veg. 2005 Disturbance intolerant species + Native species + Shrubs + Perennial species + Shade tolerant species + Low nutrient regime species + Dispersal mechanism? Seed weight? Low SLA species +  Disturbance intolerant species + (p=0.0117) Native species + (p=0.0438) Shrubs + (p=0.0154) Perennial species + (p=0.0027) Shade tolerant species + (p=0.0394) p=0.5365 p=0.0517 p=0.2332 Low SLA species + (p=0.0467) Disturbance intolerant species + (p=0.0201) Native species + (p=0.0154) Shrubs + (p=0.0026) Perennial species + (p=0.0008) Shade tolerant species + (0.0001) p=0.9525 p=0.0735 p=0.1465 Low SLA species + (p=0.0093) Colonizations ~ shape (area:perimeter) Colonizations ~ shape 1964 Colonizations ~ shape 2005 Disturbance intolerant species + Native species + Shrubs + Perennial species + Shade tolerant species + Low nutrient regime species + Dispersal mechanism? Seed weight? Low SLA species +  Disturbance intolerant species + (p=0.0147) Native species + (p=0.0358) Shrubs + (p=0.0011) Perennial species + (p=0.0004) Shade tolerant species + (p=0.0472) p=0.6366 p=0.0753 p=0.1112 Low SLA species + (p=0.0102) p=0.7009 p=0.6877 Shrubs + (p=0.0092) Perennial species + (p=0.0408) p=0.5224 p=0.9934 High nutrient regime species + p=0.3600 p=0.8635 Low SLA species + (p=0.0022)    84Table 4.5:  Predicted versus observed relationships between extirpations over 40 years, landscape structure within 500m of the relev?, and species traits.  Results are based on Fisher?s exact tests (categorical traits) or Wilcoxon rank sum tests (continuous traits) relating species traits to positive versus negative coefficients of generalised linear mixed models relating extirpation to landscape variables.  For non-significant (p >0.05) relationships, only the p-value is reported. N=57 species for all traits except shade tolerance (N=50), nutrient regime (N=49), dispersal mechanism (N=54), seed weight (N=44) and SLA (N=49). Results in bold indicate a relationship opposite to that predicted. PREDICTED TRAIT RELATIONSHIPS OBSERVED TRAIT RELATIONSHIPS Extirpations ~ road density Extirpations ~ road density 1964 Extirpations ~ road density 2005 Disturbance intolerant species + Native species + Shrubs + Perennial species + Shade tolerant species + Low nutrient regime species + Dispersal mechanism? Seed weight? Low SLA species + p=0.6877 p=0.0122 Shrubs + (p=0.0005) Perennial species + (p=0.0424) p=0.1956 p=0.2794 p=0.4151 p=0.3896 Low SLA species + (p=0.0155) p=0.7517 p=0.1872 Shrubs + (p=0.0085) p=0.1047 p=0.2601 p=0.6602 p=0.0899 p=0.0821 p=0.4099 Extirpations ~ area naturally vegetated  Extirpations ~ area 1964 Extirpations ~ area 2005 Disturbance tolerant species + Exotic species + Herbs + Annual species + Shade intolerant species + High nutrient regime species + Dispersal mechanism? Seed weight? High SLA species +  p=0.3447 p=0.1264 Herbs + (p=0.0237) p=0.2541 p=0.5541 p=0.8196 p=0.4194 p=1 High SLA species + (p=0.0111) p=0.1339 p=0.2912 p=0.0542 p=0.4987 p=0.4459 p=0.8190 p=0.9407 p=1 p=0.0825 Extirpations ~ shape (area:perimeter) Extirpations ~ shape 1964 Extirpations ~ shape 2005 Disturbance tolerant species + Exotic species + Herbs + Annual species + Shade intolerant species + High nutrient regime species + Dispersal mechanism? Seed weight? High SLA species +  p=0.8777 p=0.2628 Herbs + (p=0.0168) p=0.4610 p=0.3800 p=0.8190 p=0.4089 p=0.2579 p=0.2896 p=0.9802 p=0.8550 Herbs + (p=0.0327) p=0.9251 p=0.7399 p=0.8595 p=0.0859 p=0.4094 p=0.5035   85Table 4.6:  Observed relationships between presence of species in each year, landscape structure within 500m of the relev?, and species traits.  Results are based on Fisher?s exact tests (categorical traits) or Wilcoxon rank sum tests (continuous traits) relating species traits to positive versus negative coefficients of generalized linear mixed effects models relating species? presence to landscape variables.  For non-significant (p >0.05) relationships, only the p-value is reported. For 1968, N=63 species for all traits except shade tolerance (N=55), nutrient regime (N=53), dispersal mechanism (N=58), seed weight (N=50) and SLA (N=57).  For 2009, N=79 species for all traits except shade tolerance (N=61), nutrient regime (N=67), dispersal mechanism (N=69), seed weight (N=65) and SLA (N=69). Results in bold indicate a relationship opposite to that predicted. TRAIT Presence 1968  ~ road density 1964 Presence 2009 ~ road density 1964 Presence 2009 ~ road density 2005 Disturbance tol. Origin Form Lifespan Shade tol. Nutrient regime Dispersal mech. Seed weight SLA p=0.6586 Exotic species + (p=0.0211) p=0.2605 p=0.0967 p=0.1817 p=0.3296 p=0.3047 p=0.2687 p=0.8108 p=0.2291 Exotic species + (p=0.0047) p=0.0785 p=0.0818 Shade intolerant species + (0.0102) p=0.5330 p=0.4286 p=0.4414 p=0.0928 p=0.1691 Exotic species + (p=0.0004) Herbs + (p=0.0004) Annual species + (p=0.0000) Shade intolerant species + (p=0.0002) p=0.4437 p=0.2619 p=0.1117 High SLA species + (p=0.0012)  Presence 1968 ~ naturally vegetated          area 1964 Presence 2009 ~naturally vegetated area 1964 Presence 2009 ~ naturally vegetated area 2005 Disturbance tol. Origin Form Lifespan Shade tol. Nutrient regime Dispersal mech. Seed weight SLA p=0.6246 p=0.0585 Shrubs + (p=0.0012) Perennial species + (p=0.0069) p=0.1578 p=0.2048 p=0.5983 p=0.8556 Low SLA species + (p=0.0257) p=0.0637 Native species + (p=0.0093) Shrubs + (p=0.0171) Perennial species + (p=0.0060) Shade tolerant species + (p=0.0435) p=0.8665 p=0.7808 p=0.3445 Low SLA species + (p=0.0131) p=0.3695 Native species + (p=0.0025) Shrubs + (p=0.0001) Perennial species + (p=0.0001) Shade tolerant species + (p=0.0003) p=0.9258 p=0.1156 p=0.0506 Low SLA species + (p=0.0012)  Presence 1968 ~ shape 1964 Presence 2009 ~ shape 1964 Presence 2009 ~ shape 2005 Disturbance tol. Origin Form Lifespan Shade tol. Nutrient regime Dispersal mech. Seed weight SLA p=0.4667 p=0.1236 Shrubs + (p=0.0001) Perennial species + (p=0.0003) Shade tolerant species + (p=0.0019) p=0.3737 p=0.2338 p=0.2237 p=0.1263 Disturbance intol. spp. + (p=0.0418) Native species + (p=0.0013) Shrubs + (p=0.0013) Perennial species + (p=0.0001) Shade tolerant species + (p=0.0218) p=0.9334 p=0.3195 p=0.0862 Low SLA species + (p=0.0009) p=0.8382 p=0.5334 Shrubs + (p=0.0499) Perennial species + (p=0.0235) Shade tolerant species + (p=0.0033) p=0.9705 p=0.1165 p=0.1356 Low SLA species + (p=0.0017)   86   Figure 4.3:  (a)-(g): The percentage of species in different trait categories with a positive association between colonizations and shape index in 1964.  A positive relationship indicates a tendency to have colonized plots in patches that had a higher area:perimeter ratio in 1964.  Bar widths are proportional to the number of species in each trait category, dashed lines indicate the percent of positive relationships across all species.  Grey bars indicate a significant difference between trait categories.  (h)-(i): The average seed weight and specific leaf area (SLA) of plant species with a positive (pos) versus negative (neg) association between colonizations and shape index in 1964.  Box widths are proportional to the number of species with positive or negative associations.  Note that the y-axes are shown on a log scale.  SLA (grey boxes) is significantly lower for species having a positive association.    87  Figure 4.4:  (a)-(g): The percentage of species in different trait categories with a positive association between colonizations and shape index in 2005.  A positive association indicates a tendency to have colonized plots in patches that had a higher area:perimeter ratio in 2005.  Bar widths are proportional to the number of species in each trait category, dashed lines indicate the percent of positive associations across all species.  There was no significant difference in positive versus negative associations for categories of any trait. (h)-(i): The average seed weight and specific leaf area (SLA) of plant species with a positive (pos) versus negative (neg) association between colonizations and shape index in 2005.  Box widths are proportional to the number of species with positive or negative associations.  Note that the y-axes are shown on a log scale.  There was no significant difference in the average SLA or seed weight of species with positive versus negative relationships.  884.5.2 Predictability of colonizations and extirpations In Chapter 3, I documented a large increase in alpha diversity, with species richness going up by 8 species on average between 1968 and 2009 (McCune and Vellend, 2013).  The average number of apparent colonizations was more than double the number of apparent extirpations per plot.  The ?winners? over four decades have tended to be exotic and disturbance tolerant, suggesting that human disturbance of the landscape is likely the largest driver of these changes - but the landscape is not homogeneous.  The next step is to find out whether the success of species with different traits is predictable based on the degree of disturbance in the immediate vicinity of the plot.  The tendency for colonizations to be positively or negatively related to landscape characteristics was predictable based on traits.  Disturbance tolerant, exotic, herbaceous annuals with intolerance for shade and higher SLA were more likely to colonize areas with higher surrounding disturbance and/or fragmentation (Table 4.4).  The degree of predictability based on traits was greater based on 2005 road density compared to 1964 road density, and was the same for naturally vegetated area in either year, but more traits predicted colonization based on 1964 shape index compared to 2005 shape index (Table 4.4; Fig. 4.3, 4.4).  Similarly, the presence of disturbance intolerant species and native species in 2009 tended to be positively associated with a greater shape index in 1964, but these traits could not predict 2009 presences based on the patch shape in 2005 (Table 4.6).  This could be a result of a time lag in the response of colonizations to landscape context. Road density and total vegetated area are highly correlated between the two years, making it difficult to find an affect of time, whereas shape index is the   89least correlated between 1964 and 2005 (Fig. 4.2c).  This may be why I did not see evidence for a time lag in the data for road density or naturally vegetated area.  It is also interesting that most species showed a negative relationship between apparent colonizations and shape index in 2005:  that is, the majority of colonizations have occurred in places that now have a relatively higher perimeter:area ratio, regardless of traits (Fig. 4.4).  For example, native species are just as likely as exotic species to have colonized currently more ?edgy? areas (Fig. 4.4b).  A time lag in response to fragmentation could create this pattern.  Native colonizations over the four decade period have tended to occur in areas that had a higher shape index in 1964 (Fig. 4.3b).  However, at some point since 1964, many patches that had a high shape index have become more ?edgy?, while most patches with a low shape index have remained low (Fig. 4.2c).  This may have obscured the pattern of preferential colonization of patches with a higher shape index by natives that appears for the 1964 landscape data.  The relationship between extirpations and landscape context was much less predictable based on traits (Table 4.5).  Rogers et al. (2009) noticed a similar trend, finding patterns of colonization to be more strongly correlated with landscape context than extirpation. The only trait that seemed consistently useful in predicting extirpations was life form, with shrubs being more likely to have extirpations associated with higher road density, while herbs were more likely to be extirpated in areas with more naturally vegetated area and higher area:perimeter ratio.  This lower predictability of extirpation based on traits could represent an extinction debt, meaning that species destined for extirpation are still hanging on.  Given that over the four decade period colonizations outnumbered extirpations 2:1, the pace of colonizations in response to landscape   90change seems to be faster than extirpations.  The system has clearly not reach an equilibrium in which colonizations and extirpations are balanced.  Given this non-equilibrium condition, I predict that traits not associated with the best colonizers of a disturbed and/or fragmented landscape are likely t hose associated with higher susceptibility to extirpation.  The few significant results for extirpations support this:  for example, shrubs, perennials and low SLA species tended to be extirpated more in areas with high road density in 1964 (Table 4.5).  The nature of the relationship between colonizations, extirpations or presence and landscape context was not predictable based on seed weight, and rarely on dispersal mechanism.  The lack of consistent predictability based on dispersal-related traits could be due to three things.  First, landscape configuration and land use factors may interact to influence dispersal by vertebrates.  For example, a small forest patch may be relatively more inaccessible to vertebrate seed dispersal vectors like deer if it is surrounded by housing developments compared to a patch of equal size surrounded by agricultural fields, and my models did not take into account the potential for this kind of interaction.  Alternatively, the frequency of rare long-distance dispersal events, regardless of seed size or dispersal vector, may be enough to prevent any discernible effect of dispersal traits on colonizations (e.g. Higgins et al., 2003).  Finally, perhaps the degree of fragmentation on this landscape has not yet crossed the critical threshold for dispersal limitation to be a limiting factor (Pardini et al., 2010; With and Crist, 1995).   Given the scarcity of long-term data, many studies have relied on current presence/absence alone in order to determine the role of landscape context on plant community dynamics (e.g. Dupr? and Ehrl?n, 2002; Kolb and Diekmann, 2005; Tremlova and Munzbergova, 2007).  My results   91show that, for this system and on this timescale at least, this strategy is justified.  The predictability of apparent colonizations based on traits generally matched the results for presence/absence data.  However, it is only with temporal data that the relative contribution of colonization and extirpation to current patterns of species and traits on the landscape can be determined.  In this case, the role of landscape context in structuring plant community change over the past four decades is largely driven by colonization events.  Extirpations are less common and much less predictable (Table 4.5).  It may be that extirpations respond to changes in landscape context with an even longer time lag than I observed in the response of colonizations to patch shape.  Alternatively, extirpations may be fundamentally less predictable based on traits due to a larger role for stochastic events.  Only continual monitoring of long-term vegetation plots will enable ecologists to distinguish between these possibilities.  4.6 Conclusion There are a large number of factors that can contribute to the colonization or extirpation of a species, including local edaphic factors, competition, herbivory levels, and stochastic events.  Despite these complicating factors, I have shown that the landscape context within 500m of a plant community can influence long-term plant community dynamics by favouring colonizations of species with particular traits.  Higher density of roads, lower amount of natural vegetation, and a lower area:perimeter ratio promotes colonizations by exotic, disturbance tolerant, annual, herbaceous, shade intolerant species with high SLA.  I therefore expect native, disturbance intolerant, perennial shrubs that are shade tolerant and have lower SLA to be more vulnerable in areas of higher fragmentation.  However, extirpations are less predictable, potentially indicating a time lag of more than four decades before the consequences of landscape changes are realized.     92In highly populated regions, conservation of plant diversity will rely on an understanding of which species are most vulnerable to fragmentation and disturbance caused by human activities.  This is challenging due to the ability of many plants to withstand unsuitable conditions, and the resulting time lag between fragmentation and the response.  The use of historical data is crucial for revealing these time lags, and allows us to measure the relative rates and trait-based determinants of colonization and extirpation events.       93Chapter  5: Phytoliths of southeastern Vancouver Island   5.1 Synopsis Phytolith assemblages in terrestrial soils have the potential to reveal local shifts between open savannah and conifer forests in the past.  On southeastern Vancouver Island, this could help quantify the range of variability of Garry oak savannahs prior to European colonization, and the degree of influence of indigenous people in maintaining open savannahs.  I catalogued the phytolith morphotypes found in 72 of the most common plant species in this system, and extracted phytoliths from two soil cores on either side of the boundary between a conifer forest and an oak savannah.  Grasses are the most prolific producers of phytoliths in savannahs.  In forests, Douglas-fir (Pseudotsuga menziesii) produces large, distinctive asterosclereid phytoliths.  Phytolith assemblages from soil cores showed that phytoliths from grasses dominated, even in soils from the forest, where grasses are rare.  However, the presence of asterosclereid phytoliths reliably distinguished between Douglas-fir forest and Garry oak savannah habitats within 20m of the vegetation boundary.  This finding may be useful for reconstructing past vegetation not only on Vancouver Island, but across a large geographic area in North America where grasslands tend to be invaded by Douglas-fir following fire suppression.  The use of the phytolith record from terrestrial soils at local sites where the history of human occupation is well-known could help establish the degree of influence of indigenous peoples in maintaining open savannahs on southeastern Vancouver Island.     945.2 Introduction In many places across the Americas indigenous peoples had a long and substantial influence on vegetation long before colonization by Europeans (Delcourt and Delcourt, 2004; Denevan, 1992).  However, there are serious debates concerning the degree and extent of this influence in time and space for particular regions (e.g. Vale, 2000; Whitlock and Knox, 2002).  It is very challenging to tease apart the interacting effects of climatic and other non-human agents from cultural impacts on vegetation over long time periods.  Even with strong evidence from ethnography, dendroecology and palynology that the activities of indigenous peoples were an important influence on vegetation structure, mysteries often remain about the details.  This is definitely true for the Garry oak savannahs of Vancouver Island, Canada.  These savannahs represent the northernmost edge of the range of the Garry oak tree (Quercus garryana), which stretches as far south as northern California (Fig. 5.1).  Historical ecological studies have shown that the extent of these savannahs was much greater just prior to European settlement, and that the remnants of this vegetation type now exist mainly in steep, rocky areas not suitable for agriculture (Bjorkman and Vellend, 2010; Lea, 2006; Vellend et al., 2008).  Oak savannahs on deep soil sites are highly susceptible to invasion by the conifer Douglas-fir (Pseudotsuga menziesii; Fuchs, 2001).  Palynological studies have shown that oak savannahs were established on southeastern Vancouver Island by approximately 8,000 years ago, and persisted despite a cooling, moistening trend beginning about 6,000 years ago (Pellatt et al., 2001).   Ethnographic and historical evidence suggests that on deep soil sites in particular, management by indigenous people in the form of frequent, low-intensity prescribed burns maintained the open character of Garry oak savannahs (MacDougall et al., 2004; Turner, 1999).    95However, the spatial and temporal extent of this influence, and the timing of transitions from open savannah to Douglas-fir forest or vice versa prior to European settlement, is not clear.    Figure 5.1: Main map:  the West coast of North America showing Vancouver Island, with the range of Garry oak indicated in grey.  Inset:  Southeastern Vancouver Island and the Gulf Islands, showing Tumbo Island, the location of the soil samples.  Lower right:  photo of the boundary between oak savannah and Douglas-fir forest at the soil sampling location.  Garry oak range map based on map by Erickson (2008).     96Phytolith analysis has the potential to increase our understanding of the history of vegetation change in this system.  Phytoliths are plant microfossils formed when hydrated silicon dioxide (SiO2?nH20) is deposited within and between the cells of a living plant (Pearsall, 2000; Prychid et al., 2003).  The resulting cell-shaped silica casts are released upon decomposition of the plant, and can survive in sediments for thousands of years (e.g. Blinnikov et al., 2002).  Phytoliths may be transported by wind or water, but phytolith assemblages are derived from more local vegetation sources than pollen, which can travel for hundreds of kilometres (Blinnikov et al., 2013; Jacobson and Bradshaw, 1981; Pearsall, 2000).  Therefore, phytoliths have the potential to reveal vegetation changes on a local spatial scale.  Phytoliths are not produced by all plant taxa, but are particularly abundant in certain plant families, and in some cases are specific to the level of genus (Pearsall, 2000).  In North America, phytoliths extracted from buried paleosols and/or modern soil profiles have been used to infer past changes in vegetation in diverse regions including the Great Plains, the Columbia Basin, Arizona, Utah, Minnesota, California, and Manitoba (Bartolome et al., 1986; Blinnikov et al., 2002; Boyd, 2002; Fisher et al., 1995; Kerns et al., 2001; McLaren and Umlauf, 2000; Morris et al., 2010b).  The first steps in utilizing phytoliths to study long-term vegetation change are to establish a reference collection of the phytolith types produced by plants common to the region, and to determine whether the proportions of different phytolith types in surface soils can reliably distinguish between current vegetation types (Fredlund, 2005).  Therefore, this chapter was designed to answer the following questions:   (1) What are the phytolith types and type frequencies in common plants of Douglas-fir forests and Garry oak savannahs of southeastern Vancouver Island?    97(2) Do phytolith assemblages from terrestrial soils reliably differentiate between the two vegetation types?   (3) How sensitive are phytolith assemblages to vegetation boundaries?  5.3 Materials and methods 5.3.1 Reference collection I collected tissue samples from common plant species of the three main non-riparian vegetation types in the region:  Douglas-fir forests, Garry oak savannahs, and Douglas-fir/Arbutus forests, which are dominated by Douglas-fir but include a substantial component of arbutus (Arbutus menziesii) trees in the canopy.  I selected species based on a vegetation survey of 184 20x20m relev? plots surveyed in 2009 (see Chapter 3).  For each of the three vegetation types, I collected samples of all species that had an average cover of at least 1% in all relev?s in which they were present.  I supplemented my collection with the following:  as many grass species as I could collect, including two native bunchgrasses that are now very rare and one very rare exotic grass from the Panicoideae sub-family; one species from the Juncaceae family; and a few recently introduced exotic non-grass species whose phytoliths could potentially be used as stratigraphical markers.  In total I collected samples from 72 species in 28 plant families, including 13 herbs, 3 ferns, 15 shrubs, 10 trees, 1 sedge, and 30 grasses (Table 5.1, Table 5.2).  Samples were collected in the late spring or summer of 2010 or 2011 from mature specimens, and air-dried until they could be processed.    I extracted phytoliths from plant tissue samples using the dry-ashing technique (Pearsall, 2000).  I cut plant samples into small pieces and separated inflorescences, fruit, and vegetative portions   98(stems and leaves).  Shrubs and tree samples were separated into leaves (needles), twigs, fruit, and in some cases bark.  I did not collect below-ground plant parts except for the bulbs of camas (Camassia quamash and C. leichtlinii), which were the most important starch source for indigenous peoples prior to European colonization (Turner and Bell, 1971).  I rinsed herb and grass samples thoroughly in distilled water and placed them in a drying oven at approximately 70?C overnight.   I washed twigs and bark samples in a sonic bath for 30 minutes, and then placed them in a drying oven at approximately 70?C overnight.  I then placed each dried sample in an aluminum foil crucible, weighed it, and then incinerated it in a muffle furnace at 500?C for 30 minutes to several hours, until most material was white ash.  Some plant samples retained blackened areas even after several hours in the furnace.  After burning, I weighed each sample again and transferred it to a glass vial for storage (see Appendix B.1 for pre- and post-burning weights).   To examine the ash residue for phytoliths, I placed a small amount on a microscope slide and added a few drops of Canada balsam.  I examined each slide in its entirety using a Zeiss Imager M2 light microscope at 400x magnification.  I photographed, described and categorized all phytoliths according to the International Code for Phytolith Nomenclature (Madella et al., 2005).  I classified the abundance of each phytolith morphotype in each sample according to the following scale:  ?rare? (1-2 per slide), ?sporadic? (2-20 per slide), ?common? (20-100 per slide), or ?abundant? (>100 per slide).  I examined more than one slide for some species.     99 Table 5.1: Phytolith types and type frequencies found in grass species of southeastern Vancouver Island.  Taxonomy is according to Soreng et al., 2012.  See photos of each phytolith type in Figure 5.2.  Inflorescences = inflor., leaves and stems = veg.  A=abundant, C=common, S=sporadic, R=rare, blank cells indicate that phytolith type was not observed.  Note that all species had hairs/hairbases and stomatal complexes, and most had silicified tracheids - these are not indicated here. Subfamily Tribe Species Origin Plant part rondel (10um) ?saddle rondel? bilobate (?dumbbell?) papillae globular ovate sinuate elongate (?25um) sinuate elongate (>25um) dendriform elongate echinate elongate crenate elongate tuberculate/echinate elongate  polygonal epidermal/hairbases scrobiculate elongate (?30um) scrobiculate elongate (>50um) Danthonioideae Danthonieae Danthonia californica native inflor. C  A    R   A   S C      veg. A  A  A     C   R S Panicoideae Panicae Digitaria sanguinalis exotic inflor.   A A    A  A R   R      veg.   A       A  S R R Pooideae Stipeae Achnatherum lemmonii native inflor. A A      C   S C        veg. A  A     A   R   S  Bromeae Bromus carinatus native inflor. A   A    A    S  S      veg.     R S S       C   Bromus hordeaceus exotic inflor. C   A    A A   A R R      veg.     C S C          100Subfamily Tribe Species Origin Plant part rondel (10um) ?saddle rondel? bilobate (?dumbbell?) papillae globular ovate sinuate elongate (?25um) sinuate elongate (>25um) dendriform elongate echinate elongate crenate elongate tuberculate/echinate elongate  polygonal epidermal/hairbases scrobiculate elongate (?30um) scrobiculate elongate (>50um) Pooideae cont?d Bromeae cont?d Bromus sterilis  exotic inflor. A   A    A A   A        veg. S     S A   C   R R   Bromus vulgaris  native inflor. A   A    A     R S      veg. S     S A   A   S S  Hordeeae Elymus glaucus  native inflor. A   A    A C  S C  S      veg. A    A    C C    R   Hordeum murinum exotic inflor. A   A    A  A   R R      veg. S    R S S   A      Meliceae Melica subulata  native inflor.    A  S   A   C  R      veg.      C C   A    S  Poeae (CG1)a Agrostis exarata native inflor. A   C    A A A  C        veg. A     C A   C   R S   Agrostis pallens  native inflor. A   A    A    C  R      veg. A     A A   A       101Subfamily Tribe Species Origin Plant part rondel (10um) ?saddle rondel? bilobate (?dumbbell?) papillae globular ovate sinuate elongate (?25um) sinuate elongate (>25um) dendriform elongate echinate elongate crenate elongate tuberculate/echinate elongate  polygonal epidermal/hairbases scrobiculate elongate (?30um) scrobiculate elongate (>50um) Pooideae cont?d Pooeae (CG1) cont?d Alopecurus arundinaceus exotic inflor. A   R    A  A  C        veg.        A  A    A   Anthoxanthum odoratum exotic inflor. R   A      A  A S A      veg.     S     C   R S   Arrhenatherum elatius exotic inflor. A   A    C  C  C        veg. S     S C       A   Cynosurus echinatus exotic inflor. S   A  R  A   R C R       veg. A    A A A   A    C   Phalaris arundinacea exotic inflor. A   S      R    A      veg. A     C C   A    A  Poeae (CG2)a Aira caryophyllea exotic inflor. A   A      A  S S R      veg. A     S C   A   R S   Aira praecox  exotic inflor. A   A      A          veg. A    C S C  R A   R C   102Subfamily Tribe Species Origin Plant part rondel (10um) ?saddle rondel? bilobate (?dumbbell?) papillae globular ovate sinuate elongate (?25um) sinuate elongate (>25um) dendriform elongate echinate elongate crenate elongate tuberculate/echinate elongate  polygonal epidermal/hairbases scrobiculate elongate (?30um) scrobiculate elongate (>50um) Pooideae (cont?d) Pooeae (CG2) cont?d Dactylis glomerata exotic inflor. A   A    A C     S      veg.     S A A      R C   Festuca occidentalis native inflor. A   A     A A R  R C      veg. A     C C   A S   S   Festuca roemeri  native inflor. A   A    A A  R S S C      veg. A     A A   A    C   Festuca rubra  exotic inflor. A   A    A A  R S R       veg. A      R   A A   C   Festuca subulata native inflor. C   A    A   C  S C      veg. A     S    A A  C A   Holcus lanatus  exotic inflor. A   C      A  R        veg. A     C C   A    C   Lolium perenne  exotic inflor. A   A    A A A S   C      veg. A     A A   A C   C   103Subfamily Tribe Species Origin Plant part rondel (10um) ?saddle rondel? bilobate (?dumbbell?) papillae globular ovate sinuate elongate (?25um) sinuate elongate (>25um) dendriform elongate echinate elongate crenate elongate tuberculate/echinate elongate  polygonal epidermal/hairbases scrobiculate elongate (?30um) scrobiculate elongate (>50um) Pooideae (cont?d) Pooeae (CG2) cont?d Poa bulbosa  exotic inflor. A     A A   A R   S      veg. A     A A   A S   R   Poa pratensis  native inflor. A   A    A A  S S R S      veg. A     A A A  A R   C   Vulpia bromoides exotic inflor. A   A  S  A A A S A  S      veg. A     S S   A R   S   Vulpia myuros  exotic inflor. A   A   S A  A S C  R      veg. A     S S   A C   S a CG1 and CG2 refer to ?chloroplast group 1? and ?chloroplast group 2? of the Poeae tribe.  See Soreng et al., 2012.   104Table 5.2: Phytolith types and type frequencies in non-grass species of southeastern Vancouver Island.  Silica-based phytoliths are indicated in bold letters.  A=abundant, C=common, S=sporadic, R=rare. Family Species Origin Life form Phytoliths Aceraceae Acer macrophyllum native tree leaves:  blocky CaOx (A), stomatal complexes (A) seeds:  rhombohedral CaOx (A) twigs/petioles:  blocky CaOx (C), spiky crystals (S) Apiaceae Sanicula crassicaulis native herb stems/leaves:  spherical CaOx (A) Aquifoliaceae Ilex aquifolium exotic shrub stems and leaves:  spherical CaOx (A) Araliaceae Hedera helix exotic shrub branches/leaves/petioles:  spherical CaOx (A) Asteraceae Hypochaeris radicata exotic herb flowers:  none observed leaves:  hairs (S), tracheids (R)  stems:  none observed Berberidaceae Achlys triphylla native herb flowers:  puzzle piece epidermal (S) stems/leaves:  none observed  Mahonia aquifolium native shrub leaves:  none observed stems:   none observed  Mahonia nervosa native shrub berries:  none observed leaves:  rhombohedral CaOx (A) twigs:  none observed Betulaceae Alnus rubra native tree leaves:  spherical CaOx (A) twigs:  spherical CaOx (A) Caprifoliaceae Linnaea borealis native herb stems/leaves/flowers:  spherical CaOx (A)  Lonicera hispidula native shrub leaves:  blocky mesophyll/epidermal cells (A, poorly silicified), spherical CaOx (A), puzzle piece epidermal (C, poorly silicified) stems:  spherical CaOx (A), scrobiculate elongate (S)  Symphoricarpos albus native shrub berries:  spherical CaOx (C), rhombohedral CaOx (C) leaves:  spherical CaOx (A), tracheids (S) stems:  spherical CaOx (A) Celastraceae Paxistima myrsinites native shrub leaves:  spherical CaOx (A), puzzle piece epidermal (S) twigs:  spherical CaOx (C) Cornaceae Cornus nuttallii native tree flowers/fruit:  spherical CaOx (A), fusiform echinate 100-200um (A) leaves:  spherical CaOx (A), scrobiculate elongate (S) twigs:  rhombohedral CaOx (S), spherical CaOx (A), fusiform echinate 100-200um (A) Crassulaceae Sedum spathulifolium native herb stems/leaves:  spherical CaOx (A) Cupressaceae Thuja plicata native tree bark:  spiky crystals (A) leaves:  none observed twigs:  none observed Cyperaceae Carex inops native sedge inflorescences:  conical epidermal (C), echinate elongate (A), hairs (C), scrobiculate elongate (C), tracheids (C) stems/leaves:  conical epidermal (A), crenate elongate (A), tuberculate/echinate elongate (S), hairs (S), scrobiculate elongate (R), stomatal complexes (C) Denstaedtiaceae Pteridium aquilinium native fern leaves:  puzzle piece epidermal (A) stems:  blocky CaOx (A), spherical CaOx (S), scrobiculate elongate (S), tracheids (S)   105Family Species Origin Life form Phytoliths Dryopteridaceae Polytichum munitum native fern leaves:  puzzle piece epidermal (S, poorly silicified) stems:  tracheids (S, poorly silicified) Ericaceae Arbutus menziesii native tree leaves:  blocky CaOx (A), tracheids (S)  Gaultheria shallon native shrub leaves:  spherical CaOx (A) twigs:  rhombohedral CaOx (C), spherical CaOx (C), scrobiculate elongate (A, poorly silicified) Fabaceae Cytisus scoparius exotic shrub leaves:  none observed twigs:  none observed  Trifolium dubium exotic herb leaves/stems:  blocky CaOx (A) Fagaceae Quercus garryana native tree acorns:  spherical CaOx (A) leaves:  blocky CaOx (A), spherical CaOx (A) twigs:  blocky CaOx (A), spherical CaOx (A) Juncaceae Luzula fastigiata native herb stems/leaves:  hairs (S), spiky papillae (C), scrobiculate elongate (C), sinuate elongate (C) Liliaceae Camassia leightlinii native herb bulbs/roots:  raphide CaOx (S), tracheids (R)  Camassia quamash native herb bulbs/roots:  raphide CaOx (A)  Camassia spp. native herb flowers/seeds:  raphide CaOx (R) stems/leaves:  raphide CaOx (A) Orchidaceae Epipactis helleborine exotic herb flowers:  raphide CaOx (A) stems/leaves:  blocky CaOx (C), raphide CaOx (C), scrobiculate elongate (A) Pinaceae Abies grandis native tree needles:  blocky mesophyll/epidermal cells (C, poorly silicified), blocky CaOx (C) twigs:  blocky CaOx (A), spiky crystals (R), tracheids (S)  Pinus contorta native tree needles:  tracheary elements with bordered pits (S), pilate elongate (A), pilate oblong (A, sometimes seem to be melded with the pilate elongate cells), scrobiculate elongate (A) twigs:  scrobiculate elongate (A)  Pseudotsuga menziesii native tree needles:  asterosclereids (R), blocky mesophyll/epidermal (S, poorly silicified) twigs:  blocky CaOx, 10um (A)  Tsuga heterophylla native tree cones:  none observed needles:  none observed twigs:  scrobiculate elongate (A) Plantaginaceae Plantago lanceolata exotic herb flowers:  none observed stems/leaves:  scrobiculate elongate (A, poorly silicified) Polypodiaceae Polypodium glycyrrhiza native fern stems/leaves:  rhombohedral CaOx (S), papillate elongate (S), puzzle piece epidermal (S, poorly silicified), scrobiculate blocky mesophyll/epidermal (C), tracheids (R) Portulacaceae Claytonia perfoliata native herb flowers/buds:  none observed stems/leaves:  none observed Primulaceae Trientalis borealis ssp. latifolia native herb stems/leaves:  spherical CaOx (A), fusiform echinate 100-200um (S) Rosaceae Holodiscus discolor native shrub buds:  spherical CaOx (A) leaves:  spherical CaOx (A), tracheids (S) twigs:  spherical CaOx (A)   106Family Species Origin Life form Phytoliths  Oemleria cerasiformis native shrub fruits:  none observed leaves:  spherical CaOx (A), puzzle piece epidermal (S), tracheids (S, poorly silicified) twigs:  spherical CaOx (A) Rosaceae Prunus laurocerasus exotic shrub leaves:  spherical CaOx (A), blocky mesophyll/epidermal cells (A, poorly silicified), tracheids (S) stems:  spherical CaOx (A), blocky mesophyll/epidermal cells (S, poorly silicified), rhombohedral CaOx (C)  Rosa gymnocarpa native shrub leaves:  blocky CaOx (A), spherical CaOx (A), puzzle piece epidermal (C) stems:  blocky CaOx (A), spherical CaOx (C)  Rubus armeniacus exotic shrub flowers/fruit:  spherical CaOx (A) leaves:  spherical CaOx (A) stems:  spherical CaOx (A)  Rubus ursinus native shrub leaves:  rhombohedral CaOx (A) stems:  spherical CaOx (C)  5.3.2 Soil sampling across an ecotone Soil collection I collected two 5 cm diameter soil cores from Tumbo Island, which is part of the Gulf Islands National Park Reserve administered by Parks Canada (Fig. 5.1).  Tumbo Island is a small island, approximately 105ha in area, which includes areas of Garry oak savannah, but is largely dominated by Douglas-fir forests. I chose an area on the southeastern side of the island that consists of Garry oak savannah nearest the shoreline, but switches abruptly to Douglas-fir/Arbutus forest moving inland (Fig. 5.1).  The oak savannah vegetation consists of sparse Garry oak trees with an understory of mixed herbaceous species and exotic grasses.  The most abundant grass species is the introduced exotic sweet vernalgrass (Anthoxanthum odoratum).  The Douglas-fir/Arbutus forest consists of a canopy dominated by Douglas-fir, with some arbutus and Western red cedar (Thuja plicata), and an understory dominated by the shrub salal (Gaultheria shallon).     107I took one core from the savannah vegetation 20m from the ecotone boundary, and one core from inside the Douglas-fir forest, also 20m from the boundary.  The soil is a shallow, well-drained, sandy loam developed over sandstone bedrock (Kenney et al., 1988).  I used an AMS multi-stage sediment sampler with a slide hammer to extract the cores.  The savannah core was approximately 35cm long and the Douglas-fir core was approximately 49cm long.    My primary goals were to assess the assemblage of phytoliths present in a small sample of modern terrestrial soils in comparison to the phytoliths observed in my plant samples, and to determine whether surface soil phytolith assemblages can be used to distinguish between present-day Douglas-fir forest and savannah vegetation.  This was a necessary first step prior to any future wider study of soils in the region.  However, given the use of phytolith assemblages proceeding downwards in soil cores for paleoenvironmental reconstruction in other studies, and the lack of any such studies in the region, I was also interested in observing patterns in phytolith assemblages with soil depth.  Although phytoliths have been shown to be somewhat mobile in soil (e.g. Fishkis et al., 2010), there is also some evidence that the average age of the phytoliths in a soil layer increases with depth (Alexandre et al., 1999), validating paleoenvironmental interpretations.    Phytolith extraction I divided each soil core into 2cm increments and processed approximately every other increment.  I used a wet oxidation and heavy liquid flotation procedure modified from Pearsall (2000) to extract phytoliths (see Appendix B.2 for the detailed protocol).  First I air-dried the soil sample overnight, and then filtered it through a 500 micron sieve.  I weighed out approximately 5g of the   108filtered soil to process for phytoliths.  I removed carbonates by heating in 10% hydrochloric acid for one hour.  Following rinsing to remove the acid, I removed organic material with a five minute bleach treatment followed by heating in 30% hydrogen peroxide until no further reaction was visible.  This required several hours for highly organic samples.  I then dispersed clay particles by shaking for 24 hours in a 0.1% solution of ethylenediaminetetraacetic acid (EDTA) disodium salt dihydrate.    I removed clay particles by sedimentation in a 10cm column of water for 9.3 hours followed by removal of the supernatant.  I repeated the sedimentation process at least 4 times.  Finally, I separated phytoliths using heavy liquid flotation with sodium polytungstate at 2.3g/cm3 density.  I decanted the supernatant (with suspended phytoliths), and then sank the phytoliths by adding distilled water and centrifuging. I carried out this process six times to ensure complete flotation of phytoliths.  I rinsed the final extract twice, and then dried it at approximately 60? C in a drying oven until all water had evaporated (2-3 days).  I weighed the final extract and stored it in glass vials.  Phytolith counts I used an analytical balance to add between 0.5mg and 1.2mg of phytolith extract to a microscope slide (measured to the nearest 0.1mg).  I then added 3-4 drops of Canada balsam, mixed the extract in thoroughly using a clean dissecting needle, and placed a cover slip gently on top.  I scanned the entire slide at 200x magnification to count the large asterosclereid phytoliths produced by Douglas-fir.  I then counted other phytolith morphotypes at 400x magnification across a transect of 16-18 consecutive microscope fields from the centre of the cover slip to the   109edge.  This resulted in a total count of between 205 and 1,516 phytoliths, depending on the sample.  I estimated the number of each phytolith morphotype (other than asterosclereids) per slide by first calculating the weighted average number of phytoliths per field by weighting each field by its distance from the cover slip centre.  For example, the first field counted is one field distant from the cover slip centre, and there are eight fields equidistant from the centre, so I multiplied the counts for this field by eight.  The counts for the next field were multiplied by 16, and so on.  I then divided by the total number of fields to find the average number of each morphotype per field.  This gave more weight to fields more distant from the centre. I used this calculation because no matter how well I stirred the extract into the Canada balsam, phytoliths were usually more concentrated near the centre of the cover slip, and I did not want to overestimate the actual number of phytoliths per slide.  I then multiplied this weighted average by the number of fields that would be needed to cover the entire area of the cover slip.  I counted at least one slide for each 2cm increment, and graphed the total count per gram of dried soil and the percentage of each morphotype against depth in the core.  I also examined the total phytolith count per gram of the final extract, but as this was qualitatively very similar to the graphed amounts per gram of soil I chose to display the latter only.  I predicted that asterosclereids would be significantly more abundant in the Douglas-fir forest core and morphotypes produced by grasses would be more abundant in samples from the oak savannah soil core.    1105.4 Results 5.4.1 Phytoliths found in grasses Almost all of the grasses in my sample are from the Pooideae subfamily (members of Twiss et al.?s Festucoid class), and as expected rondels are abundant in most of these species (Table 5.1; Fig. 5.2a; Twiss et al., 1969).  I did not differentiate between different rondel forms (e.g. ?horned? versus ?keeled? rondels).  Bilobate short cells (also called ?dumbbells?; Fig. 5.2e) were restricted to the two native bunchgrasses, Danthonia californica (of the Danthoniodieae subfamily) and Achnatherum lemmonnii (of the Stipeae tribe), and to the Panicoid exotic Digitaria sanguinalis, which is highly rare in this region.  The inflorescences of A. lemmonnii also contained abundant cylindrical short cells, about the same diameter as rondels, but with concave ends and edges that appear ornamented with small round bumps (Fig. 5.2b,c).  I have nicknamed these ?saddle rondels?.  A few of my grass species also had globular ovate phytoliths in varying abundance in their vegetative tissue (Fig. 5.2d; Table 5.1).  Unlike Morris et al. (2010b), I did not find a consistent difference in rondel production between exotic and native grasses.  Most of the grass species had abundant papillae phytoliths in their inflorescence tissue (Fig. 5.2f).  In terms of elongate cells (long cells), morphology ranged from smooth scrobiculate to dendriform (Fig. 5.2g-n).  In all grasses more than one of these elongate morphotypes were found, and the phytoliths in one plant species often grade between different ornamentation types.  Overall, dendriform elongates were more often found in inflorescence tissue than in vegetative tissue, while sinuate elongates were more common in vegetative tissue (Table 5.1).  I separated sinuate elongate phytoliths into two lengths because they seemed to consistently fall within these   111two classes (Fig. 5.2g,h).  Similarly, scrobiculate elongate cells were divided into two size classes (Fig. 5.2l,m).  In general, elongate cells ranged in length from less than 25 microns to over 100 microns.  In addition to short cells and elongate cells, I often observed silicified tracheids (Fig. 5.2r), hairs and hairbases (Fig. 5.2s), and stomatal complexes (Fig. 5.2t).  Many species also had a class of phytolith I have called ?polygonal epidermal/hairbases?, as they appear to be epidermal cells, but in some cases are subtended by hairs (Fig. 5.2o).  I did not observe cuneiform bulliform cells (also known as ?fan shaped? phytoliths) in any of the ashed samples, although these phytolith types are frequently noted in grass species (Pearsall, 2000).   It could be that bulliform cells were poorly silicified in my samples due to dry conditions (Parry and Smithson, 1964).  In their study of grasses of the Great Basin, Morris et al. (2009) found bulliform cells in only two of 33 grass species they examined.  Two phytolith types were unique each to only one grass species in my sample. I found abundant phytoliths in Agrostis pallens inflorescences that appear to be elongate cells with regular thickened bands along their length (Fig. 5.2p).  These resemble the abaxial epidermis multicells described by University College London researchers in Agrostis gigantea husks (Fuller, 2007).   The stems and leaves of Digitaria sanguinalis had a rare occurrence of interesting Y-shaped phytoliths located between bilobate cells (Fig. 5.2q).  These may be the ?cross-shaped and nodular? cells referred to by Metcalfe (1960, p. 162).        112 Figure 5.2: Phytolith morphotypes found in grass species:  (a) rondels, from vegetative tissue of Vulpia bromoides, (b) ?saddle rondel? (side), from inflorescences of Achnatherum lemmonii, (c) ?saddle rondel? (top), from inflorescences of A. lemmonii, (d) globular ovate, from vegetative tissue of Danthonia californica, (e) bilobate, from vegetative tissue of D. californica, (f) papillae, from inflorescences of Bromus carinatus, (g) sinuate elongate (? 25um), from vegetative tissue of Poa pratensis, (h) sinuate elongate (>25um), from vegetative tissue of Arrhenatherum elatius, (i) dendriform elongate, from inflorescences of Bromus hordeaceus, (j) echinate/papillate elongate, from vegetative tissue of A. lemmonii, (k) crenate elongate, from vegetative tissue of Festuca occidentalis, (l) scrobiculate elongate (?50um), from vegetative tissue of Bromus vulgaris, (m) scrobiculate elongate (?30um), from inflorescences of Festuca subulata, (n) echinate/tuberculate/papillate elongate, from vegetative tissue of F. occidentalis, (o) polygonal epidermal/hairbases, from inflorescences of B. carinatus, (p) banded elongate cells, from inflorescences of Agrostis pallens, (q) nodular phytoliths, from inflorescences of Digitaria sanguinalis, (r) tracheid, from vegetative tissue of Lolium perenne, (s) hair and hairbase, from inflorescences of F. subulata, (t) stomatal complex, from inflorescences of Festuca rubra.    1135.4.2 Phytoliths found in non-grasses Most of the non-grass species I observed contained calcium oxalate (CaOx) crystals (Table 5.2, Fig. 5.3p-u).  These crystals are widespread in higher plants and can be diagnostic of certain taxonomic groups (Pearsall, 2000; Franceschi and Nakata, 2005).  However, they are prone to dissolution in acidic environments, and therefore they are difficult to recover from soil with extraction procedures that use strong acids to remove carbonates (Pearsall, 2000).  One study (McLaran and Coder, 2003) failed to isolate them from soil even when using an extraction protocol that did not use acid.  Given that the soils in my study region tend to be quite acidic (Kenney et al., 1988), it is unlikely that CaOx crystals can be used in paleoecological reconstructions from soil in this region.  My samples contained four distinct shapes of CaOx crystals, the most abundant of which are spherical, also known as druses (Fig. 5.3p,q; Franceschi and Nakata, 2005).  I observed spherical CaOx crystals in 12 of my 15 shrub species, three trees, four herbs, and one fern.  I found rhombohedral CaOx crystals of varying sizes in eight species (Fig. 5.3r,s) and blocky CaOx crystals in nine species (Fig. 5.3t).  I observed raphide CaOx crystals (Fig. 5.3u) only in species of the Liliaceae and Orchidaceae families (Table 5.2).  Silica-based phytoliths were relatively rare in my non-grass samples, but there were a few species with distinctive phytoliths.  As expected, I found asterosclereid phytoliths in the needles of Douglas-fir (Pseudotsuga menziesii), a conifer in the Pinaceae family (Fig. 5.3a).  Asterosclereid phytoliths are formed by the deposition of silica within asterosclereids (also called stellate sclereids), which are large, non-living lignified plant cells thought to provide structural   114support (Al-Talib and Torrey, 1961; Apple et al., 2002).  They appear as branched bodies, with some of the branches often broken off in soil (Fig. 5.3a, 5.4).  Rovner described them as ?antler-shaped? (Rovner, 1983).  These large, distinctive phytoliths were first noted in British Columbia in the early 1960s, when soil scientists came across the distinctive bodies in the silt fraction of a Vancouver Island soil (Brydon et al., 1963).  Since then, Norgren (1973), Klein and Geis (1978) and Blinnikov (2005) observed them in plant samples from Oregon, New York, and Washington states, respectively.  Due to the large size of these phytoliths (up to 200 microns) I found them quite rare in my ashed samples, such that I examined at least three slides before finding one.  The relative abundance of asterosclereids in Douglas-fir needles can vary depending on the location of the needles along the shoot axis (Al-Talib and Torrey, 1961), the age of the tree (Apple et al., 2002), and the availability of silica in the soil upon which the tree is growing (Norgren, 1973).  I confirmed the presence of the characteristic conical epidermal cells of the Cyperaceae family in Carex inops (Fig. 5.3b,c).  This species also has crenate, scrobiculate, and echinate elongate cells, as were found in many grasses (see Fig. 5.2j-m).   Luzula fastigiata, from the Juncaceae family, also has elongate cells resembling those of the Poaceae, including sinuate and scrobiculate elongate phytoliths (Table 5.2).  This species also has phytoliths that resemble spiky or papillate papillae (Fig. 5.3o).  Both Cornus nuttalli (a tree), and Trientalis borealis ssp. latifolia (an herb) have large fusiform echinate phytoliths, ranging in size from 100-200 microns (Fig. 5.3d).  Pinus contorta (Pinaceae) contains pilate oblong phytoliths identical in appearance to the phytoliths reported by Blinnikov (2005) and Norgren (1973) for Pinus ponderosa (Fig. 5.3f).   Blinnikov did not report this   115morphotype in his examination of P. contorta.  Also found in P. contorta are pilate elongate phytoliths (Fig. 5.3g) which sometimes seem fused with the oblong types (Fig. 5.3h).  The needles also contain sporadic tracheary elements with bordered pits (Fig. 5.3i), which Bozarth (1993) considers characteristic of the Pinaceae family.  Several of the non-grass species contained scrobiculate elongate phytoliths similar to those seen in grasses (see Fig. 5.2m).  I noted sheets of brick-like blocky mesophyll or epidermal cells in five species, although these were often poorly silicified (Fig. 5.3l,m,n).  These cells appeared pitted (scrobiculate) in the case of the fern Polypodium glycyrrhiza (Fig 5.3n).  The silicified epidermal cells of eight of my species resembled puzzle-pieces (Fig. 5.3j).  I also noted silicified tracheids and stomatal complexes (Fig. 5.3k) in a few species (see Table 5.2).  Lastly, in three of my tree species I noted very fine, spiny crystals in the twigs or bark (Fig. 5.3e).   I am not certain if these are silica-based bodies.     116    117Figure 5.3: (Previous page) Phytolith morphotypes found in non-grass species:  (a) asterosclereid, from needles of Pseudotsuga menziesii, (b) conical epidermal (top), from vegetative tissue of Carex inops, (c) conical epidermal (side), from vegetative tissue of C. inops, (d) fusiform echinate, from leaves of Cornus nuttallii, (e) spiky crystal, from bark of Thuja plicata, (f) pilate oblong, from needles of Pinus contorta, (g) pilate elongate, from needles of P. contorta, (h) pilate elongate/oblong, from needles of P. contorta, (i) tracheary element with bordered pits, from needles of P. contorta, (j) puzzle-piece epidermal, from leaves of Pteridium aquilinum, (k) stomatal complexes, from leaves of Prunus laurocerasus, (l) mesophyll, from leaves of P. laurocerasus, (m) mesophyll, from leaves of Lonicera hispidula, (n) mesophyll, from leaves of Polypodium glycyrrhiza, (o) ?spiky papillae?, from tissue of Luzula fatigata, (p) calcium oxalate spherical (large), from leaves of L. hispidula, (q) calcium oxalate spherical (small), from leaves of Alnus rubra, (r) calcium oxalate rhombohedral (large), from stems of P. laurocerasus, (s) calcium oxalate rhombohedral (small), from leaves of Mahonia nervosa, (t) calcium oxalate blocky, from twigs of P. menziesii, (u) calcium oxalate raphide, from bulbs/roots of Camassia quamash.   5.4.3 Soil core analyses The weight of the final extract ranged from 1.99% to 7.22% of the original dry soil sample (uncorrected for non-phytolith particles remaining in the extract).  Based on my examination of phytoliths in plant tissue, I counted five categories of phytoliths:  asterosclereids, elongate cells, rondels, bilobates, and ?other? phytoliths.  The final category included hairs/hairbases, the conical epidermal cells from Carex, tracheids, and unknown potential phytoliths.  This category was dominated by hairs/hairbases, with 59% of 192 total ?other? phytoliths being classified as hairs/hairbases.  8.3% were tracheids and 1% (2 out of 192) were Carex epidermal cells.  I did not observe any papillae, stomatal complex, ?saddle rondels?, puzzle-piece epidermal or Pinus contorta-type morphotypes in the Tumbo Island soil samples.  I also did not see any calcium oxalate crystals in the phytolith extract.   As predicted, asterosclereids were consistently more abundant in the Douglas-fir forest core than in the savannah core, although they declined sharply with depth, and were not found at the   118deepest increment sampled, 46 cm (Fig. 5.4, 5.5).  Excluding this depth, Douglas-fir forest samples ranged from 382 to 5,101 asterosclereids per gram of soil, while oak savannah samples ranged from 0 to 128 asterosclereids per gram of soil.  The maximum number of asterosclereids counted per slide for oak savannah samples was 5, while Douglas-fir forest samples ranged from 8 to 106 asterosclereids per slide (not counting the 46 cm depth; data not shown).  Asterosclereids represented a maximum of 0.19% of total phytoliths per slide (Fig. 5.6).  Soil samples were dominated by elongate phytolith types, which made up 60-83% of the estimated total number of phytoliths per slide (Fig. 5.6).  They did not follow my prediction of being more abundant in the grass-dominated oak savannah core.  At similar depths, elongates were usually more abundant in the forest core (Fig. 5.5).  However, they did tend to represent a higher proportion of the total number of phytoliths per slide in the oak savannah core (Fig. 5.6).  The second most abundant of the morphotypes were the rondels, which represented between 18% and 36% of total phytoliths per slide (Fig. 5.6).  Rondels also tended to be more abundant in the forest core (Fig. 5.5), and to represent a higher percentage of total phytoliths than in the savannah core (Fig. 5.6). Figure 5.4: Asterosclereid phytolith extracted from the soil core located within the Douglas-fir-dominated forest on Tumbo Island.        119 Bilobate phytoliths were quite rare in my soil samples, being between 0 and 3% of the estimated phytoliths per slide (Fig. 5.6).  These grass-produced phytoliths were also unexpectedly more abundant and represented higher proportions of total phytoliths per slide in the forest core samples than in the savannah core samples (Fig. 5.5, 5.6).  The proportion of phytoliths represented by the bilobate morphotype seemed to increase with depth in the soil core, in particular in the savannah core, however the proportion remained relatively low in all cases (Fig. 5.6).  Figure 5.5: Estimated total number (in thousands of phytoliths) of (a) asterosclereids, (b) elongate phytoliths, (c) rondels and (d) bilobates per gram of dry soil plotted against depth in the soil core.  Black triangles represent the Douglas-fir forest core; open circles represent the Garry oak savannah core.  Note that the scale of the y-axis differs for each phytolith morphotype.   120  Figure 5.6: Percentage of total estimated number of phytoliths per slide represented by (a) asterosclereids, (b) elongate phytoliths, (c) rondels and (d) bilobates plotted against depth in the soil core.  Black triangles are the Douglas-fir forest core; open circles represent the Garry oak savannah core.  Note that the scale of the y-axis differs for each phytolith morphotype.  5.5 Discussion 5.5.1 Documenting changes in savannah understorey composition Garry oak savannahs in British Columbia today are dominated by introduced exotic grasses in the Pooideae subfamily (Fuchs, 2001; Maslovat, 2002).  The understorey composition of these savannahs in the past is not known.  Some researchers have suggested that the two native bunchgrasses, Achnatherum lemmonii and Danthonia californica, were much more abundant   121prior to the introduction of agronomic grass species (Maslovat, 2002), while others suggest that there was a mosaic of mixed native grasses and forbs, with some areas dominated by camas (MacDougall et al., 2004).  Phytolith research has the potential to corroborate the former hypothesis, if the bilobate phytoliths found only in the two native bunchgrasses are more abundant in the lower (older) layers of soils that have a well-established stratigraphy.  Indeed, researchers in California have used the increased presence of bilobates in subsurface soils as evidence of the former dominance of native bunchgrasses on certain sites (Bartolome et al., 1986).  On average, the Tumbo Island soil samples seemed to show an increase in bilobate morphotypes with depth (Fig. 5.5, 5.6), however I did not attempt any form of soil layer dating to confirm that deeper soil layers are in fact older.  I did not find any silica-based phytoliths in any part of the camas plant, and therefore phytolith analysis cannot shed light on its abundance in the past.     Nearly all the grass species I examined from the region produce abundant rondels and elongate phytoliths.  However, in the test soil cores these morphotypes were as abundant or more abundant in the forest core, where grasses were very rare.  This finding is similar to a study in southern Arizona, which found phytoliths characteristic of warm-season grasses to be abundant even in heavily forested areas where these grasses were absent (Kerns et al., 2001).  Bozarth (1993) also found soils of boreal forest and aspen stands to be dominated by grass phytoliths even though grasses were rare.  This could be caused by two broad processes.  First, we must consider that phytolith assemblages are formed over decades to centuries, a process termed ?inheritance? (Fredlund and Tieszen, 1994).  This incorporation of phytoliths into a soil layer may be affected by the rate of soil accumulation and/or erosion, shifts in the plant community   122over time, as well as the degree of preservation of phytoliths due to soil chemistry or texture (Fredlund and Teiszen, 1994; Morris et al., 2010b).  A difference in one or more of these factors between the two core locations could have caused differences in the concentration of grass phytoliths in the soil.  For example, the higher abundance of grass phytoliths in the forest core could be a result of greater preservation of phytoliths in the Douglas-fir forest soil, or alternatively, the Douglas-fir forest may be a relatively recent formation on a soil that was previously dominated by grass species.  Secondly, conditions in the open savannah could permit the relatively small grass phytoliths to be translocated into the nearby forest from the savannah before they are deposited in the soil.  The original ?decay-in-place? model of phytolith deposition has been found inaccurate in grasslands where wind, fire and herbivory can transport grass phytoliths moderate to long distances (Piperno, 1988; Fredlund and Tieszen, 1994; Kerns et al., 2001; Fredlund, 2005; Madella and Lancelotti, 2012; Blinnikov et al., 2013).  The location of the Tumbo Island savannah core is in shallow soil exposed to the nearby coastline and winds.   Regardless of the cause, grass phytoliths in surface soils do not appear to be accurate indicators of the shift between Douglas-fir forest and oak savannah at this scale.  More work is needed to determine if grass-produced morphotypes are rarer in forest interiors, where a source of grass phytoliths is more distant.  In Chapter 6, I examine soil cores from a wider range of deep-soil savannah remnants and Douglas-fir forests that are farther away from coastal influence.  5.5.2 Asterosclereids as indicators of the presence of Douglas-fir The large asterosclereid phytoliths produced in Douglas-fir needles were significantly more abundant in the forest core samples, and were found very rarely in the savannah soil core, just   12320m from the forest boundary.  This along with their very distinctive appearance and their specificity to Douglas-fir makes them an excellent indicator of the local presence of Douglas-fir in the past, if a reliable soil chronology can be established.  Care must be taken to establish that deeper soil layers do in fact contain older phytolith assemblages.  In the Tumbo Island forest core, for example, a spike in asterosclereids, elongate and rondel phytoliths occurs at 31cm depth in the core (Fig. 5.5), and this could result from a buried surface horizon.  Examination of surface soil samples from Douglas-fir forests of varying ages and tree densities is needed to determine whether the abundance of asterosclereids can be used to indicate the density of forest in an area.  However, the presence of asterosclereids alone can be used to document the encroachment of Douglas-fir into savannah areas over time in specific locations.  This finding is applicable not only to the Garry oak savannahs of British Columbia, but across a large geographic area in which grasslands are infilling with Douglas-fir, including the entire range of Garry oak (Fig. 5.1), as well as other regions of California (e.g. Cocking et al., 2012), Montana (e.g. Heyerdahl et al., 2006), and New Mexico (e.g. Coop and Givnish, 2007).    5.6 Conclusion The most common plants of the Douglas-fir forests and Garry oak savannahs of southeastern Vancouver Island produce a few distinctive phytolith morphotypes.  Most importantly, the grasses which dominate savannahs produce abundant elongate and rondel phytoliths, and Douglas-fir produces a large, distinctive asterosclereid phytolith.  The relative abundance of these phytolith types in terrestrial soil layers has the potential to help document local shifts between open savannahs and closed-canopy Douglas-fir forests in the past.  However, test cores   124revealed that soil formed under Douglas-fir forests near grassy savannahs can contain relatively high amounts of grass-produced phytolith morphotypes even though grass is sparse in the forest understorey.  In Chapter 6, I explore whether this holds for deeper soils and in forested areas that are more distant from sources of grass phytoliths.  Given the tendency of Douglas-fir to invade deep soil savannahs in the absence of fire, the use of the phytolith record in terrestrial soils at sites where archaeological research has established the history of human occupation and palynological research has documented regional vegetation history could provide important insights into the local variability of this ecosystem before European settlement, and the degree of influence of indigenous peoples on local plant communities in the past.   125Chapter  6:  Combining phytolith analysis with historical ecology to better understand the long-term dynamics of an endangered ecosystem  6.1 Synopsis The historical range of variation in the vegetation of local sites on southeastern Vancouver Island prior to European settlement is unknown.  In this chapter, I use an extensive sample of surface soils to establish a phytolith index that distinguishes current vegetation types.  I then measure changes in this index with depth in soil profiles from sites where historical ecological research has quantified shifts in vegetation since the first land surveys in 1859.  I also examine the proportions of burnt phytoliths and bilobate phytoliths with depth to investigate whether phytolith coloration is a good index of fire frequency in this region, and whether native bunchgrasses were formerly a dominant understorey component before the introduction of exotic agronomic grasses.   The log ratio of asterosclereid to rondel phytoliths in soil surface samples accurately distinguishes between current vegetation types.  Shifts in this ratio with depth were sensitive to known historical changes in most of the sampled cores.  The soil core analyses reveal that some sites have likely supported open savannah vegetation for at least two thousand years, while others were formerly open but have been filled in by Douglas-fir.  However, this infilling appears to have started at different times for different sites.  Only one site showed evidence for an increased component of native bunchgrasses in the past, and burnt phytoliths were found not to be a good indication of past fire history.  I suggest a phytolith-based metric to determine the relative timing of Douglas-fir infilling at different sites across the landscape.     1266.2 Introduction The way an ecosystem looks and functions today is both a product of current interactions between the components of that ecosystem, and the result of its unique history, including climatic and cultural changes over hundreds and thousands of years (Egan and Howell, 2005; Russell, 1997).  In North America, conservationists tend to focus on the past few centuries of change brought about following colonization by European settlers:  introduction of exotic species, fire suppression, and landscape conversion for agriculture and urban development (e.g. Nowacki and Abrams, 2008; Wilcove et al., 1998).  Understanding the consequences of these massive recent changes is necessary for understanding how best to conserve endangered ecosystems.  However, looking back even farther can provide important context in understanding an ecosystem:  its origin, its range of variability through time, the nature of its response to past climatic or cultural shifts, and legacies of these changes that remain today (Foster et al., 1990, 2003; Landres et al., 1999; Swetnam et al., 1999).  Compiling this kind of long-term history is not easy, because it requires a layering of data sources with different temporal and spatial scales.  But doing so can reveal surprises that challenge our ideas about how an ecosystem functions, with implications for conservation and restoration both in terms of setting goals for the system, and in designing strategies to meet those goals (e.g. Delcourt and Delcourt, 2004; McCune et al., 2013b; Rick and Lockwood, 2013; Swetnam et al., 1999).  The Garry oak ecosystems of southwestern British Columbia are highly endangered due to habitat destruction and fragmentation, introduced species, and fire suppression (Fuchs, 2001; Lea, 2006; Parks Canada Agency, 2006a).  In Canada, these ecosystems are limited to a very small area on southeastern Vancouver Island and the Gulf Islands in the rainshadow of the   127Olympic and Vancouver Island mountain ranges, where the climate is drier than anywhere else on the coast of British Columbia (Erickson, 2008; Meidinger and Pojar, 1991).  They represent the northernmost edge of the range of Garry oak (Quercus garryana), which has become the flagship species for a complex of associated vegetation types including savannahs, oak woodlands, meadows, vernal pools and coastal bluffs (Fuchs, 2001; GOERT, 2011).   It is estimated that over 90% of the Garry oak ecosystems present just prior to European settlement have been lost to agricultural or urban land use (Lea, 2006).  The remnants are highly fragmented, invaded by introduced exotic species, and concentrated in higher elevation, rocky areas that are less suitable for agriculture or urban development (Parks Canada Agency, 2006a; Vellend et al., 2008).  The few oak savannahs left on deep soil sites are susceptible to infilling by the conifer Douglas-fir (Pseudotsuga menziesii; Fuchs, 2001).    This region has a rich human history long before the arrival of Europeans.  The indigenous cultures of the Northwest Coast of North America were highlighted by early anthropologists as rare examples of high-density, sedentary, socially stratified societies that had developed complex traditions of technology and art although they lacked agriculture (Ames and Maschner, 1999; Deur, 2002).  This idea likely contributed to the assumption on the part of most colonist Europeans that the Coast Salish peoples of southeastern Vancouver Island and the Gulf Islands did not manage the land in any way, but merely harvested nature?s bounty (Deur, 2002; Deur and Turner, 2005).  Yet, ethnographic and historical research suggests that Garry oak savannahs, particularly on deep soil sites, were maintained by frequent, low intensity fires set purposely by people (Turner, 1999; MacDougall et al., 2004).  These fires preserved the open conditions that favoured important food plants like camas (Cammassia spp.; Turner, 1999; Turner and Kuhnlein,   1281983).  Pollen records show the continued presence of oak savannah vegetation in the region even after a cooling, moistening trend approximately 6,000 years ago that favoured moisture-loving coniferous forests (Pellatt et al., 2001).  Human management may have contributed to the maintenance of oak savannahs for thousands of years (Chapter 2).  However, the decimation of the Coast Salish population by introduced diseases, and European-enforced fire suppression, put an end to management by fire (Harris, 1994; MacDougall et al., 2004; Turner, 1999).  The long-term history of Garry oak ecosystems in British Columbia is relatively well-known (Chapter 2; McCune et al., 2013b).   However, several important questions remain unanswered.     First, what was the degree of variation in space and time in the extent of open vegetation types before the first land surveys?  Land survey records have shown that open vegetation types with low or very low tree density were much more widespread at the time of European settlement than they are today (Bjorkman and Vellend, 2010).  But the range of variation prior to that is not known.  When did infilling by conifers begin, and can this infilling be linked to cultural and/or climatic changes?  Were the areas of open vegetation types consistently open going back through time, or was there a shifting mosaic of prairies, savannahs, and forest?  Second, what was the composition of the understorey of savannahs prior to the introduction of exotic agronomic grasses, which now dominate?  Land survey records are mostly mute on this question, as the surveyors usually made reference only to trees.  Historical references mention large swards of clover (MacDougall et al., 2004), while some researchers suggest native bunchgrasses had an important role (Maslovat, 2002), but there is little quantitative evidence.   129 Finally, how frequent was fire on the landscape in the past?  The ethnographic literature suggests fire was used as a tool following the harvest of important plant foods like camas, and may have been as frequent as yearly or every few years (Turner, 1999).  One study using fire-scarred tree rings found mean fire return intervals of approximately 7 years between 1700 and 1879 (Sprenger and Dunwiddie, 2011).  However, low-intensity fires may not always produce fire scars, and dendroecological data can only extend back as far as the oldest tree in a stand.  Charcoal records from lake sediment cores have yielded mean fire return intervals ranging from 27 to 88 years (Lucas and Lacourse, 2013; McCoy, 2006; Pellatt et al., 2007), but the low-fuel, low-intensity fires characterizing savannah management may not have produced sufficient charcoal to be detected above the background charcoal signal.  Therefore other potential evidence for historical fire frequency needs to be examined.   In this chapter, I use the soil phytolith record to address these questions.  Phytoliths are microscopic silica casts formed when hydrated silicon dioxide (SiO2?nH20) is deposited within and between the cells of a living plant (Pearsall, 2000; Prychid et al., 2003).  They are released into the soil upon decomposition of the plant, and can survive intact for thousands of years (e.g. Blinnikov et al., 2002).  Phytoliths generally represent local vegetation patterns, as they do not travel as far as pollen (Blinnikov et al., 2013; Pearsall, 2000).  In Chapter 5, I showed that the unique asterosclereid phytolith produced in Douglas-fir needles allows the reliable differentiation of a current forest from a nearby savannah based on surface soil phytolith assemblages (McCune and Pellatt, 2013).  Now I explore the use of the phytolith record from a larger sample of soils to address the questions listed above.   130 First, I sample surface soils from a wide range of vegetation types throughout the historical range of Garry oak vegetation in order to develop a phytolith index that reliably differentiates between current vegetation types.  Then I examine changes in this index with depth in seven soil cores from sites where historical ecological research provides information on past vegetation change, or lack of change.  If the phytolith record is an effective proxy for vegetation change, I expect the phytolith index to shift with depth according to known vegetation changes.   I can then use shifts at other sites to infer the nature and timing of vegetation change.  I also examine the change in the proportion of bilobate phytoliths with depth in the soil cores.  In this region, bilobates are produced only by two native bunchgrasses (Chapter 5; McCune and Pellatt, 2013).  Therefore, if bunchgrasses were a more significant component of the understorey vegetation of savannahs in the past, I expect them to increase in proportion with depth.  Finally, I test the use of discoloured phytoliths as an index of fire frequency.  Phytoliths often become darkened when burnt, and a high proportion of burnt phytoliths has been used as an indication of past fire in other systems (e.g. Boyd, 2002; Morris et al., 2010a).  I calculate the proportion of discoloured phytoliths in the soil core samples, including one site that was experimentally burnt each year for over a decade, to investigate whether or not discoloured phytoliths are a reliable indication of fire in this system.  131 Table 6.1:  Site location and characteristics of the 25 sites used in the study. Cells are empty where information was not available.  Tree densities are reported in trees per hectare. Sites in bold are cores for which sub-surface layers were analyzed. SITE REGION LOCATION %  GRASS COVER %  DOUGLAS-FIR COVER %  OAK COVER BROAD VEGETATION TYPE 1859 SURVEYOR DESCRIPTION 1859 TREE DENSITY 2007 TREE DENSITY 316 Saanich Mount Douglas Park 100 0 55 savannah    304 Saanich Mount Douglas Park 95 0 85 savannah    SOM1 Cowichan Somenos Preserve 80 0 80 savannah oak plains 13-102   330 Saanich Summit Park 72 0 40 savannah    NCC2 Cowichan Cowichan Preserve 65 0 90 savannah rich oak plains 13-102  NCC1 Cowichan Cowichan Preserve    savannah rich oak plains   303 Saanich Mount Douglas Park 44 0 85 savannah    335 Saanich Bear Hill Park 11 0 25 savannah    16 Saanich Mount Douglas Park 0 5 0 other    356 Saanich John Dean Park 17 10 32 transition    COW4 Cowichan Mountain View Cemetery 1 12 0 other wood land with fallen small timber 13-102 662 363 Saanich John Dean Park 44 35 35 transition    108 Saanich Francis King Park 8 40 20 transition    288 Saanich Bear Hill Park 42 70 0.5 transition    365 Saanich John Dean Park 17 55 0.5 transition    SOM2 Cowichan Somenos Preserve 1 60 3 transition oak plains 13-102 232 COW12 Cowichan Arbutus Ridge 1 55 0 forest good open land   COW10 Cowichan Providence Farm 0.5 65 0 forest pine plains 13-102 593 COW14 Cowichan Private land 0.5 70 0 forest open pine plains 13-102 1369 163 Saanich Municipal Forest 0.5 80 0 forest    SOM3 Cowichan Somenos Preserve 1 85 0 forest oak plains 13-102  COW7 Cowichan Municipal Forest 0.5 85 0 forest open woods 13-102 748 57 Saanich Elk Lake Park 0 90 0 forest    COW1 Cowichan Private land 0.5 85 0 forest thick heavy timber 405-700 1297 COW2 Cowichan Private land 0.5 85 0 forest heavily timbered 405-700 378   132 Figure 6.1:  Map of southeastern Vancouver Island showing the location of soil samples (red dots), and the estimated historical range of Garry oak ecosystems (grey shading; Lea, 2006).  6.3 Materials and methods 6.3.1 Site selection I took 25 soil samples within the historical range of Garry oak ecosystems on southeastern Vancouver Island (Lea, 2006; Fig. 6.1, Table 6.1).  I selected sites based on three criteria.  First, I   133aimed to sample a range of current vegetation, from open Garry oak savannahs with few trees and a grassy understorey, through transition vegetation with open oak savannahs in various stages of encroachment by Douglas-fir, through closed canopy Douglas-fir forests.  I also included two sites that are heavily forested, but with minimal cover by Douglas-fir (COW4, dominated by bigleaf maple, Acer macrophyllum, and plot 16, dominated by grand fir, Abies grandis).  Second, I selected some sites where the vegetation type and tree density in 1859 and in 2007 had been quantified in a previous study using the original surveyor notes and maps from the first land survey of the Cowichan Valley (Bjorkman, 2008; Bjorkman and Vellend, 2010).  Third, I chose deep-soil oak savannahs at two locations thought to have been minimally disturbed (e.g. not ploughed) since European settlement.  Both of these sites are now protected, one by the province of British Columbia (Somenos Garry Oak Preserve) and the other by the Nature Conservancy of Canada (Cowichan Garry Oak Preserve).  The former is also adjacent to the most well researched archaeological site in the Cowichan Valley, with a large shell midden (Brown, 1996).  I chose some sites based on quantitative plant cover data from an earlier survey of 184 20x20m vegetation plots (Chapter 3; McCune and Vellend, 2013).  The original notes taken during the 1859 land surveys are kept at the archives of the British Columbia Land Title and Survey Office in Victoria.  Land surveyors walked gridlines across the landscape, and placed posts at the intersections to mark property boundaries.  The distance and direction of bearing trees in relation to these posts were noted in case the post was lost (Whitney and Decant, 2005).  The 1859 description in Table 6.1 is based on the surveyor?s handwritten notes at that precise location and/or the descriptive text in that area on the accompanying maps, as compiled by Anne Bjorkman (2008).  Note that the surveyors used the term ?pine? when   134referring to Douglas-fir trees (Bjorkman, 2008).  Bjorkman also calculated the tree density for landscape descriptions used consistently in the notes by using the bearing tree data from all points described using the same phrase.  For example, areas described as ?oak plains? or ?pine plains? had tree densities of between 13 and 102 trees per hectare (Bjorkman, 2008; Bjorkman and Vellend, 2010).   Bjorkman revisited the same gridline intersections in 2007, and calculated current forest density based on the 10 nearest trees.  Where available, this 2007 forest density is also shown in Table 6.1. I chose some sites which have maintained a consistent character of either Douglas-fir forest or open savannah since 1859, and some which were ?plains? in 1859, but had become forest by 2007.  By analyzing soil phytoliths from these locations I can test whether the documented historical changes are also evident in the phytolith record.    At each chosen site, I set up a 20x20m vegetation plot.  Phytoliths may be deposited in soil from an area as large as 1 ha or even larger (Blinnikov et al., 2002; Fredlund and Tieszen, 1994), however I chose the 20x20m plot size as it is a more manageable size for plant surveys, and it has been shown to be large enough to accurately sample the species richness of both Douglas-fir forest and oak savannah communities (Roemer, 1972).  I recorded a list of all vascular plant species present in the plot, and estimated percent cover to the nearest 1%.  I then took a composite soil surface sample by collecting a small amount of soil from within the top 2 cm (after removing leaf litter) near each of the four corners of the plot.  Finally, I extracted at least one 5cm diameter soil core from somewhere within the plot using a multi-stage sediment sampler with a slide hammer (AMS, American Falls, ID, USA).  I selected areas for coring that appeared to have minimal soil disturbance, were relatively flat, and free of visible rocks.  Soil samples were refrigerated until they could be processed for phytoliths.   135 At the Cowichan Garry Oak preserve I took two cores (NCC1 and NCC2).  One was within the sampled vegetation plot, and one was from the center of a 1x1m quadrat that had been experimentally burned each year since 2000 for 11 years (MacDougall et al., 2013).  At the Somenos Garry Oak Preserve I took three cores (SOM1, SOM2, and SOM3), each within its own 20x20m vegetation plot.  I took these cores along a transect proceeding northwards, from the grassy, open-canopied oak savannah at the southern end of the property and nearest the ancient shell midden (SOM1) up to what is now quite a dense Douglas-fir dominated forest (SOM3).  SOM2 is currently in an area with a few very large old Garry oak trees that have been completely surrounded by younger Douglas-firs (Fig. 6.2).  Figure 6.2:  Location of soil cores taken from the Somenos Garry Oak Preserve showing the landscape in 1859 (left) and today (right).  The description on the 1859 map to the East of Somenos Lake reads ?Oak Plains?.  Non-marshy sites that are currently still naturally vegetated have mostly become filled in with thick Douglas-fir forest (darkest areas in photo at right).     136 6.3.2 Phytolith extraction I used a wet oxidation and heavy liquid flotation procedure modified from Pearsall (2000) to extract phytoliths.  I filtered each soil sample through a 500 micron sieve after air-drying overnight.  Approximately 5g of the dried, filtered soil was processed for each sample.  I removed carbonates by heating in 10% hydrochloric acid for one hour.  I removed organic material with a five minute bleach treatment followed by heating in 30% hydrogen peroxide until no further reaction was visible.  This required several hours for highly organic samples.  I dried and weighed each sample following organic removal to estimate the inorganic fraction.  I then dispersed clay particles by shaking for 24 hours in a 0.1% solution of ethylenediaminetetraacetic acid (EDTA) disodium salt dihydrate.  I removed clay particles by sedimentation in a 10cm column of water for 9.3 hours followed by removal of the supernatant.  I repeated the sedimentation process at least 4 times.  Finally, I separated phytoliths using heavy liquid flotation with sodium polytungstate at 2.3g/cm3 density.  I decanted the supernatant (with suspended phytoliths), and then sank the phytoliths by adding distilled water and centrifuging.  I carried out this process six times to ensure complete flotation of phytoliths.  I rinsed the final extract twice, and then dried it at approximately 60? C in a drying oven until all water had evaporated (2-3 days).  I weighed the final extract and stored it in glass vials.  For more details on the phytolith extraction protocol, see Appendix B.2.    I chose seven full soil cores to analyze changes in phytolith assemblages with depth (Table 6.1).  I also extracted phytoliths from the top 12 cm of the NCC1 core.  I sliced each core in half, photographed it, and then divided it into 2cm increments for analysis.  I extracted phytoliths   137from every other 2cm increment starting with the 0-2cm increment, excluding the bottom 3cm (except I extracted phytoliths from all 2cm increments to a depth of 12cm for NCC1).  6.3.3 Phytolith counts Based on the reference collection of phytoliths produced by plants in the region (Chapter 5; McCune and Pellatt, 2013), I counted five different phytolith morphotypes:  elongates (produced almost exclusively by grasses), rondels and bilobates (produced by grasses only), asterosclereids (produced by Douglas-fir only) and ?other? (Fig. 6.3).  The final category included hair and hairbase phytoliths, conical Carex-type phytoliths, tracheid phytoliths, and various rare unknown phytoliths (Fig. 6.3; also see Fig. 5.1 and 5.2 in Chapter 5, and Appendix B.4).  I mounted between 0.5mg and 1.2mg of phytolith extract on a microscope slide (measured to the nearest 0.1mg using an analytical balance) in 3-4 drops of Canada balsam mounting medium.  I scanned the entire slide at 200x magnification to count the large asterosclereid phytoliths produced by Douglas-fir.  I then counted other phytolith morphotypes at 400x magnification across a transect of 16-18 consecutive microscope fields from the centre of the cover slip to the edge.  For each phytolith morphotype other than asterosclereids, I calculated the weighted average number per field by weighting each field by its distance from the cover slip centre (see Chapter 5).  I then multiplied this weighted average by the number of fields that would be needed to cover the entire area of the cover slip.  I obtained the mean and standard error for the estimated number of each phytolith morphotype per slide, and for the ratio of asterosclereids to rondels, using a bootstrapping procedure with 1000 runs.  I calculated concentrations of each morphotype per gram of soil using these means based on the weight of the final extract and the initial weight of the soil sample.   138  Figure 6.3:  :  Examples of the five phytolith morphotypes counted in the study, (a) elongate, (b) rondel, (c) bilobate, (d) asterosclereid, (e) a tracheid phytolith that would be counted in the ?other? class.  The phytoliths in (d) and (e) would be considered brown phytoliths.  Phytoliths are usually colourless and transparent, but may become brown or blackened and more opaque when exposed to fire (Parr, 2006).  The ratio of burnt to unburnt phytoliths in a sample (as a percentage), called the Burnt Phytolith Index (BPI), can indicate the prevalence of fire   139(Boyd, 2002).  I counted the total number of brown or darkened phytoliths from each of my soil core samples in order to calculate the BPI (Fig. 6.3).  6.3.4 Analyses I assessed site-to-site variation in vegetation composition by ordinating plots using non-metric multidimensional scaling (NMDS) with Bray-Curtis dissimilarities based on square root transformed cover values of all vascular plant species.  Then I superimposed vectors for the total percent cover of all grasses and the percent cover of Douglas-fir (the two main phytolith producing groups) to confirm that these are strongly correlated with the main gradient of plant community composition.  I then tested for correspondence between surface phytolith samples and vegetation composition by plotting the concentration and proportions of the five different phytolith morphotypes versus the difference between the percent cover of Douglas-fir and the percent cover of grass for each plot.  I tested for pairwise differences in concentration or proportion of each morphotype between the three broad vegetation types using Wilcoxon rank sum tests.  I used this data to establish an index that best discriminated between the vegetation types.  Lastly, in the complete soil cores I examined the changes in this index, the weight of the inorganic soil fraction, the total concentration of phytoliths, the BPI, and the proportions of each phytolith morphotype with depth.  All analyses, including bootstrapping, were carried out in R, using the ?vegan? package for NMDS ordination (Oksanen et al., 2009; R Core Development Team, 2012).   140 6.3.5 Radiocarbon dating For five of the seven complete soil cores, I obtained radiocarbon dates for 2-3 small pieces of charcoal or wood extracted from the soil.  The number of samples was limited due to the expense of radiocarbon dating.  Beta Analytic, Ltd. (Miami, Florida) processed the samples to obtain accelerator mass spectrometry (AMS) 14C ages.  I calibrated the reported conventional radiocarbon ages with the OxCal calibration program using the INTCAL09 calibration curve (OxCal version 4.2; Bronk Ramsey, 2009; Reimer et al., 2009; Table 6.2).  Table 6.2:  AMS (accelerator mass spectrometry) radiocarbon and calibrated calendar ages of charcoal or wood samples from soil cores taken within the historical range of Garry oak ecosystems on southeastern Vancouver Island, British Columbia. Core Depth below surface (cm) Lab number Material Radiocarbon age  (14C years BP ? 1?) Calendar age  (cal years BP)* SOM1 12 Beta-322820 charred material 60 ? 30 93 ? 77 SOM1 34 Beta-327609 charred material 2010 ? 30 1959 ? 38 SOM2 10 Beta-322821 wood 220 ? 30 181 ? 98 SOM2 22 Beta-322822 charred material 2050 ? 30 2013 ? 48 SOM3 4 Beta-327610 charred material 120 ? 30 119 ? 80 SOM3 24 Beta-322824 charred material 720 ? 30 672 ? 28 COW2 12 Beta-351518 charred material 3269 ? 30 3480 ? 44 COW2 24 Beta-350718 charred material 2390 ? 30 2417 ? 78 COW2 48 Beta-350719 charred material 3280 ? 30 3509 ? 40 COW14 16 Beta-350720 charred material 1850 ? 30 1784 ? 43 COW14 24 Beta-350721 charred material 2860 ? 30 2978 ? 53 *shown is the median age ? 1? with 95.4% probability as calibrated by OxCal (Bronk Ramsey, 2009).  6.4 Results 6.4.1 Surface calibration The NMDS ordination of current vegetation confirmed that the total cover of Douglas-fir and the total cover of all grasses are highly correlated with the main vegetation gradient in these plots   141(Fig. 6.4).  The sites separate out in vegetation space as I expected, with ?transition? plots located between oak savannahs and Douglas-fir forests in plant species space.  Figure 6.4:  Non-metric multidimensional scaling ordination of all soil sample sites based on current vascular plant species composition within a 20x20m plot.  The Bray-Curtis dissimilarity measure was used after standardization and square root transformation.  The ordination has been centred, the axes scaled to half-change units, and the configuration rotated so that the variation among plots is maximized on axis NMDS1.  The vectors show the direction of correlation of the total grass cover (?GRASS?) and total Douglas-fir cover (?D.FIR?) with the ordination axes.  Stress in three dimensions is 7.37, which indicates a good correlation of distances in ordination space with the calculated ecological dissimilarities (McCune and Grace, 2002).  The current vegetation types showed differences in the concentration of elongate, rondel, bilobate and asterosclereid phytoliths per gram of surface soil (Fig. 6.5).  On average, the amount of grass-produced phytoliths increased from dense Douglas-fir forest plots towards plots with   142more grass cover.  However, there was quite a lot of overlap between vegetation types, and the concentration of any one grass-produced phytolith morphotype did not reliably distinguish between vegetation types.  For example, the median number of rondels in the surface soils of ?transition? and ?other? plots was approximately the same as that for savannah vegetation plots, and the two distributions were not significantly different (Fig. 6.5).  Asterosclereid phytoliths were almost always absent from surface soils under savannah vegetation, and the distributions of asterosclereid concentration differed between all three vegetation types; however the range of overlap in asterosclereid concentration between Douglas-fir forest and ?transition? vegetation was quite large (Fig. 6.5).    Figure 6.5:  Differences in total phytoliths per gram of surface soil estimated for each phytolith morphotype by vegetation type, Douglas-fir dominated forest (?forest?), ?transition? or ?other? vegetation types (?trans/oth?), and Garry oak savannah (?savannah?).  Note that the scale of the y axis differs for each plot.  Different letters above the boxes indicate significantly different distributions according to pairwise tests.    143Similar patterns were observed when expressing phytolith counts of each type as a proportion of the total (Fig. 6.6).  As I found in Chapter 5, elongates and rondels made up the bulk of the phytoliths in my samples.  Douglas-fir plots tended to have lower proportions of rondels and bilobate types, but they had the highest median proportion of elongates of all three vegetation types (Fig. 6.6).  Bilobates, which are produced only by two native bunchgrasses that are now extremely rare (Achnatherum lemmonii and Danthonia californica), represented less than 5% of phytoliths counted in all surface soil samples.  Each of these two species was found in only one plot, with less than 1% cover in both cases.  Phytoliths classed as ?other? represented on average less than 2% of phytoliths in surface samples.  The majority of ?other? phytoliths (66%) counted in both surface and core samples were hairbase phytoliths (see Appendix B.4).  Figure 6.6:  Differences in the percentage of all phytoliths represented by each morphotype in surface samples by vegetation type, Douglas-fir dominated forest (?forest?), ?transition? or ?other? vegetation types (?trans/oth?), and Garry oak savannah (?savannah?).  Note that the scale of the y axis differs for each plot.  Different letters above the boxes indicate significantly different distributions according to pairwise tests.    144 Other studies have successfully derived phytolith-based indices of relative forest cover using ratios of dicotyledon- to grass-produced phytolith morphotypes (e.g. Alexandre et al., 1999; Delhon et al., 2003).  I examined the ratio of asterosclereids, produced solely by Douglas-fir, and rondels, the most abundant phytolith morphotype produced solely by grasses.  I found that the log ratio of these two morphotypes (ln(A:R)) quite clearly differentiates the three broad vegetation types, with little overlap (Fig. 6.7a).  I calculated approximate threshold lines between the three vegetation types by halving the distance between the uppermost and lowermost values (plus or minus the bootstrapped standard error) between each of the three groups.  When estimating these thresholds, I excluded the two plots classified as ?other? (COW4 and plot 16) as well as plots COW12 and 57, which had unexpectedly lower asterosclereid concentrations compared to the other Douglas-fir dominated plots (Fig. 6.7b).   These thresholds correspond to approximate asterosclereid:rondel ratios of less than 1:1250 for savannah vegetation, between 1:1250 and 3:500 for ?transition? vegetation, and above 3:500 for Douglas-fir vegetation.  The largest asterosclereid to rondel ratio was approximately 1:10 for plot COW1 (Fig. 6.7b).  6.4.2 Full core analysis According to detailed soil maps of the region, the locations of the seven full cores I analyzed are dominated by dystric brunisols that are strongly acidic (Jungen, 1985).  Brunisols, as classified by the Canadian System of Soil Classification, are called ?Inceptisols? by the US Soil Taxonomy system (Smith et al., 2011).  Brunisolic soils are typically found in semiarid regions where low soil moisture results in limited chemical weathering (Valentine et al., 1978).  The brunisols of southeastern Vancouver Island are characterized by dark brown to black surface horizons that   145result from the accumulation of high amounts of organic matter because of high litter inputs and slow decomposition due to low moisture conditions (Broersma, 1973).  I do not have evidence that these soils have been aggrading over time, but the presence of clearly defined soil horizons in five of my seven full cores, and an increase in the inorganic fraction with depth in all cores is evidence against significant soil mixing or the presence of buried surface horizons (Fig. 6.8-6.14).  In addition, in four of the five cores with radiocarbon dates, older charcoal is found below younger charcoal (Table 6.2).   I found a steep decline in phytolith concentration once entering the B horizon of soil cores, if present (Figs. 6.8-6.14).  This is common pattern of phytolith distribution in soils (Hart and Humphreys, 2003; Jones and Beavers, 1964).   Current savannah sites SOM1 and NCC2, currently oak savannah sites, both maintain ln(A:R) ratios in the savannah vegetation range throughout the length of the cores, suggesting long-term absence of Douglas-fir.  The SOM1 core was 59cm in length in total, and the entire length of the core was dark organic soil (Fig. 6.8).  The NCC2 core was 45cm long with an abrupt change from dark, organic soil to light yellowish, hard clay soil at 12cm (Fig. 6.9).  I obtained radiocarbon dates on two charcoal samples from SOM1.  The deepest (34cm) had a calibrated age of 1959 years before present (BP, where ?present? is considered the year 1950; Table 6.2).   SOM2 transition site  SOM2 is less than 200m away from SOM1 and is currently dominated by young (<100 year old) Douglas-firs surrounding a few very old oaks.  The core was 49cm long with a dark organic soil from the surface to 14cm, followed by a gradual transition to a yellowish clay horizon that began    146 Figure 6.7:  (a) The log ratio of asterosclereid to rondel phytoliths in surface samples by vegetation type, Douglas-fir dominated forest (?forest?), ?transition? or ?other? vegetation types (?trans/oth?), and Garry oak savannah (?savannah?).  (b) The log ratio of asterosclereid to rondel phytoliths in surface soil samples plotted against the difference in current percent cover of Douglas-fir and all grasses.  Dotted lines are the estimated thresholds between vegetation types.  Error bars show ? the bootstrapped standard error.  Some samples have more than one estimate if their composite surface soil sample and the 0-2cm increment of the soil core were both analyzed, see methods.   147around 32cm.  The ln(A:R) ratio begins in the transition zone near the surface, but then falls below the threshold to savannah levels by 14cm below the surface.  This is what I would expect for a former savannah site recently encroached upon by Douglas-fir.  The ratio generally remains at this level, although there is greater uncertainty in deeper layers where the overall phytolith concentration is quite low (Fig. 6.10).  I obtained radiocarbon dates on two charcoal samples from SOM2, the deepest (from 22cm depth) had a calibrated age of 2013 years BP (Table 6.2).  SOM3 and COW14 These cores were taken at sites currently dominated by Douglas-fir forests, but described as ?oak plains? or ?open pine plains? at the time of the first land survey (Table 6.1).  The SOM3 core (total length 63cm) had a very shallow layer of dark organic soil about 6cm deep, followed by a transition to a reddish clay layer from 16-22cm in depth.  Below this was a heavily charred layer of about 4cm that contained high amounts of charcoal and burnt wood (Fig. 6.11).  There was a spike in phytolith concentration just before this layer, which I suspect was due to a build-up of phytoliths prevented from moving further down due to impermeability of the charcoal layer.  Below the charred layer was an abrupt transition to a yellowish clay horizon at approximately 28cm below the surface.  The ln(A:R) ratio is well above the threshold of Douglas-fir forest until 16cm below the surface, where it drops down into the ?transition? zone (Fig. 6.11).  This matches the expectation for a site that was formerly ?open plains?.  Below 30cm there are minimal phytoliths, and no asterosclereids were observed, dropping the ratio down below the ?transition? threshold.  I obtained radiocarbon dates on two charcoal samples from SOM3, the deepest (from within the charred layer at 24cm depth) had a calibrated age of 672 years BP.    148COW14 also had a shallow layer of dark organic soil for the top 6cm, and then gradually changed to a yellowish clay layer by about 20cm in depth.  Unlike SOM3, the ln(A:R) ratio is maintained within the Douglas-fir forest zone for almost the entire core length, although the site formerly supported ?plains? vegetation.  The phytolith concentration is very low below 28cm, and the ratio estimates become more uncertain (Fig. 6.12).  No rondels were observed in the 32-34cm layer or from 40-52cm.  A charcoal sample from 24cm below the surface was estimated at 2978 calibrated years BP (Table 6.2).   Long-term forested sites COW1 and COW2 cores were both taken from sites that are now and were in 1859 dominated by Douglas-fir forest (Table 6.1).  The soil at COW1 was extremely rocky, and I was unable to extract more than 28cm of soil.  This core did not show any definite horizons, and appeared mixed and rocky throughout, although there was more yellowish clay-coloured soil near the bottom of the core, and the organic fraction declined with depth (Fig. 6.13).  This core contained an abundance of visible charcoal.  Throughout this core, the ln(A:R) ratio is maintained well into the forest zone, as predicted for a long-term forested site.  Although the organic fraction declined with depth, the phytolith concentration did not (Fig. 6.13).  At COW2 I was able to obtain a 53cm core.  This core had a shallow dark organic layer to 6-8cm below the surface, followed by a reddish hard clay horizon from 8cm to about 36cm, and then a yellowish clay layer below that.  The phytolith concentration falls dramatically in this lowest layer (Fig. 6.14).  The ln(A:R) ratio in the top 6cm was well above the forest threshold, but declined into the ?transition? zone below that, even dipping below the ?savannah? level a few   149times.  A charcoal sample from near the bottom of this core (48cm) yielded an estimated age of 3509 years BP.  However, a sample from 12cm was dated to nearly the same age (Table 6.2, Fig. 6.14).    Figure 6.8:  The SOM1 core showing, top: the opened core; middle: the change with depth in % organic by weight (filled circles with dotted line smoothed using locally weighted regression) and total phytolith concentration (open circles with solid line); and bottom: change in the log ratio of asterosclereids to rondel phytoliths (ln(asteros:rondels)) with depth.  Error bars are ? the boostrapped standard error.  Grey vertical lines indicate layers with radiocarbon dated charcoal, the dates above the lines indicate median calibrated years before present (cal BP).    150  Figure 6.9:  The NCC2 core.  See caption for Fig. 6.8.  No radiocarbon dates were obtained for this core.      151 Figure 6.10:  The SOM2 core.  See caption for Fig. 6.8.      152 Figure 6.11:  The SOM3 core.  See caption for Fig. 6.8.      153 Figure 6.12:  The COW14 core.  See caption for Fig. 6.8.       154 Figure 6.13:  The COW1 core.  See caption for Fig. 6.8.  No radiocarbon dates were obtained for this core.    155 Figure 6.14:  The COW2 core.  See caption for Fig. 6.8.        156Burnt phytoliths and morphotype proportions with depth The ratio of brown to colourless phytoliths in the core samples was very low overall:  the BPI ranged from 0% to 14%, with an average of 2.4%.  I noticed more visible small burnt charcoal pieces in the top 4cm of the NCC1 core, which had been experimentally burnt, than any other cores.  However, the BPI was low throughout the top 12cm of this core, with a maximum of only 3.5%.   The BPI for SOM3 was highest in the charcoal layer at 24cm below the surface, but it was only 3.4%, not much higher than a value of 3.2% reached at a depth of 16cm.  The proportion of elongate, rondel and bilobate phytoliths was relatively stable with depth, except for NCC2, which showed a spike in the bilobate proportion between 16 and 22cm depth (Fig. 6.15a,b,d).  The proportion of phytoliths represented by asterosclereids generally declined with depth for all cores with a significant concentration of asterosclereids (Fig. 6.15c).   157 Figure 6.15:  Changes in the proportion of all phytoliths represented by (a) elongate phytoliths, (b) rondels, (c) asterosclereids, and (d) bilobates with depth in the first 25cm below the surface of all cores.  Legend is shown in (d). Asterosclereid proportions are shown only for cores with significant amounts of asterosclereids.    1586.5 Discussion 6.5.1 The landscape before 1859 This study shows that phytoliths from soils are a promising tool to document local shifts in vegetation over the past several centuries.  The log ratio of asterosclereids, produced by Douglas-fir, and rondels, produced by grasses, extracted from surface soils is a remarkably accurate indicator of the current vegetation community.  In addition, the phytolith record was sensitive to known changes, or lack of change, since the first land surveys for most of the soil cores.  The two deep-soil sites currently under oak savannahs show no evidence of Douglas-fir presence for at least the last 2000 years.  Both SOM1, at the Somenos Garry Oak Preserve, and NCC2, at the Cowichan Garry Oak Preserve, maintain very low concentrations of asterosclereid phytoliths throughout their depth (Fig. 6.8, 6.9).  I cannot be certain about the density of oak or other trees on the landscape throughout this period, but I am confident that Douglas-fir has not been a significant component of the vegetation on these sites for a very long time.  The profile for SOM2, just uphill from SOM1, matches expectations for a savannah recently filled in with Douglas-fir.  The crossing of the ln(A:R) ratio into the ?transition? zone coincides with a charcoal sample dated 181?98 calibrated years BP, which coincides roughly with the time of European settlement, or just before (1671-1867AD).  This timeframe overlaps with a particularly wet period that occurred on southern Vancouver Island from the 1560s to the 1760s, at the end of the cold period known as the Little Ice Age (Zhang and Hebda, 2005).  Dendroecological studies have also documented pulses of oak and Douglas-fir recruitment at various sites in the region in the early- to mid-1800s, and generally attribute these pulses to fire   159suppression, climatic changes, changes in herbivory levels, or a combination of these factors (Gedalof et al., 2006; Dunwiddie et al., 2011; Smith, 2007).   The coincidence of the wet period with changes in human management due to Coast Salish population decline and European land appropriation make it difficult to determine which was responsible for the infilling of the savannah at SOM2 with Douglas-fir.  However, it appears that the site was an open savannah for at least 2000 years prior to European colonization.  The SOM3 site is currently a Douglas-fir forest with little grass cover, which is in agreement with the high ln(A:R) ratio in the surface layer of soil.  However, by 16cm below the soil surface, the ratio has dropped into the ?transition? zone (Fig. 6.11).  The increase in Douglas-fir asterosclereids occurs above a charcoal fragment dated at 672 calibrated years BP (1250-1306AD), but below charcoal dated at 119 calibrated years BP (1751-1911AD).  This suggests that the transition may have happened well before European settlement in the region, perhaps triggered by the onset of the Little Ice Age climate anomaly, which brought higher precipitation and lower growing season temperatures from approximately 1400AD (Mann et al., 2009).  Several charcoal and tree ring-based studies have found reduced fire frequencies in the region during this time (e.g. Brown and Hebda, 2002; Lucas and Lacourse, 2013), which would favour increased recruitment of young Douglas-firs.  It is curious that the entire landscape east of Somenos Lake is described as ?oak plains? on the 1859 map (Fig. 6.2), given this inferred increase in Douglas-fir at SOM3 well before the original surveys.  However, the bearing tree for the gridline intersection nearest SOM3 was a Douglas-fir, as was the bearing tree for the next three intersections heading north from SOM3.  The   160description at the intersection less than 500m north of SOM3 reads: ?oak and pine plains, excellent land?.  It is clear that there was a significant presence of Douglas-fir in this area at the time.  The COW14 core was taken from a site described as ?open pine plains? in 1859, which now has a forest density of over 1300 trees/ha (Table 6.1).  I was surprised, therefore, to find that the ln(A:R) ratio remains at a high level throughout the depth of this core, and particularly in the depth range before the sharp decline in phytolith concentration (Fig. 6.12).  A charcoal sample from the 24cm layer was dated to 2978 calibrated years BP (1081-975 BC).  This suggests Douglas-fir forest vegetation has existed here for the past 3,000 years, which does not match with historical descriptions of a ?pine plain? in 1859.  There are two possibilities that could explain this finding.  It could be that low-density ?pine plains? contained sporadic groves of more dense forest, including the location of COW14.  However, the bearing tree for the gridline intersection at COW14 was a Douglas-fir located approximately 15m away.  That seems too far to indicate a forest with a density near current levels.  The second possible explanation is that ?pine plains? can produce ln(A:R) ratios as high as Douglas-fir forests.  These ?pine plains? were described by a surveyor as ?land of the best quality, open, and little wood upon it, which usually grows in clumps with an occasional isolated tree.  The picturesque effect being very similar to that of an extensive cultivated park? (Bjorkman, 2008).  I assume the understorey was mainly grass; however COW14 (along with SOM3) has the lowest concentration, and proportion, of rondels among the seven cores (Fig. 6.15).  The few references to understorey species made by the surveyors include mention of plants other than grass in the understory.  For example, one description reads:  ?continued in fine open plain land growing wild peas, fern and grass?.  These   161?pine plains? are practically nonexistent on the landscape in the present.  They represent a sort of no-analog community, for which we do not have an example in the present (Williams and Jackson, 2007).     Sites COW1 and COW2 both are described as being heavily timbered in 1859 and are still relatively dense Douglas-fir forests today, although COW2 has seen a slight decline in estimated tree density (Table 6.1).  The COW1 core had the highest level of asterosclereids of all core samples, and maintains a ln(A:R) ratio well into the forest zone throughout its length (Fig. 6.13).  I consider this evidence that this site has been dominated by Douglas-fir since well before European settlement, although I have no dates from this core to estimate a time frame.  COW2, on the other hand, shows a decline in the ln(A:R) ratio into the ?transition? zone by 8cm below the surface (Fig. 6.14).  The charcoal sample at 12cm yielded a radiocarbon date older than one near the base of the core (Table 6.2).  There are several processes that can lead to older charcoal being found closer to the surface, including animal burrowing and tree uprooting (Piperno and Becker, 1996).  Therefore, I cannot estimate when this more open phase in the history of COW2 may have occurred.  However, this site is right on the western boundary of the estimated historical range of Garry oak savannahs (Fig. 6.1), and it is very possible that this now forested location was more open before the surveyors came through, and may have begun filling in with Douglas-fir due to climatic change centuries before, as observed for SOM3.  Taken together, the results of the soil core analysis suggest that the occurrence of open vegetation types with little or no Douglas-fir was even higher in the centuries before the first European land surveyors came through.  I did not see any evidence of a shift in the opposite   162direction, from forested to more open and grassy.  However, the timing of infilling by Douglas-fir appears to have differed by site.  At some sites, Douglas-fir began to increase well before European settlement, while at others the increase in Douglas-fir appears to have been coincident with climate and cultural changes that occurred just before or during European settlement.  Some current savannah sites have maintained open conditions for at least two thousand years.  6.5.2 Evidence for changing abundance of native bunchgrasses I did not find evidence for increased dominance of bilobate-producing native bunchgrasses prior to European settlement except for at the Cowichan Garry Oak Preserve, plot NCC2, where the proportion of bilobates increases quite significantly below 12cm (Fig. 6.15).  This is also the location in the core where the change to harder, clay soil coincides with a dramatic decline in total phytolith concentration (Fig. 6.9).  However, bilobate and rondel phytoliths are similar in size, so I do not expect bilobates to be more prone to percolation into clay horizons than rondels.  Fredlund and Tieszen (1994) note that bilobate forms are usually underrepresented relative to other grass phytoliths in relation to the biomass of bilobate-producing grasses on the landscape.  An increase in the proportion of bilobates, therefore, may represent an even greater increase of bilobate-producing species.  Bunchgrasses were likely a larger component of the understorey vegetation in the past than they are currently at the NCC2 site.  However, my analysis does not support the idea that bunchgrasses were the most dominant component of savannah understorey vegetation throughout the Garry oak range.  It is likely that a mixture of other native grasses and herbs dominated at different sites.  Unfortunately, the phytolith record cannot provide a more detailed indication of understorey composition.    1636.5.3 Phytoliths and fire on the landscape I did not find burnt phytoliths to be an accurate indicator of fire frequency in this region.  The BPI at all depths of all soil cores was well within the range of unburned sites in other studies.  For example, in the pinyon-juniper and sagebrush steppe ecosystems of the Great Basin Desert, Morris et al. (2010a) found that the BPI of sites that had not burned in the last century ranged from 0% to 20%, with an average of 6%.  The BPI of a site that burned in the year 2000 was 38%.  In the grasslands of southern Manitoba, Boyd (2002) found that soil surface samples from areas known not to have burned in 100 years had a mean BPI of 8.2%.  He measured a BPI as high as 70% in buried soil horizons and 73% in an occupation layer of an archaeological site with abundant evidence of fire.  While I did observe fluctuations in BPI with depth in some of my cores, the BPI never exceeded 8% in the top 25cm of any core, including the one from an experimental plot burnt yearly for more than a decade (NCC1).  Despite these low BPI levels, I cannot conclude that fire has been absent from my sites.  Firstly, I know that fire was present in the last 10 years for NCC1, and at the 24cm depth for SOM3, yet I do not see corresponding dramatic increases in BPI for the NCC1 core or in the charred layer of SOM3.  Similarly, Morris et al. (2010a) were unable to find an increase in BPI with depth at sites known to have burned within the past century, and the BPI up to 10cm deep in these sites was less than 8% in all cases.  The production of significant burnt phytoliths likely depends on the extent of the fire, the temperature at which the plant material and/or the litter layer burns, and the type of vegetation present at the site (Morris et al., 2010a).  For example, a small fire (like within a 1x1m quadrat) may not contribute enough darkened phytoliths to the soil to cause a noticeable peak in the BPI.  In addition, the surface temperatures produced by fire in current oak savannahs   164range from 70-400?C, and these temperatures are sustained for less than 15 seconds (MacDougall, 2005).  These fires are considered ?cool? in contrast to fires in pinyon-juniper woodlands, where temperatures can reach 800?C and be sustained at 200-400?C for more than an hour (Morris et al., 2010a).  Finally, there is good evidence that phytoliths can be naturally brown or blackened without having been burned (Parr, 2006), and also that burnt phytoliths may not change colour (Elbaum et al., 2003).  I conclude that BPI is not a good proxy for historical fire frequency or intensity in this region.  6.5.4 The phytolith record in soil as a paleoenvironmental indicator The use of the soil phytolith record for paleoenvironmental interpretation is still relatively new compared to long-established methods like the interpretation of fossil pollen assemblages from lake sediments (Piperno, 1988).   The nature of the record stored in a soil profile is different than the record stored in a regularly deposited sedimentary record, such as a lake sediment core with a consistent sedimentation rate.  The soil phytolith record has higher spatial resolution, due to the more limited dispersal of phytoliths relative to pollen, but may have a lower degree of temporal resolution (Fredlund, 2005; Targulian and Goryachkin, 2004).  Soil formation can occur in much more irregular patterns, with periods of gradual soil accumulation followed by periods of erosion or relative stability, and is strongly influenced by the nature of the vegetation growing in the soil and the parent material on which the soil is formed (Fredlund, 2005).  The incorporation of phytoliths into the soil profile is affected by the rate of soil accumulation and/or erosion, as well as shifts in the plant community over time (Fredlund and Teiszen, 1994).      165In addition, terrestrial soil phytolith analyses must consider the movement of phytoliths downwards in the soil profile.  Phytolith researchers have often assumed that phytoliths are immobile in soil (e.g. Rovner, 1986).  They are after all quite strongly chemically and physically bonded to clays and other soil materials, which is why the protocol for extracting them from soils is so complicated (Pearsall, 2000).  However, it has been shown both experimentally using fluorescently labeled phytoliths (Fishkis et al., 2010a,b), and with observed changes in phytolith surface ornamentation with depth (Alexandre et al., 1999) that phytoliths can and do move downwards in the soil profile.  The extent of movement depends on the type of soil, the amount of precipitation, and the size of phytoliths, but can be as much as 40mm per year (Fishkis et al., 2010a).  Nevertheless, the movement of phytoliths over longer time periods is poorly understood.  One model, which fit experimental results over 5 months quite well, predicted that even after 1000 years, 99% of phytoliths would be found no more than 5cm below the layer in which they were initially deposited (Fishkis et al., 2009).    The index I developed for distinguishing Douglas-fir forests from savannah vegetation is a ratio between a very large phytolith (the asterosclereid, approximately 50-200um in diameter) and a very small phytolith (the rondel, c. 10-20um in diameter, Fig. 6.3).  Therefore, differential movement of these phytoliths in the soil profile based on size could cause shifts in this ratio independent of vegetation change.  However, if this were the cause of changes in the ratio, I would not expect to be able to find patterns such as those observed in COW1 and COW14 cores, in which a high ln(A:R) ratio is maintained with depth (Fig. 6.12, 6.13).    166Despite the challenges, phytolith assemblages from soil profiles can provide a useful stratigraphic record of vegetation change over time.  McNamee (2013) found that even at sites with non-aggrading or erosional soil profiles, the phytolith record of known historical vegetation changes was not completely obscured.  Alexandre et al. (1999) provide evidence that the movement of phytoliths downward in soil can be likened to the movement of soil organic matter.  Although a certain proportion of phytoliths is mobile in the soil, a stable pool of phytoliths is maintained such that the average age of phytoliths increases systematically with depth.  Therefore, as long as strong vertical mixing and soil inversion due to disturbances like ploughing or landslides can be ruled out, the phytolith record in soils can provide good evidence of past vegetation change.  Once phytolith evidence of a vegetation shift is found, there is still the difficulty of dating the shift given the potentially uneven rate of soil formation over time.  Some researchers have avoided this issue by simply assigning the top 2cm or so of soil as ?modern?, and samples from depths below that as ?pre-modern? (e.g. Bartolome et al., 1986; Evett et al., 2007; Kerns et al., 2001).  This works well in situations where the goal is to ascertain whether or not a shift in vegetation has occurred, not when it occurred.  The alternative is to radiocarbon date bulk soil organic material from the same layer as phytoliths, small charcoal pieces or macrofossils from the same layer, or the phytoliths themselves.  Phytoliths often have carbon occluded within them, and this carbon can be dated if enough pure extract can be obtained (Fredlund, 2005).  However, it is difficult to obtain such a pure sample, and recent research yielded inexplicably old radiocarbon dates from recently formed phytoliths (Santos et al., 2010).  Bulk soil organic matter gives an average date for all the carbon in a soil layer, and is considered a minimum age due to   167the continued movement of younger carbon down in the soil profile (Kerns et al., 2001).  Radiocarbon dating of individual macroscopic charcoal gives a single date, and this was the technique I chose for this study.  There is some evidence that phytoliths can move downwards in soil more quickly than charcoal due to their smaller size (Alexandre et al., 1999), in which case phytoliths may be younger on average than macroscopic charcoal in the same soil layer.  For this reason, I consider my radiocarbon dates to be the earliest possible timeframe for observed changes to have occurred.  I did not attempt to build age-depth relationships given my lack of knowledge about the rate and consistency of soil formation at my study sites.  The phytolith record preserved in terrestrial soils is best thought of as a palimpsest (sensu Targulian and Goryachkin, 2004), in which the continuous input of phytoliths to the soil surface combines with organic matter accumulation, weathering of the parent material, translocation of phytoliths and other materials down in the soil layer, bioturbation and other soil-forming processes to create the observed phytolith pattern.  This gradual but continuous process of phytolith incorporation into the soil profile, called ?inheritance? by some phytolith researchers (Fredlund and Tieszen, 1994), can document vegetation shifts, as long as major erosional events and deep soil mixing can be ruled out.  For example, a high concentration of grass-produced phytoliths throughout a soil profile indicates that grass species have formed a significant proportion of the vegetation on the site for a considerable length of time (e.g. Evett et al., 2007).  I propose that shifts in the relative amounts of rondels and asterosclereids with depth in soils formed on similar parent materials in the same climate, as in my study region, can indicate shifts in the relative predominance of grass versus Douglas-fir on a site throughout the formation of the soil profile.   168 The key to successfully using the soil phytolith record to interpret past vegetation changes is to compare changes in the phytolith record with independent lines of evidence for vegetation shifts from historical or other paleoecological data.  This way, the sensitivity of the phytolith record to known vegetation changes, and the reliability of various methods for dating these changes, can be tested and confirmed before interpretations are made at other sites.  This strategy is still relatively new, but it has already been successfully used by integrating phytolith studies with legacy data and aerial photographs (McNamee, 2013), land survey records (Evett et al., 2012), written and oral records of vegetation change and fire occurrence (Morris et al., 2009, 2010a,b), palynological data (Alexandre et al., 1999; Okunaka et al., 2012; Piperno, 1985), and data from long-term experiments (Blinnikov et al., 2013).   My study provides further support for the usefulness of this technique.  6.5.5 Outliers in the surface calibration The ln(A:R) ratio of phytoliths in surface soils is quite a robust indication of current vegetation type (Fig. 6.7).  However, there were three surface samples for which the ln(A:R) ratio did not match expectations based on the current proportions of grass and Douglas-fir.  I believe these mismatches are a result of the long-term incorporation of phytoliths into the surface soil assemblage (Fredlund and Tieszen, 1994).  In the case of plot COW4, the ln(A:R) ratio placed it decidedly in the range of Douglas-fir dominated vegetation, although the current canopy is dominated by bigleaf maple, with only 12% cover of Douglas-fir.  Given the prevalence of bigleaf maple on mesic sites following logging or other disturbance of the Douglas-fir canopy, and the fact that the 1859 description is of ?woodland?, with the bearing tree being identified as a   169Douglas-fir, I suspect that at least 150 years of Douglas-fir prevalence at the site has maintained a level of asterosclereids:rondels in the range of Douglas-fir forest, although Douglas-fir is not currently dominant.    Plot COW12 has 55% cover Douglas-fir, very little grass, and no Garry oaks, and yet its ln(A:R) ratio is in the ?transition? range, similar to plot SOM2.  This plot occurs in an area described as ?good open land? in 1859, and given its similarity to SOM2 in terms of Douglas-fir and grass cover, it is most likely a ?transition? plot in which Douglas-fir has only increased recently, and should have been classified as such.    Finally, one of my surface soil extractions for plot 57 yielded relatively low asterosclereid counts, in spite of its dominance by Douglas-firs.  It had a similar rondel concentration to the consistently forested site COW1, therefore the lower ln(A:R) ratio is caused by a lower asterosclereid concentration rather than a higher rondel concentration.  It is possible that this Douglas-fir vegetation is a relatively recent formation on this site, and therefore asterosclereids have not had time to build up in the soil to the same extent as other sites.  In any case, these three plots emphasize that the ln(A:R) ratio of phytoliths in the top two centimetres of soil is not always an accurate indicator of instantaneous current vegetation composition.  6.5.6 A possible index for dating Douglas-fir infilling In light of my findings, I believe it may be possible to define an index for sites currently dominated by Douglas-fir forest that would indicate how long they have existed in that state, based on the concentration of asterosclereids in the top 2cm of soil.  I noticed that the   170concentration, much like the proportion of asterosclereids (Fig. 6.15), seems to vary with the length of time Douglas-fir has been estimated to dominate based on the ln(A:R) ratio.  That is, the highest concentrations are found in long-term Douglas-fir forests, followed by those that began to see an increase in Douglas-fir some centuries prior to European settlement (e.g. SOM3), followed by those that did not begin to infill with Douglas-fir until just before or during European settlement (e.g. SOM2).  This makes sense according to the ?inheritance? model of soil phytolith assemblage formation (Fredlunch and Tieszen, 1994), whereby sites that have been forested with Douglas-fir for longer periods of time have had longer to build up high concentrations of asterosclereids.  If we assume that COW1 has been a Douglas-fir forest for about 3000 years (the age of the oldest radiocarbon date), then we could construct a curve as shown in Figure 6.16.  Ideally, more full cores in a range of current and historical vegetation types would be analyzed, starting with those I collected, in order to confirm that this relationship holds.  I could then estimate the time since afforestation based on the asterosclereid concentration.  For example, I would predict that COW2 and COW7 have had significant cover of Douglas-fir for approximately 1500 years (Fig. 6.16).    171 Figure 6.16:  Potential relationship between the concentration of asterosclereids in surface soil phytolith assemblages, and the date of encroachment by Douglas-fir.  Solid circles represent cores with which I constructed the curve, dotted lines project the asterosclereid concentration for other sites onto the curve.  Once this relationship was established, it would be relatively easy to estimate times of afforestation for other sites, since surface soil samples are easy to collect, and asterosclereids are easy to count due to their large size.  Future research could then test for relationships between the date of afforestation and landscape factors, such as soil depth, elevation, slope, and proximity to areas of high human use.  This would help us to understand how and why different areas on the landscape were filled in by Douglas-fir at different times.    1726.6 Conclusion The phytolith record appears to be a useful potential indicator of local vegetation shifts that occurred prior to the first written records in this region.  The historical range of variation of the plant communities of southeastern Vancouver Island seems to differ depending on the specific location.  Some of my sites showed relative stability, maintaining open, grassy vegetation or Douglas-fir forest continuously for at least 2000 years.  Other sites have shifted from more open vegetation to a greater presence by Douglas-fir; but this seems to have happened at different times for different sites.  In some areas, infilling may have begun around the time of the onset of cooler, wetter Little Ice Age period, which likely reduced fire frequency and allowed more Douglas-fir seedlings to survive.  In other areas, infilling did not begin until just before or during the time of European settlement, and remnant old oak trees can still be found in the midst of dense Douglas-fir stands.  This infilling may have been climatically driven, a result of changes in human activities, or a combination.  More research is needed to confirm that some sites saw an increase in Douglas-fir sooner than others, and to disentangle which site factors, be they edaphic, topographic, or cultural, predispose a site to Douglas-fir colonization.  There is still mystery surrounding the nature of the ?pine plains? so common on the landscape at the time of the first land surveys.  Further examination of soil phytoliths from sites of former ?pine plains? is needed to determine the range of the ln(A:R) ratio possible in these areas.  We also need to find sites that approximate this vegetation type now, in order to understand what its phytolith assemblage might look like.  The phytolith record suggests that the landscape of southeastern Vancouver Island was even more open in the centuries before the first European settlers arrived than it was at the time of the first surveys.  If we want to preserve the species   173adapted to these open conditions, active management to maintain lower forest densities will be necessary.   174Chapter  7: Conclusion  Landscape architect Anne Whiston Spirn (1996, p.110) wrote:    ?Our short individual and collective memories present a major human conundrum.  How can human communities manage landscape change that takes place over a hundred years or more, when people?s perceptions and priorities change from generation to generation, or even from election to election? [?] Humans may not have the right ?attention span? to manage environmental change, and this may be the species?s fatal flaw.?   I was reminded of this as a teaching assistant in the third year ecology course at UBC.   My students had some difficulty with one laboratory report, in which they were asked to describe the likely trajectory of succession based on the measured sizes of different tree species in a forest.  They found this a challenge even though I had pointed out to them during their sampling that the understorey beneath the big old Douglas-firs was dominated by a layer of young hemlock and red cedar saplings, but young Douglas-firs were practically absent.  Our short attention span certainly is a challenge when it comes to understanding the structure and function of plant communities.  If it is hard to grasp one hundred years of forest succession, it is even more difficult to fathom that the massive old growth Douglas-fir forests of the Pacific Northwest as seen by the first European explorers were actually relatively ?recent? formations, having existed for only five to ten Douglas-fir life spans (Sprugel, 1991).    175The idea that the long-term history of ecological communities is important is not a new one (e.g. Cowles, 1901), but over the past few decades there has been a resurgence of interest in long-term ecology (e.g. Dearing et al., 2006; Foster et al., 1990; Landres et al., 1999; Swetnam, 1999).  Not only do plant community dynamics occur over relatively long timeframes, but past disturbances and the resulting changes in plant communities can leave lasting legacies that influence plant community structure in the present (Foster et al., 2003).  It is this contingency on past events that requires an understanding of the past in order to understand the present, and hopefully, to predict the future.  The challenge for ecologists is that long-term research often requires the use of unconventional data sources, and the integration of information from discplines outside traditional ecology, like history, paleoecology, and archaeology (McCune et al., 2013b; Vellend et al., 2013).  The goal of my research was to contribute to the understanding of the long-term history of plant communities on southeastern Vancouver Island.  I aimed to integrate studies at two temporal scales to fill the knowledge gaps in the history of this system.  I asked the following broad questions:  (1) What have been the consequences of recent urban expansion on understorey plant community diversity and composition?  (2) Are these consequences predictable based on changing landscape configuration and land use, and plant life history traits?  (3) Can phytoliths be used to provide a local-scale indication of changes in plant communities prior to written records?  (4) If so, what was the historical range of variability in forest density at local sites before the first written records, and did changes coincide with particular climatic or cultural changes?  To address questions (1) and (2), I used legacy data from a regional vegetation study of the 1960s which provided detailed plant community data at a fine spatial scale.  To address   176questions (3) and (4), I used phytolith analysis, which provided a coarser indication of plant community composition, but at the same local spatial scale.  These are diverse methods in terms of techniques, but they illuminate different parts of the same story.  7.1 Chapters 3 and 4:  the invisible present In Chapter 3, I quantified changes in plant community diversity and composition during a period in which the human population of the area doubled, with a resulting surge of suburban development.  Ecologists working in the area have generally observed an increase in exotic species over time, but there is little quantitative data on local changes in exotic species abundance.  I predicted that exotic species would have increased over forty years, with a concomitant decline in native species.  I also predicted that the spread of exotic species would drive a reduction in the variation in plant community composition between plots across the landscape.  This reduction in beta diversity ? called biotic homogenization ? is often associated with the spread of exotic species (McKinney and Lockwood, 1999).    Surprisingly, I found that exotic and native species richness at the plot-scale both increased, and that plot-level changes in exotics and natives were positively correlated.  While exotics were more likely than natives to have increased over the four decade period, gains in exotics were not significantly associated with homogenization of plant communities.  In fact, there was a tendency for gains in native species to promote increased community similarity.  This pattern is likely a result of the higher average frequency of natives in 1968, which meant that a gain in a native species was more likely to be a gain of a species already present in many other plots.    177Therefore, colonizations by natives were more likely than colonizations by exotics to promote biotic homogenization.  I also discovered that the finding of biotic homogenization was conditional on the length of the plant community gradient I sampled.  When I considered vegetation types separately, only Douglas-fir forests showed significant homogenization, while Garry oak communities did not.  Many of the Garry oak sites I sampled were already highly isolated and fragmented at the time of the 1968 survey, whereas Douglas-fir plots tended to occur in more intact locations.  This difference in starting conditions resulted in a much greater relative increase in understorey species for Douglas-fir plots, as disturbance tolerant colonizers moved in following fragmentation, leading to biotic homogenization. It is already well known that beta diversity is dependent on the spatial and temporal scales at which it is measured (McKinney, 2004).  My findings highlight the importance of considering the length of the plant community gradient as well.  In order to determine whether species with certain traits were more likely to increase in frequency or abundance over the four decade period, I compiled data on traits hypothesized to influence the success of species faced with increased disturbance, warming climate, changing herbivore densities, or changing nutrient availability.  Species that increased significantly tended to be exotic and/or disturbance tolerant, which suggests that disturbance is the greatest driver of change.  I did not find evidence that increased nitrogen deposition, climatic warming, or changing deer populations are driving changes in plant communities.      178In Chapter 4, I tested whether plant traits mediate the relationship between apparent colonizations and extirpations based on landscape context within 500m of a plot.  Landscape context refers to the configuration of vegetation patches, and the type of surrounding land use.  I predicted that exotic, disturbance tolerant, annual, herbaceous, shade intolerant, high-nutrient loving species with higher specific leaf area would be more likely to colonize plots surrounded by a higher degree of disturbance and fragmentation.  Conversely, I expected species with these traits to be more likely to be extirpated in more intact, less disturbed areas.  I found that colonizations largely followed these predictions, but that extirpations were much less predictable.  This is a challenge for conservation, since we would like to be able to predict which species are most vulnerable to extirpation.  The lack of predictability of extirpations may be caused by a time lag of greater than 40 years in plant community response following landscape alteration.  If this is the case, we have an extinction debt (Kuussaari et al., 2009), and predict that many species will eventually be extirpated as a result of landscape changes that have already occurred.  7.2 Chapters 5 and 6:  vegetation before European settlement In Chapter 5, I compiled a reference collection of the most common plant species of non-riparian habitats in the region in order to determine which species produce phytoliths and in what proportions.  This work was necessary before phytoliths stored in soils could be used to infer past vegetation changes.  The grass family is known to be a major producer of phytoliths (Pearsall, 2000), so I was not surprised to find that that the most prolific producers of phytoliths in the region are the grasses, which produce abundant rondel and elongate phytoliths.  The two native bunchgrasses, Achnatherum lemmonnii and Danthonia californica, are the only producers   179of bilobate phytoliths in this region.  The conifer Douglas-fir produces large, distinctive asterosclereid phytoliths.  I observed distinctive phytolith types in a few other taxa, including Pinus contorta and Carex inops, but these types were extremely rare in soil samples.  I also collected two soil cores, each 20m on either side of an ecotone between a current Douglas-fir/Arbutus forest and an open, grassy Garry oak savannah.  I extracted the phytoliths from these samples to determine whether phytoliths could accurately distinguish between these two vegetation types.  Asterosclereid phytoliths were found almost exclusively in the soil from the Douglas-fir dominated forest.  However, rondels and other grass phytoliths were as abundant, and sometimes more abundant, in the forest as in the savannah.  Given the much smaller size of the grass-produced phytoliths, it is possible that they are more mobile in windy conditions.  Alternatively, it may be that the Douglas-fir forest is a relatively recent formation on a former open, grassy site.  Nevertheless, asterosclereids are an accurate indicator of the presence of Douglas-fir on a site.  In Chapter 6, I aimed to determine whether phytolith assemblages in soil cores were sensitive to known historical vegetation changes, and then to determine the relative timing of shifts in vegetation prior to European colonization at local sites.  An earlier study based on the 1859 land surveys of the Cowichan Valley established that forest density is currently at least twice as high as it was then, and that open, grassy habitats were much more common on the landscape than they are today (Bjorkman, 2008; Bjorkman and Vellend, 2010).  I extracted phytoliths from 25 soil surface samples and seven full soil cores from sites throughout the historical range of Garry oak savannahs.  Some of these cores were taken where the present and 1859 forest density had   180been estimated.  I used the soil surface samples to establish that the log ratio of asterosclereids to rondels is a reliable predictor of current vegetation type.  I then confirmed that most of the soil cores from sites known to be open ?plains? or ?prairies? according to the original land surveys, but that had subsequently been colonized by Douglas-fir forest, showed a pattern of decreasing asterosclereid:rondel ratio with depth in the soil core.    To estimate the timing of vegetation shifts, I obtained radiocarbon dates on charcoal fragments from five cores.  Some sites have been continuously open or continuously forested for at least two thousand years. Other sites have seen infilling by Douglas-fir forest, but the timing of this infilling seems to have varied depending on the site.  Some sites show evidence of infilling centuries prior to European settlement, perhaps around the time of a cooling and moistening climatic trend known as the Little Ice Age, which began around 1400AD (Mann et al., 2009).  This time period coincides with a reduction in fires in the area (e.g. Brown and Hebda, 2002; Lucas and Lacourse, 2013), which would have increased survival of Douglas-fir seedlings.  Other sites show no sign of Douglas-fir encroachment until just prior to or during European settlement.  Given the concurrence of a particularly moist period near the end of the Little Ice Age with European settlement, and the resulting changes in First Nations activities, it is difficult to say what caused these changes.  However, the phytolith record suggests that in this region prior to European settlement, the landscape supported vegetation that was even more open than was observed during the first land surveys.    1817.3 General conclusions There is a belief amongst scientists and the general public that human actions are decimating global biodiversity (e.g. Barnosky et al., 2011; Butchart et al., 2010; Hooper et al., 2012).  My research shows that the effect of human disturbance on biodiversity depends on the spatial and temporal scale at which it is examined.  On the Saanich peninsula over the past four decades, plot-level plant diversity actually increased by 38%, and the total diversity of all plots pooled increased by 32%.  In fact, this is not a unique result amongst studies using historical data to track changes in plant diversity over time.  Increases in plot-level plant species richness have been documented quite often in European forests and grasslands (Aggemyr and Cousins, 2012; Baeten et al., 2009; Naaf and Wulf, 2010; Peter et al., 2009; Thimonier et al., 1994; Van Calster et al., 2007).  These results do not preclude the existence of an extinction debt, which, when paid, may result in a net decline in species richness as a result of human disturbance.  In addition, increases in local and overall species richness may be accompanied by a decline in beta diversity, as I observed.  However, these results show that human disturbance does not automatically lead to immediate declines in local species richness.   There is also great concern that invasions by exotic species are causing the extirpation of native species and the homogenization of plant communities (e.g. McKinney and Lockwood, 1999; Vitousek et al., 1996; Wilcove et al., 1998).  The increased disturbance and fragmentation of plant communities on the Saanich peninsula over the past four decades has been accompanied by an increase in abundance and richness of exotic plant species, as predicted.  However, gains in exotic species are not correlated with losses in native species - in fact, the net change in exotics and natives per plot were positively correlated ? nor was colonization by exotic species linked to   182the decline in beta diversity.   Other studies in the region have provided evidence that the spread of exotic species is not driven by competitive superiority over natives, but a result of their greater ability to take advantage of altered disturbance regimes (Lilley and Vellend, 2009; Gonzales and Arcese, 2008).  My research adds support for this idea, showing that disturbance tolerant species, both exotic and native, have been increasing across the landscape.  Most exotics are disturbance tolerant, and therefore a higher proportion of exotics are thriving.  These surprising results do not mean that disturbance, fragmentation, and the invasion of exotic species are beneficial in terms of promoting native plant survival and plant community diversity on this landscape.  The direct effects of suburban development are undisputable:  of the 400 or so plots Hans Roemer surveyed in 1968, about 70 had been replaced by houses and pavement by 2009.  Human disturbance via recreation and habitat destruction for residential and commercial development are the top two threats listed in recovery strategies for plants listed under Canada?s Species at Risk Act (McCune et al., 2013a).  However, the long-term perspective shows that human activities are not always detrimental to all plant species in the remaining fragments, and that the response of plant communities is complex.  So far, colonizations have been remarkably predictable based on plant species traits, and the condition of the landscape surrounding a plant community.  Extirpations have been less common, but also less predictable.  It is possible that this system is facing an extinction debt, and that many more species will be extirpated over the next few decades.  In this case, it is likely that those most vulnerable will be disturbance intolerant, native, perennial species.  Continued monitoring is needed to find out if this is the case.  If so, and we wish to prevent those extirpations from occurring, we will have to find a way to reduce the threats to these disturbance-sensitive species.   183 The results of my phytolith research suggest that many of the plant communities we see on southeastern Vancouver Island today are the result of a centuries-long trajectory of infilling of much more open, savannah-like vegetation types by Douglas-fir forest.  But the onset of this infilling was likely different for different sites.  At some sites, infilling may have begun well before European settlement, perhaps around the time of the onset of the cooler, wetter Little Ice Age period, which likely reduced fire frequency and allowed more Douglas-fir seedlings to survive.  At other sites, infilling did not began until just before or during the time of European settlement.  This infilling may have been climatically driven, a result of changes in human activities, or a combination.  More research is needed to determine why some sites saw an increase in Douglas-fir sooner than others, and to disentangle which site factors, be they edaphic, topographic, or cultural, predispose a site to Douglas-fir colonization.  On the other hand, some current savannah sites have maintained open conditions for at least two thousand years.  Many of the threatened plant species of the Garry oak ecosystem are shade-intolerant species that most likely thrived in these more open conditions (Parks Canada Agency, 2006a,b).  If our goal is to maintain these species, active management to restore and maintain open conditions will be required.  7.4 Study limitations and future research The use of legacy data from past ecological studies for long-term research comes with particular challenges because thes data were not collected with the purpose of long-term monitoring in mind (Vellend et al., 2013).  First there is the issue of imprecise relocation:  Hans Roemer?s plots were not permanently marked, and therefore some of the changes observed between 1968 and   1842009 could be associated with relocation error (Fischer and Stocklin, 1997).  In addition, there can be issues with taxonomic uncertainty where species names have changed, or if a different taxonomic resolution was used by each surveyor (Vellend et al., 2013).  The original survey by Dr. Roemer was done for the purpose of phytosociological classification, and as such the plots were not randomly chosen, but carefully selected to maximize vegetation homogeneity and minimize disturbance within each plot.  This bias towards homogeneity in the original selection of plots can result in a finding of increased heterogeneity simply due to spatial variation in disturbance with time (Palmer, 1993; Ross et al., 2010).    I was able to minimize many of these limitations with the assistance of the original surveyor himself.  Dr. Roemer advised me on the criteria he used for selecting plots, and explained his system for referencing plots using the aerial photo grid system, so I could relocate plots with as much accuracy as possible.  I was also able to clarify changes in taxonomy with him.  I maximized stand homogeneity as much as possible within the relocated plots.  In addition, I was very careful not to make conclusions about particular plots, given the chance of relocation error, instead looking at patterns across the set of all plots.    The historical data from my resurvey study is limited to only two time points:  1968 and 2009.  There is no record of the speed or timing with which changes I observed occurred.  Future work to monitor these sites on a more frequent basis could help reveal how quickly species respond to changes on the landscape.  For example, it seems that there may be a lag of over four decades between disturbances and extirpations on the landscape.  Only future long-term monitoring can   185determine exactly how long these lags may be.  Now that we have more precise GPS coordinates for each plot, it will be easier for future researchers to relocate and resurvey the plots.  The ability of resurvey data to measure changes in rare species is also limited.  Out of 266 understorey taxa observed in at least one year, only three are provincially listed as imperiled or critically imperiled:  Lomatium dissectum var. dissectum, Sanicula bipinnatifida, and Sericocarpus rigidus.  These species were each found in 3 or fewer plots in each year, and so there is insufficient statistical power to determine whether they have declined significantly in abundance on the landscape.  In order to determine the dynamics of rare species, their populations must be targeted for long-term monitoring.  Indeed, this is the approach advocated in the recovery strategies for Garry oak-associated plants (Parks Canada Agency, 2006a,b).  There are a number of challenges associated with using the phytolith record in soil to document historical changes in vegetation. The incorporation of phytoliths into the soil profile is affected by the rate of soil accumulation and/or erosion, as well as shifts in the plant community over time (Fredlund and Teiszen, 1994).  In addition, terrestrial soil phytolith analyses must consider the documented movement of phytoliths downwards in the soil profile (Fishkis et al., 2010a,b).  Despite the challenges, phytolith assemblages from soil profiles can provide a useful stratigraphic record of vegetation change over time (Alexandre et al., 1999; McNamee, 2013), as long as strong vertical mixing and soil inversion due to disturbances like ploughing or landslides can be ruled out.      186There is also difficulty in dating vegetation shifts indicated by phytolith assemblages due to the potentially uneven rate of soil formation over time.  Radiocarbon dating of macroscopic charcoal fragments within the same layer as a shift in phytolith assemblages must be considered only a rough estimate of the timing of a shift in vegetation.  There is some evidence that phytoliths can move downwards in soil more quickly than charcoal due to their smaller size (Alexandre et al., 1999), in which case phytoliths may be younger on average than macroscopic charcoal in the same soil layer.  For this reason, I consider my radiocarbon dates to be a rough estimate of the earliest possible timeframe for observed changes to have occurred.    The results of my phytolith research suggest that there is good potential for using the concentration of asterosclereids in surface soils as an index of the relative date of forest infilling at local sites across the landscape.  Future research could confirm this by analyzing and dating a broader set of soil cores. Once this relationship was well established, it could be used to test for relationships between the date of afforestation at particular sites and landscape factors, such as soil depth, elevation, slope, and proximity to areas of high human use.  Ideally, this research would be combined with archaeological research at inland sites, to find out more about the extent of human influence at particular locations on the landscape prior to European settlement.  This would help to disentangle which site factors make an open savannah site susceptible to infilling by Douglas-fir.  Delving into the long-term history of a plant community provides fresh insight into the complex processes that have produced the patterns we see today.  The plant communities of southeastern Vancouver Island are a product of many centuries of gradual infilling and thickening of Douglas-  187fir forests, as well as accelerated fragmentation, disturbance, and colonization by introduced species in the last few decades.  The response of local plant communities to these changes is sometimes predictable, but often unexpected.  Long-term research can help us to extend our short attention span, and thus to understand how best to manage plant communities in order to maintain the characteristics we value.         188References Aggemyr, E. and S. A. O. Cousins. 2012. Landscape structure and land use history influence changes in island plant composition after 100 years. Journal of Biogeography 39:1645-1656. Agresti, A. 2002. Categorical data analysis. Wiley, New York, USA. Albert, R. M., O. Lavi, L. Estroff, and S. Weiner. 1999. Mode of occupation of Tabun cave, Mt Carmel, Israel during the Mousterian period:  a study of the sediments and phytoliths. Journal of Archaeological Science 26:1249-1260. Alexander, H. M., B. L. Foster, F. Ballantyne, C. D. Collins, J. Antonovic, and R. D. Holt. 2012. Metapopulations and metacommunities:  combining spatial and temporal perspectives in plant ecology. Journal of Ecology 100:88-103. Alexandre, A., J. D. Meunier, A. Mariotti, and F. Soubies. 1999. Late holocene phytolith and carbon-isotope record from a latosol at Salitre, South-Central Brazil. Quaternary Research 51:187-194. Allison, S.K. 2004. What do we mean when we talk about ecological restoration? Ecological Restoration 22:281-286. Al-Talib, K. H. and J. G. Torrey. 1961. Sclereid distribution in the leaves of Pseudotsuga under natural and experimental conditions. American Journal of Botany 48:71-79. Ames, K. M. and H. D. G. Maschner. 1999. Peoples of the Northwest Coast:  their archaeology and prehistory. Thames and Hudson, London, UK. Anderson, D.R. 2008. Model Based Inference in the Life Sciences:  A Primer on Evidence. Springer, New York, USA.   189Anderson, M. J. 2001a. A new method for non-parametric multivariate analysis of variance. Austral Ecology 26:32-46. Anderson, M. J. 2001b. Permutation tests for univariate or multivariate analysis of variance and regression.  Canadian Journal of Fisheries and Aquatic Science 58:626-639. Anderson, M. J., K. E. Ellingsen, and B. H. McArdle. 2006. Multivariate dispersion as a measure of beta diversity. Ecology Letters 9:683-693. Anderson, M.J., R.N. Gorley, and K.R. Clarke. 2008. PERMANOVA+ for PRIMER: Guide to Software and Statistical Methods. PRIMER-E, Plymouth, UK. Apple, M., K. Tiekotter, M. Snow, J. Young, A. Soeldner, D. Phillips, D. Tingey, and B. J. Bond. 2002. Needle anatomy changes with increasing tree age in Douglas-fir. Tree Physiology 22:129-136. Athens, J. S. and J. V. Ward. 1993. Environmental change and prehistoric Polynesian settlement in Hawai'i. Asian Perspectives 32:205-223. Baeten, L., B. Bauwens, A. De Schrijver, L. De Keersmaeker, H. Van Calster, K. Vandekerkhove, B. Roelandt, H. Beeckman, and K. Verheyen. 2009. Herb layers changes (1954-2000) related to the conversion of coppice-with-standards forest and soil acidification. Applied Vegetation Science 12:187-197. Bagaria, G., J. Pino, F. Roda, and M. Guardiola. 2012. Species traits weakly involved in plant responses to landscape properties in Mediterranean grasslands. Journal of Vegetation Science 23:432-442. Baiser, B., J. D. Olden, S. Record, J. L. Lockwood, and M. L. McKinney. 2012. Pattern and process of biotic homogenization in the New Pangaea. Proceedings of the Royal Society B: Biological Sciences 279:4772-4777.   190Barnosky, A. D., N. Matzke, S. Tomiya, G. O. U. Wogan, B. Swatzh, T. B. Quental, C. Marshall, J. L. McGuire, E. L. Lindsay, K. C. Maguire, B. Mersey, and E. A. Ferrer. 2011. Has the Earth's sixth mass extinction already arrived? Nature 471:51-57. Bartolome, J. W., S. E. Klukkert, and W. J. Barry. 1986. Opal phytoliths as evidence for displacement of native Californian grassland. Madrono 33:217-222. Bayliss-Smith, T., E. Hviding, and T. Whitmore. 2003. Rainforest composition and histories of human disturbance in Solomon Islands. Ambio 32:346-352. Beckwith, B.R. 2004. "The Queen Root of This Clime":  Ethnoecological investigations of blue camas (Camassia leichtlinii (Baker) Wats., C. quamash (Pursh) Greene; Liliaceae) and its landscapes on Southern Vancouver Island, British Columbia. PhD thesis. University of Victoria, Victoria, Canada. Bennett, J.R. 2012. Multi-scale comparison of native and exotic communities in the Garry oak ecosystem of British Columbia. PhD Thesis. University of British Columbia, Vancouver, Canada. Bennett, J. R., M. Vellend, P. L. Lilley, W. K. Cornwell, and P. Arcese. 2013. Abundance, rarity and invasion debt among exotic species in a patchy ecosystem. Biological Invasions 15:707-716. Berglund, B.E. 1991. The project:  background, aims, and organization. Ecological Bulletins 41:13-17. Bignal, E.M., and D.I. McCracken. 2000. The nature conservation value of European traditional farming systems. Environmental Reviews 8:149-171.   191Bjorkman, A. D. 2008. Changes in the landscape and vegetation of southeastern Vancouver Island and Saltspring Island, Canada since European settlement. MSc Thesis. University of British Columbia, Vancouver, Canada. Bjorkman, A.D., and M. Vellend. 2010. Defining historical baselines for conservation:  ecological changes since European settlement on Vancouver Island, Canada. Conservation Biology 24:1559-1568. Bjornstad, O.N. 2012. Package ?ncf?: spatial nonparametric covariance functions. R package version 1.1-4. Bjornstad, O.N., and W. Falck. 2001. Nonparametric spatial covariance functions:  estimation and testing. Environmental and Ecological Statistics 8:53-70. Blinnikov, M. S., C. M. Bagent, and P. E. Reyerson. 2013. Phytolith assemblages and opal concentrations from modern soils differentiate temperate grasslands of controlled composition on experimental plots at Cedar Creek, Minnesota. Quaternary International 287:101-113. Blinnikov, M., A. Busacca, and C. Whitlock. 2002. Reconstruction of the late Pleistocene grassland of the Columbia basin, Washington, USA, based on phytolith records in loess. Palaeogeography Palaeoclimatology Palaeoecology 177:77-101. Bobbink, R., K. Hicks, J. Galloway, T. Spranger, R. Alkemade, R., M. Ashmore, M. Bustamante, S. Cinderby, E. Davidson, F.  Dentener, B. Emmett, J.-W. Erisman, M. Fenn, F. Gilliam, A. Nordin, L. Pardo, and W. De Vries. 2010. Global assessment of nitrogen deposition effects on terrestrial plant diversity: a synthesis. Ecological Applications 20:30-59.   192Boyd, M. 2002. Identification of Anthropogenic Burning in the Paleoecological Record of the Northern Prairies: A New Approach. Annals of the Association of American Geographers 92:471-487. Bozarth, S. 1993. Biosilicate assemblages of Boreal forests and aspen parklands. Pages 95-105 in D. M. Pearsall and D. R. Piperno, editors. Current Research in Phytolith Analysis:  Applications in Archaeology and Paleoecology. MASCA Research Papers in Science and Archaeology, Philadelphia, USA. Briggs, J.M., M. Kruse, K.A. Spielmann, K. Morehouse, H. Schaafsma, and K. Schollmeyer. 2006. Why ecology needs archaeologists and archaeology needs ecologists. Frontiers in Ecology and the Environment 4:180-188. Broersma, K. 1973.  Dark soils of the Victoria area, British Columbia. M.Sc. Thesis. University of British Columbia, Vancouver, Canada. Bronk Ramsey, C. 2009. Bayesian analysis of radiocarbon dates. Radiocarbon 51:337-360. Brostr?m, G., and H. Holmberg. 2011. Package ?glmmML?:  Generalized linear models with clustering. R package version 0.82-1. Brown, D. R. 1996. Disposing of the dead:  a shell midden cemetery in British Columbia's Gulf of Georgia Region. M.A. Thesis. University of British Columbia, Vancouver, Canada. Brown, K.J., and R.J. Hebda. 2002. Ancient fires on southern Vancouver Island, British Columbia, Canada:  A change in causal mechanisms at about 2,000 ybp. Environmental Archaeology 7:1-12. Brown, K. J., A. B. Nielsen, R. J. Fitton, and R. J. Hebda. 2008. Postglacial evolution and spatial differentiation of seasonal temperate rainforest in western Canada. Holocene 18:715-727.   193Brydon, J. E., W. G. Dore, and J. S. Clark. 1963. Silicified plant asterosclereids preserved in soil. Soil Science Society Proceedings 27:476-477. Burnham, K. P., D. R. Anderson, and K. P. Huyvaert. 2011. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behavioral Ecology and Sociobiology 65:23-35. Butchart, S. H. M., M. Walpole, B. Collen, A. van Strien, J. P. W. Scharlemann, R. E. A. Almond, J. E. M. Baillie, B. Bomhard, C. Brown, J. Bruno, K. E. Carpenter, G. M. Carr, J. Chanson, A. M. Chenery, J. Csirke, N. C. Davidson, F. Dentener, M. Foster, A. Galli, J. N. Galloway, P. Genovesi, R. D. Gregory, M. Hockings, V. Kapos, J.-F. Lamarque, F. Leverington, J. Loh, M. A. McGeoch, L. McRae, A. Minasyan, M. H. n. Morcillo, T. E. E. Oldfield, D. Pauly, S. Quader, C. Revenga, J. R. Sauer, B. Skolnik, D. Spear, D. Stanwell-Smith, S. N. Stuart, A. Symes, M. Tierney, T. D. Tyrrell, J.-C. Vi?, and R. Watson. 2010. Global Biodiversity: Indicators of Recent Declines. Science 328:1164-1168. Cavers, M. 2009. "Victoria's own oak tree":  a brief cultural history of Victoria's Garry oaks after 1843. BC Studies Autumn:65-85. Christensen, R.H.B. 2013. Ordinal: regression models for ordinal data. R package version 2012.09-11. Clarke, K. R., and R.N. Gorley. 2006. Primer v6 Permanova+. Primer-E Ltd., Plymouth, UK. Cocking, M. I., J. M. Varner, and R. L. Sherriff. 2012. California black oak responses to fire severity and native conifer encroachment in the Klamath Mountains. Forest Ecology and Management 270:25-34.   194Coop, J. D. and T. J. Givnish. 2007. Spatial and temporal patterns of recent forest encroachment in montane grasslands of the Valles Caldera, New Mexico, USA. Journal of Biogeography 34:914-927. Cordeiro, N.J., and H.F. Howe. 2001. Low recruitment of trees dispersed by animals in African forest fragments. Conservation Biology 15:1733-1741. Cousins, S. A. O. and E. Aggemyr. 2008. The influence of field shape, area and surrounding landscape on plant species richness in grazed ex-fields. Biological Conservation 141:126-135. Cousins, S. A. O., H. Ohlson, and O. Eriksson. 2007. Effects of historical and present fragmentation on plant species diversity in semi-natural grasslands in Swedish rural landscapes. Landscape Ecology 22:723-730. Cowan, I. M. 1945. The ecological relationships of the food of the columbian black-tailed deer, Odocoileus hemionus columbianus (Richardson), in the coast forest region of southern Vancouver Island, British Columbia. Ecological Monographs 15:109-139. Cowles, H. C. 1901. The physiographic ecology of Chicago and vicinity; a study of the origin, development, and classification of plant societies. Botanical Gazette 31:73-108. Cowles, H. C. 1911. The causes of vegetative cycles - contributions from the Hull Botanical Laboratory 143. Botanical Gazette 51:0161-0183. Damschen, E. I., S. Harrison, and J. B. Grace. 2010. Climate change effects on an endemic-rich edaphic flora:  resurveying Robert H. Whittaker's Siskiyou sites (Oregon, USA). Ecology 91:3609-3619. Davis, M. B. 1994. Ecology and paleoecology begin to merge. Trends in Ecology & Evolution 9:357-358.   195Day, J.H., L. Farstad, and D.G. Laird. 1959. Soil survey of southeast Vancouver Island and Gulf Islands, British Columbia. Research Branch, Canada Department of Agriculture, Ottawa, Canada. Dearing, J.A., R.W. Battarbee, R. Dikau, I. Larocque, and F. Oldfield, F., 2006. Human-environment interactions:  learning from the past. Regional Environmental Change 6:1-16. Debinski, D. M. and R. D. Holt. 2000. A survey and overview of habitat fragmentation experiments. Conservation Biology 14:342-355. de Blois, S., G. Domon, and A. Bouchard. 2001. Environmental, historical, and contextual determinants of vegetation cover:  a landscape perspective. Landscape Ecology 16:421-436. de Blois, S., G. Domon, and A. Bouchard. 2002. Landscape issues in plant ecology. Ecography 25:244-256. Deevey, E. S. 1969. Coaxing history to conduct experiments. Bioscience 19:40-43. Delcourt, P.A., and H.R. Delcourt. 2004. Prehistoric Native Americans and Ecological Change:  human ecosystems in Eastern North America since the Pleistocene. Cambridge University Press, Cambridge. Delhon, C., A. Alexandre, J. F. Berger, S. Thiebault, J. L. Brochier, and J. D. Meunier. 2003. Phytolith assemblages as a promising tool for reconstructing Mediterranean Holocene vegetation. Quaternary Research 59:48-60. Denevan, W.M. 1992. The pristine myth:  the landscape of the Americas in 1492. Annals of the Association of American Geographers 82:369-385.   196Deur, D. 2002. Rethinking precolonial plant cultivation on the Northwest Coast of North America. Professional Geographer 54:140-157. Deur, D., and N.J. Turner. 2005. Introduction: reconstructing indigenous resource management, reconstructing the history of an idea. Pages 3-34 in D. Deur and N.J. Turner, editors. Keeping it Living, Traditions of Plant Use and Cultivation on the Northwest Coast of North America. University of Washington Press, Seattle, USA. Digiovinazzo, P., G. F. Ficetola, L. Bottoni, C. Andreis, and E. Padoa-Schioppa. 2010. Ecological thresholds in herb communities for the management of suburban fragmented forests. Forest Ecology and Management 259:343-349. Douglas, G.W., G.B. Straley, D. Meidinger, and J. Pojar. 1998. Illustrated flora of British Columbia. Ministry of Environment, Lands and Parks and Ministry of Forests, Victoria, BC, Canada. Duff, W. 1997. The Indian History of British Columbia:  The Impact of the White Man. Royal British Columbia Museum, Victoria, Canada. Dufr?ne, M., and P. Legendre. 1997. Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecological Monographs 67:345-366. Duncan, R. P., S. E. Clemants, R. T. Corlett, A. K. Hahs, M. A. McCarthy, M. J. McDonnell, M. W. Schwartz, K. Thompson, P. A. Vesk, and N. S. G. Williams. 2011. Plant traits and extinction in urban areas:  a meta-analysis of 11 cities. Global Ecology and Biogeography 20:509-519. Dunwiddie, P.W. 2005. Using historical data in ecological restoration:  a case study from Nantucket. Pages 367-390 in D. Egan and E. Howell, editors. The Historical Ecology Handbook. Island Press, Washington, USA.   197Dunwiddie, P.W., J.D. Bakker, M. Almaguer-Bay, and C.B. Sprenger. 2011. Environmental history of a Garry oak/Douglas-fir woodland on Waldron Island, Washington.  Northwest Science 85:130-140. Dupr?, C. and J. Ehrl?n. 2002. Habitat configuration, species traits and plant distributions. Journal of Ecology 90:796-805. Egan, B. 1999. The ecology of the coastal Douglas-fir zone. Ministry of Forests, British Columbia, Canada. Egan, D. and E. Howell. 2005. Introduction. Pages 1-23 in D. Egan and E. Howell, editors. The Historical Ecology Handbook. Island Press, Washington, DC. Elbaum, R., S. Weiner, R. M. Albert, and M. Elbaum. 2003. Detection of burning of plant materials in the archaeological record by changes in the refractive indices of siliceous phytoliths. Journal of Archaeological Science 30:217-226. Erickson, W.R. 2008. Results and data from an ecological study of Garry oak (Quercus garryana) ecosystems in Southwestern British Columbia. B.C. Ministry of Forests and Range, Research Branch. Technical Report 043. Victoria. http://www.for.gov.bc.ca/hfd/pubs/Docs/Tr/Tr043.htm Eriksson, O. 1996. Regional dynamics of plants: a review of evidence for remnant, source-sink and metapopulations. Oikos 77:248?258. Eriksson, O. and A. Jakobsson. 1998. Abundance, distribution and life histories of grassland plants: a comparative study of 81 species. Journal of Ecology 86:922-933. Evett, R. R., A. Dawson, and J. W. Bartolome. 2012. Estimating vegetation reference conditions by combining historical source analysis and soil phytolith analysis at Pepperwood Preserve, Northern California Coast Ranges, U.S.A. Restoration Ecology 21:464-473.   198Evett, R. R., E. Franco-Vizcaino, and S. L. Stephens. 2007. Phytolith evidence for the absence of a prehistoric grass understory in a Jeffrey pine - mixed conifer forest in the Sierra San Pedro Martir, Mexico. Canadian Journal of Forest Research 37:306-317. Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology and Systematics 34:487-515. Flynn, S. 1999. Ecosystems in British Columbia at risk: coastal Douglas-fir ecosystems. Ministry of Environment, Lands and Parks, Victoria, BC, Canada. Fischer, M. and J. Stocklin. 1997. Local extinctions of plants in remnants of extensively used calcareous grasslands 1950-1985. Conservation Biology 11:727-737. Fisher, R. F., C. N. Bourn, and W. F. Fisher. 1995. Opal phytoliths as an indicator of the floristics of prehistoric grasslands. Geoderma 68:243-255. Fishkis, O., J. Ingwersen, and T. Streck. 2009. Phytolith transport in sandy sediment:  experiments and modeling. Geoderma 151:168-178. Fishkis, O., J. Ingwersen, M. Lamers, D. Denysenko, and T. Streck. 2010a. Phytolith transport in soil:  a field study using fluorescent labelling. Geoderma 157:27-36. Fishkis, O., J. Ingwersen, M. Lamers, D. Denysenko, and T. Streck. 2010b. Phytolith transport in soil:  a laboratory study on intact soil cores. European Journal of Soil Science 61:445-455. Flynn, S. 1999. Ecosystems in British Columbia at risk: coastal Douglas-fir ecosystems. Ministry of Environment, Lands and Parks, British Columbia, Canada. Foster, D. R., G. Motzkin, and B. Slater. 1998. Land-use history as long-term broad-scale disturbance:  regional forest dynamics in central New England. Ecosystems 1:96-119.   199Foster, D.R., P.K. Schoonmaker and S.T.A. Pickett. 1990. Insights from paleoecology to community ecology. Trends in Ecology and Evolution 5: 119-122. Foster, D., F. Swanson, J. Aber, I. Burke, N. Brokaw, D. Tilman, and A. Knapp. 2003. The importance of land-use legacies to ecology and conservation. Bioscience 53:77-88. Franceschi, V. R. and P. A. Nakata. 2005. Calcium oxalate in plants:  formation and function. Annual Review of Plant Biology 56:41-71. Freckleton, R. P., P. H. Harvey, and M. Pagel. 2002. Phylogenetic analysis and comparative data:  a test and review of evidence. The American Naturalist 160:712-726. Freckleton, R. P. and A. R. Watkinson. 2002. Large-scale spatial dynamics of plants:  metapopulations, regional ensembles and patchy populations. Journal of Ecology 90:419-434. Fredlund, G. G. 2005. Inferring vegetation history from phytoliths. Pages 335-362 in D. Egan and E. Howell, editors. The Historical Ecology Handbook. Island Press, Washington, DC. Fredlund, G. G. and L. T. Tieszen. 1994. Modern Phytolith Assemblages from the North American Great Plains. Journal of Biogeography 21:321-335. Fuchs, M.A. 2001. Towards a recovery strategy for Garry oak and associated ecosystems in Canada:  ecological assessment and literature review. Technical Report GBEI/EC-00-030.  Environment Canada, Canadian Wildlife Service, Pacific and Yukon Region. Fuller, D. 2007. Archaeobotany:  Phytolith teaching and research images. Old World Reference Phytoliths version 1.3. University College London, London. http://www.homepages.ucl.ac.uk/~tcrndfu/phytoliths.html#poo   200Gedalof, Z., M. Pellatt, and D.J. Smith, D.J. 2006. From prairie to forest: Three centuries of environmental change at Rocky Point, Vancouver Island, British Columbia. Northwest Science 80: 34-46. Gergel, S. E., M. G. Turner, J. R. Miller, J. M. Melack, and E. H. Stanley. 2002. Landscape indicators of human impacts to riverine systems. Aquatic Sciences 64:118-128. GOERT, 2011. What are Garry oak and associated ecosystems? Garry Oak Ecosystem Recovery Team.  http://www.goert.ca/about/index.php. Accessed 22 March 2012. Gonzales, E.K., and P. Arcese. 2008. Herbivory more limiting than competition on early and established native plants in an invaded meadow. Ecology 89:3282-3289. Grier, C., P. Dolan, K. Derr, and E. McLay. 2009. Assessing sea level changes in the southern Gulf Islands of British Columbia using archaeological data from coastal spit locations. Canadian Journal of Archaeology 33: 254-280. Hanski, I. 1999. Metapopulation ecology. Oxford University Press, Oxford, UK. Harris, C. 1994. Voices of disaster:  smallpox around the Strait of Georgia in 1782. Ethnohistory 41:591-626. Hart, D. M. and G. S. Humphreys. 2003. Phytolith depth functions in surface regolith materials. Pages 159-163 in I. C. Roach, editor. Advances in Regolith. Cooperative Research. Centre for Landscape Environments and Mineral Exploration, Australia. Hebda, R.J. 1995. British Columbia vegetation and climate history with focus on 6 ka BP. G?ographie Physique et Quaternaire 49:55-79. Heegaard, E., R. H. Okland, H. Bratli, W. E. Dramstad, G. Engan, O. Pedersen, and H. Solstad. 2007. Regularity of species richness relationships to patch size and shape. Ecography 30:589-597.   201Helm, A., I. Hanski, and M. P?rtel. 2006. Slow response of plant species richness to habitat loss and fragmentation. Ecology Letters 9:72-77. Heusser, L.E.1983. Palynology and paleoecology of post-glacial sediments in an anoxic basin, Saanich Inlet, British Columbia. Canadian Journal of Earth Science 20:873-885. Heyerdahl, E. K., R. F. Miller, and R. A. Parsons. 2006. History of fire and Douglas-fir establishment in a savanna and sagebrush-grassland mosaic, southwestern Montana, USA. Forest Ecology and Management 230:107-118. Higgins, S. I., R. Nathan, and M. L. Cain. 2003. Are long-distance dispersal events in plants usually caused by nonstandard means of dispersal? Ecology 84:1945-1956. Hobbie, J. E., S. R. Carpenter, N. B. Grimm, J. R. Gosz, and T. R. Seastedt. 2003. The US Long Term Ecological Research Program. Bioscience 53:21-32. Hooper, D. U., E. C. Adair, B. J. Cardinale, J. E. K. Byrnes, B. A. Hungate, K. L. Matulich, A. Gonzalez, J. E. Duffy, L. Gamfeldt and M. I. O?Connor. 2012. A global synthesis reveals biodiversity loss as a major driver of ecosystem change. Nature 486: 105-108. Jackson, S. T. 2006. Vegetation, environment, and time: the origination and termination of ecosystems. Journal of Vegetation Science 17:549-557. Jackson, S. T. and D. F. Sax. 2010. Balancing biodiversity in a changing environment: extinction debt, immigration credit and species turnover. Trends in Ecology & Evolution 25:153-160. Jacobson, G.L., and R.H.W. Bradshaw. 1981.  The selection of sites for paleovegetation studies.  Quaternary Research 16:80-96.   202Jacquemyn, H., J. Butaye, and M. Hermy. 2001. Forest plant species richness in small fragmented mixed deciduous forest patches:  the role of area, time and dispersal limitation. Journal of Biogeography 28:801-812. Jacquemyn, H., J. Butaye, and M. Hermy. 2003. Influence of environmental and spatial variables on regional distribution of forest plant species in a fragmented and changing landscape. Ecography 26:768-776. Jones, R. L. and A. H. Beavers. 1964. Aspects of catenary and depth distribution of opal phytoliths in Illinois soils. Soil Science Society of America Proceedings 28:413-416. Jungen, J. R. 1985. Soils of Southern Vancouver Island. Ministry of Environment, Victoria, BC, Canada. Keith, S. A., A. C. Newton, M. D. Morecroft, C. E. Bealey, and J. M. Bullock. 2009. Taxonomic homogenization of woodland plant communities over 70 years. Proceedings of the Royal Society B: Biological Sciences 276:3539-3544. Kenney, E. A., L. J. P. van Vliet, and A. J. Green. 1988. Soils of the Gulf Islands of British Columbia:  Volume 2, Soils of North Pender, South Pender, Prevost, Mayne, Saturna, and lesser islands. Agriculture Canada, Vancouver, Canada. Kerns, B. K., M. M. Moore, and S. C. Hart. 2001. Estimating forest-grassland dynamics using soil phytolith assemblages and delta C-13 of soil organic matter. Ecoscience 8:478-488. Kerr, J.T., and J. Cihlar, J. 2004. Patterns and causes of species endangerment in Canada.  Ecological Applications 14:743-753. Kirch, P. V. 2005. Archaeology and global change: the Holocene record. Annual Review of Environment and Resources 30:409-440. Klein, R. L. and J. W. Geis. 1978. Biogenic silica in the Pinaceae. Soil Science 126:145-156.   203Klinkenberg, B. 2013. E-Flora BC: Electronic Atlas of the Plants of British Columbia. Lab for Advanced Spatial Analysis, Department of Geography, University of British Columbia, Vancouver, Canada. eflora.bc.ca Knapp, S., I. Kuhn, R. Wittig, W. A. Ozinga, P. Poschlod, and S. Klotz. 2008. Urbanization causes shifts in species' trait state frequencies. Preslia 80:375-388. Kolb, A. and M. Diekmann. 2005. Effects of Life-History Traits on Responses of Plant Species to Forest Fragmentation. Conservation Biology 19:929-938. Kuussaari, M., R. Bommarco, R. K. Heikkinen, A. Helm, J. Krauss, R. Lindborg, E. Ockinger, M. Partel, J. Pino, F. Roda, C. Stefanescu, T. Teder, M. Zobel, and I. Steffan-Dewenter. 2009. Extinction debt:  a challenge for biodiversity conservation. Trends in Ecology & Evolution 24:564-571. Landres, P.B., P. Morgan, and F.J. Swanson. 1999. Overview of the use of natural variability concepts in managing ecological systems. Ecological Applications 9:1179-1188. Lea, T. 2006. Historical Garry oak ecosystems of Vancouver Island, British Columbia, pre-European contact to the present. Davidsonia 17:34-50. Lentfer, C. and R. Torrence. 2007. Holocene volcanic activity, vegetation succession, and ancient human land use: unravelling the interactions on Garua Island, Papua New Guinea. Review of Palaeobotany and Palynology 143:83-105. Lepofsky, D. and K. Lertzman. 2008. Documenting ancient plant management in the northwest of North America. Botany 86:129-145. Lepofsky, D., K. Lertzman, D. Hallett, and R.W. Mathewes. 2005. Climate change and culture change on the southern coast of British Columbia 2400-1200 B.P.:  An hypothesis. American Antiquity 70:267-293.   204Lilley, P.L., and M. Vellend, M. 2009. Negative native-exotic diversity relationship in oak savannas explained by human influence and climate. Oikos 118:1373-1382. Lindborg, R., and O. Eriksson. 2004. Historical landscape connectivity affects present plant species diversity. Ecology 85:1840-1845. Lindborg, R. 2007. Evaluating the distribution of plant life-history traits in relation to current and historical landscape configurations. Journal of Ecology 95:555-564. Lindenmayer, D. B. and J. Fischer. 2006. Habitat Fragmentation and Landscape Change:  An Ecological and Conservation Synthesis. Island Press, Washington, USA. Lucas, J.D., and T. Lacourse. 2013. Holocene vegetation history and fire regimes of Pseudotsuga menziesii forests in the Gulf Islands National Park Reserve, southwestern British Columbia, Canada.  Quaternary Research 79:366-376. Lunt, I. D. and P. G. Spooner. 2005. Using historical ecology to understand patterns of biodiversity in fragmented agricultural landscapes. Journal of Biogeography 32:1859-1873. MacArthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, USA. MacDougall, A.S. 2004. Joint effects of competition, recruitment limitation, and fire suppression in an invaded oak savanna ecosystem. PhD Thesis. University of British Columbia, Vancouver, Canada. MacDougall, A. S. 2005. Responses of diversity and invasibility to burning in a northern oak savanna. Ecology 86:3354-3363.   205MacDougall, A.S., B.R. Beckwith, and C.Y. Maslovat. 2004. Defining conservation strategies with historical perspectives: a case study from a degraded oak grassland ecosystem. Conservation Biology 18:455-465. MacDougall, A. S., K. S. McCann, G. Gellner, and R. Turkington. 2013. Diversity loss with persistent human disturbance increases vulnerability to ecosystem collapse. Nature 494:86-90. MacDougall, A.S., and R. Turkington. 2005. Are invasive species the drivers or passengers of change in degraded ecosystems? Ecology 86:42-55. MacDougall, A.S., and R. Turkington, R. 2006. Dispersal, competition, and shifting patterns of diversity in a degraded oak savanna. Ecology 87:1831-1843. MacDougall, A.S., and R. Turkington. 2007. Does the type of disturbance matter when restoring disturbance-dependent grasslands? Restoration Ecology 15:263-272. Madella, M., A. Alexandre, T. Ball, and I. W. Grp. 2005. International Code for Phytolith Nomenclature 1.0. Annals of Botany 96:253-260. Madella, M. and C. Lancelotti. 2012. Taphonomy and phytoliths: A user manual. Quaternary International 275:76-83. Magnuson, J.J. 1990. Long-term ecological research and the invisible present ? uncovering the processes hidden because they occur slowly or because effects lag years behind causes. Bioscience 40:495-501. Malmer, N. 1991. The Ystad project and nature conservation. Ecological Bulletins 41:454-455. Mann, M. E., Z. Zhang, S. Rutherford, R. S. Bradley, M. K. Hughes, D. Shindell, C. Ammann, G. Faluvegi, and F. Ni. 2009. Global signatures and dynamical origins of the Little Ice Age and Medieval Climate Anomaly. Science 27:1256-1260.   206Marini, L., H. H. Bruun, R. K. Heikkinen, A. Helm, O. Honnay, J. Krauss, I. Kuhn, R. Lindborg, M. Partel, and R. Bommarco. 2012. Traits related to species persistence and dispersal explain changes in plant communities subjected to habitat loss. Diversity and Distributions 18:898-908. Maslovat, C. 2002. Historical jigsaw puzzles:  piecing together the understory of Garry oak (Quercus garryana) ecosystems and the implications for restoration. USDA Forest Service Gen. Tech. Rep.:141-149. Matson, R.G., and G. Coupland. 1995. The Prehistory of the Northwest Coast. Academic Press, San Diego, USA. McArdle, B.H., and M.J. Anderson. 2001.  Fitting multivariate models to community data:  a comment on distance-based redundancy analysis.  Ecology 82: 290-297. McCann, J.M. 1999. Before 1492:  the making of the pre-Columbian landscape. Part I:  The Environment. Ecological Restoration 17:15-30. McClaran, M. P. and M. Coder. 2003. Phytolith shape and chemical composition in desert grassland plants. Journal of the Arizona-Nevada Academy of Science 36:21-33. McClaran, M. P. and M. Umlauf. 2000. Desert Grassland Dynamics Estimated from Carbon Isotopes in Grass Phytoliths and Soil Organic Matter. Journal of Vegetation Science 11:71-76. McCoy, M.M. 2006. High resolution fire and vegetation history of Garry oak ecosystems in British Columbia.  MSc Thesis. Simon Fraser University, Burnaby, Canada. McCune, B., and B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, Oregon, USA.   207McCune, J.L., W.L. Harrower, S. Avery-Gomm, J.M. Brogan, A.M. Csergo, L.N. Davidson, A. Garani, L.R. Halpin, L.P.J. Lipsen, C. Lee, J.C. Nelson, L.R. Prugh, C. Stinson, C.K. Whitney, and J. Whitton. 2013a. Threats to Canadian species at risk:  an analysis of finalized recovery strategies. Biological Conservation 166: 254-265. McCune, J.L., and M.G. Pellatt. 2013. Phytoliths of southeastern Vancouver Island, Canada, and their potential use to reconstruct shifting boundaries between Douglas-fir forest and oak savannah. Palaeogeography Palaeoclimatology Palaeoecology 383-384:59-71.   McCune, J. L., M. G. Pellatt, and M. Vellend. 2013b. Multidisciplinary synthesis of long-term human-ecosystem interactions:  a perspective from the Garry oak ecosystem of British Columbia. Biological Conservation 166:293-300. McCune, J. L. and M. Vellend. 2013. Gains in native species promote biotic homogenization over four decades in a human-dominated landscape. Journal of Ecology 101:1542-1551. McIntyre, S. and S. Lavorel. 1994. Predicting richness of native, rare, and exotic plants in response to habitat and disturbance variables across a variegated landscape. Conservation Biology 8:521-531. McIntyre, S., S. Lavorel, and R. M. Tremont. 1995. Plant life-history attributes:  their relationship to disturbance response in herbaceous vegetation. Journal of Ecology 83:31-44. McKinney, M. L. 2004. Do exotics homogenize or differentiate communities?  Roles of sampling and exotic species richness. Biological Invasions 6:495-504. McKinney, M. L. 2008. Do humans homogenize or differentiate biotas?  It depends. Journal of Biogeography 35:1960-1961.   208McKinney, M. L. and J. L. Lockwood. 1999. Biotic homogenization: a few winners replacing many losers in the next mass extinction. Trends in Ecology & Evolution 14:450-453. McLay, E. 2010. In search of camas:  archaeological correlates of camas root intensification on Vancouver Island and Southern Gulf Islands, British Columbia. 33rd Annual Conference of the Society of Ethnobiology. May 5-8, Victoria, Canada. McNamee, C. 2013. Soil phytolith assemblages of the American Southwest:  the use of historical ecology in taphonomic studies. PhD Thesis. University of Calgary, Calgary, Canada. Meidinger, D., and J. Pojar. 1991. Ecosystems of British Columbia. BC Ministry of Forests, Victoria, British Columbia. Metcalfe, C. R. 1960. Anatomy of the Monocotyledons. Clarendon Press, Oxford, UK. Metzger, J.P. 2000. Tree functional group richness and landscape structure in a Brazilian tropical fragmented landscape. Ecological Applications 10:1147-1161. Miller, T. J., P. F. Quintana-Ascencio, S. Maliakal-Witt, and E. S. Menges. 2012. Metacommunity dynamics over 16 years in a pyrogenic shrubland. Conservation Biology 26:357-366. Morlan, R.E. 2005. CARD, Canadian Archaeological Radiocarbon Database.  Canadian Museum of Civilization.  http://www.canadianarchaeology.ca. Accessed 3 May 2012 Morris, L. R., F. A. Baker, C. Morris, and R. J. Ryel. 2009. Phytolith types and type-frequencies in native and introduced species of the sagebrush steppe and pinyon-juniper woodlands of the Great Basin, USA. Review of Palaeobotany and Palynology 157:339-357. Morris, L. R., R. J. Ryel, and N. E. West. 2010a. Can soil phytolith analysis and charcoal be used as indicators of historic fire in the pinyon-juniper and sagebrush steppe ecosystem types of the Great Basin Desert, USA? The Holocene 20:105-114.   209Morris, L. R., N. E. West, and R. J. Ryel. 2010b. Testing soil phytolith analysis as a tool to understand vegetation change in the sagebrush steppe and pinyon-juniper woodlands of the Great Basin Desert, USA. The Holocene 20:697-709. Naaf, T. and M. Wulf. 2010. Habitat specialists and generalists drive homogenization and differentiation of temperate forest plant communities at the regional scale. Biological Conservation 143:848-855. Nasr, M., M. Castonguay, J. Ogilvie, B. A. Raymond, and P. A. Arp. 2012. Modelling and mapping critical loads and exceedances for the Georgia Basin, British Columbia, using a zero base-cation depletion criterion. Journal of Limnology 69 (Suppl. 1):181-192. National Archives of Canada. 1999. Canada Land Inventory Level II UTM Digital Data, Soil Capability for Agriculture.  The Canada Centre for Remote Sensing, Natural Resources Canada, Government of Canada, Ottawa, Canada. Norgren, J. A. 1973. Distribution, form and significance of plant opal in Oregon soils. PhD Thesis. Oregon State University, Corvallis, USA. Norment, C. 2002. On grassland bird conservation in the Northeast. The Auk 199:271-269. Nowacki, G. J. and M. D. Abrams. 2008. The demise of fire and 'mesophication' of forests in the Eastern United States. Bioscience 58:123-138. Ockinger, E., O. Schweiger, T. O. Crist, D. M. Debinski, J. Krauss, M. Kuussaari, J. D. Peterson, J. Poyry, J. Settele, K. S. Summerville, and R. Bommarco. 2010. Life-history traits predict species responses to habitat area and isolation:  a cross-continental synthesis. Ecology Letters 13:969-979.   210Oksanen, J., R. Kindt, P. Legendre, B. O?Hara, G.L. Simpson, P. Solymos, M. Henry, H. Stevens, and H. Wagner. 2009. Vegan: community ecology package. R package version 1.15-4. Okunaka, R., T. Kawano, and J. Inoue. 2012. Holocene history of intentional fires and grassland development on the Soni Plateau, Central Japan, reconstructed from phytolith and macroscopic charcoal records within cumulative soils, combined with paleoenvironmental data from mire sediments. The Holocene 22:793-800. Olden, J. D. and N. L. Poff. 2003. Toward a mechanistic understanding and prediction of biotic homogenization. The American Naturalist 162:442-460. Olden, J. D. and T. P. Rooney. 2006. On defining and quantifying biotic homogenization. Global Ecology and Biogeography 15:113-120. Orme, D., R. Freckleton, G. Thomas, T. Petzoldt, S. Fritz, N. Isaac, and W. Pearse. 2013. Comparative analyses of phylogenetics and evolution in R. R package version 0.5. Palmer, M. W. 1993. Potential biases in site and species selection for ecological monitoring. Environmental Monitoring and Assessment 26:277-282. Paradis, E., B. Bolker, J. Claude, H.S. Cuong, R. Desper, B. Durand, J. Dutheil, O. Gascuel, C. Heibl, D. Lawson, V. Lefort, P. Legendre, J. Lemon, Y. Noel, J. Nylander, R. Opgen-Rhein, A.A. Popescu, K. Schliep, K. Strimmer, and D. de Vienne. 2013. Analyses of phylogenetics and evolution. R package version 3.0-8. Pardini, R., A. D. Bueno, T. A. Gardner, P. I. Prado, and J. P. Metzger. 2010. Beyond the fragmentation threshold hypothesis: regime shifts in biodiversity across fragmented landscapes. PLoS one 5:1-10.   211Parks Canada Agency. 2006a. Recovery strategy for multi-species at risk in Garry Oak woodlands in Canada. Parks Canada Agency, Ottawa, Canada. Parks Canada Agency. 2006b. Recovery strategy for multi-species at risk in maritime meadows associated with Garry Oak ecosystems in Canada.  Parks Canada Agency, Ottawa, Canada. Parr, J. F. 2006. Effect of fire on phytolith coloration. Geoarchaeology-an International Journal 21:171-185. Parry, D. W. and F. Smithson. 1964. Types of opaline silica depositions in the leaves of British grasses. Annals of Botany 28:169-185. Pearsall, D. M. 2000. Phytolith Analysis. Pages 355-496  Paleoethnobotany:  A Handbook of Procedures, Second Edition. Academic Press, Inc., San Diego. Pellatt, M.G., Z. Gedalof, M. McCoy, K. Bodtker, A. Cannon, S. Smith, B. Beckwith, R. Mathewes, and D. Smith. 2007. Fire History and Ecology of Garry Oak and Associated Ecosystems in British Columbia:  Final Report for the Interdepartmental Recovery Fund, Project 733.  Parks Canada, Vancouver, Canada. Pellatt, M.G., R.J. Hebda, and R.W. Mathewes. 2001. High-resolution Holocene vegetation history and climate from Hole 1034B, ODP leg 169S, Saanich Inlet, Canada. Marine Geology 174:211-226. Peter, M., A. Gigon, P. J. Edwards, and A. Luscher. 2009. Changes over three decades in the floristic composition of nutrient-poor grasslands in the Swiss Alps. Biodiversity and Conservation 18:547-567.   212Petit, S., L. Griffiths, S. M. Smart, G. M. Smith, R. C. Stuart, and S. M. Wright. 2004. Effects of area and isolation of woodland patches on herbaceous plant species richness across Great Britain. Landscape Ecology 19:463-471. Piperno, D. 1985. Phytolithic analysis of geological sediments from Panama. Antiquity LIX:13-19. Piperno, D. R. 1988. Phytolith Analysis:  An Archaeological and Geological Perspective. Academic Press, San Diego, CA, USA. Piperno, D. R. and P. Becker. 1996. Vegetational history of a site in the central Amazon basin derived from phytolith and charcoal records from natural soils. Quaternary Research 45:202-209. Piqueray, J., E. Bisteau, S. Cristofoli, R. Palm, P. Poschlod, and G. Mahy. 2011a. Plant species extinction debt in a temperate biodiversity hotspot:  community, species and functional traits approaches. Biological Conservation 144:1619-1629. Piqueray, J., S. Cristofoli, E. Bisteau, R. Palm, and G. Mahy. 2011b. Testing coexistence of extinction debt and colonization credit in fragmented calcareous grasslands with complex historical dynamics. Landscape Ecology 26:823-836. Proctor, K. 2010. Introducing traditional blue camas (Camassia leightlinii and C. quamash) cultivation practices into a Garry oak (Quercus garryana) meadow preserve in British Columbia.  Pages 28-30 in Garry Oak Ecosystems Recovery Team Research Colloquium Proceedings. Victoria, BC, Canada.  Prychid, C. J., P. J. Rudall, and M. Gregory. 2003. Systematics and biology of silica bodies in monocotyledons. Botanical Review 69:377-440.   213Purschke, O., M. T. Sykes, T. Reitalu, P. Poschlod, and H. C. Prentice. 2012. Linking landscape history and dispersal traits in grassland plant communities. Oecologia 168:773-783. Pyne, S. J. 1995. World Fire:  The Culture of Fire on Earth. Holt, New York, USA. R Core Development Team. 2012. R version 2.15.0. R Foundation for Statistical Computing, Vienna, Austria. Rackham, O. 1998. Trees and woodland in a cultural landscape:  the history of woods in England. Pages 139-147 in K. Sassa, editor. Environmental Forest Science. Springer Science+Business Media, Dordrecht, The Netherlands. Redman, C. L. 1999. Human Impact on Ancient Environments. University of Arizona Press, Tucson, USA. Reimer, P. J., M. G. L. Baillie, E. Bard, A. Bayliss, J. W. Beck, P. G. Blackwell, C. Bronk Ramsey, C. E. Buck, G. S. Burr, R. L. Edwards, M. Friedrich, P. M. Grootes, T. P. Guilderson, I. Hajdas, T. J. Heaton, A. G. Hogg, K. A. Hughen, K. F. Kaiser, B. Kromer, F. G. McCormac, S. W. Manning, R. W. Reimer, D. A. Richards, J. R. Southon, S. Talamo, C. S. M. Turney, J. van der Plicht, and C. E. Weyhenmeyer. 2009. INTCAL09 and MARINE09 radiocarbon age calibration curves, 0-50,000 years cal BP. Radiocarbon 51:1111-1150. Rick, T.C., and R. Lockwood. 2013. Integrating paleobiology, archaeology, and history to inform biological conservation. Conservation Biology 27:45-54. Rival, L. 2006. Amazonian historical ecologies. Journal of the Royal Anthropological Institute N.S.:79-94. Roberts, D.W. 2010. Package ?labdsv?: ordination and multivariate analysis for ecology. R package version 1.4-1.   214Roemer, H.L. 1972. Forest vegetation and environments on the Saanich peninsula, Vancouver Island. PhD Thesis. University of Victoria, Victoria, Canada. Roemer, H.L. 1993. Vegetation and ecology of Garry oak woodlands. Pages 19-24 in R.J. Hebda and F. Aitkens, editors. Garry Oak Meadow Colloquium. Garry Oak Meadow Preservation Society, Victoria, Canada. Rogers, D. A., T. P. Rooney, D. Olson, and D. M. Waller. 2008. Shifts in southern Wisconsin forest canopy and understory richness, composition, and heterogeneity. Ecology 89:2482-2492. Rogers, D. A., T. P. Rooney, T. J. Hawbaker, V. C. Radeloff, and D. M. Waller. 2009. Paying the Extinction Debt in Southern Wisconsin Forest Understories. Conservation Biology 23:1497-1506. Rooney, T. P., S. M. Wiegmann, D. A. Rogers, and D. M. Waller. 2004. Biotic impoverishment and homogenization in unfragmented forest understory communities. Conservation Biology 18:787-798. Ross, L. C., S. J. Woodin, A. Hester, D. B. A. Thompson, and H. J. B. Birks. 2010. How important is plot relocation accuracy when interpreting re-visitation studies of vegetation change? Plant Ecology & Diversity 3:1-8. Rovner, I. 1983. Plant opal phytolith analysis: major advances in archaeobotanical research. Advances in Archaeological Method and Theory 6:225-266. Rovner, I. 1986. Downward percolation of phytoliths in stable soils:  a non-issue. Pages 23-30 in I. Rovner, editor. Occasional Paper of the Phytolitharien. North Carolina State University, Raleigh, USA.   215Royal Botanic Gardens Kew. 2008. Seed Information Database (SID). Version 7.1. http://data.kew.org/sid/. Russell, E. W. B. 1997. People and the Land through Time - Linking Ecology and History. Yale University Press, New Haven. Saar, L., K. Takkis, M. Partel, and A. Helm. 2012. Which plant traits predict species loss in calcareous grasslands with extinction debt? Diversity and Distributions 18:808-817. Santos, G. M., A. Alexandre, H. H. G. Coe, P. E. Reyerson, J. R. Southon, and C. N. De Carvalho. 2010. The phytolith 14C puzzle:  a tale of background determinations and accuracy tests. Radiocarbon 52:113-128. Santos, T., J.L. Teller?a, and E. Virg?s. 1999. Dispersal of Spanish juniper Juniperus thurifera by birds and mammals in a fragmented landscape.  Ecography 22:193-204. Sax, D. F. and S. D. Gaines. 2003. Species diversity:  from global decreases to local increases. Trends in Ecology & Evolution 18:561-566. Sinclair, A.R.E. 2012. Ecological history guides the future of conservation:  lessons from Africa. Pages 265-272 in J.A. Wiens, G.D. Hayword, H.D. Safford, and C.M. Giffen, editors. Historical Environmental Variation in Conservation and Natural Resource Management. John Wiley & Sons Ltd., Hoboken, NJ, USA.  Smart, S. M., R. G. H. Bunce, R. Marrs, M. Le Duc, L. G. Firbank, L. C. Maskell, W. A. Scott, K. Thompson, and K. J. Walker. 2005. Large-scale changes in the abundance of common higher plant species across Britain between 1978, 1990 and 1998 as a consequence of human activity:  tests of hypothesised changes in trait representation. Biological Conservation 124:355-371.   216Smart, S. M., K. Thompson, R. H. Marrs, M. G. Le Duc, L. C. Maskell, and L. G. Firbank. 2006. Biotic homogenization and changes in species diversity across human-modified ecosystems. Proceedings of the Royal Society B-Biological Sciences 273:2659-2665. Smith, S. 2007. Garry oak savannah stand history and change in coastal southern British Columbia. MSc Thesis. University of Guelph, Guelph, Canada. Smith, F.A., and A.G. Boyer. 2012. Losing time? Incorporating a deeper temporal perspective into modern ecology. Frontiers in Biogeography 4:26-39. Smith, C. A. S., K. T. Webb, E. Kenney, A. Anderson, and D. Kroetsch. 2011. Brunisolic soils of Canada:  genesis, distribution and classification. Canadian Journal of Soil Science 91:695-717. Soreng, R.J., G. Davidse, P.M. Peterson, F.O. Zuloaga, E.J. Judziewicz, T.S. Filgueiras, O. Morrone, and K. Romaschenko, K. 2012. A world-wide phylogenetic classification of Poaceae (Gramineae). Missouri Botanical Garden, Saint Louis, MS.  http://www.tropicos.org/docs/meso/CLASSIFICATION%20OF%20world%20grasses%202012%20Oct%2018c.htm  Sprenger, C. B. and P. W. Dunwiddie. 2011. Fire history of a Douglas-Fir-Oregon White Oak woodland, Waldron Island, Washington. Northwest Science 85:108-119. Sprugel, D. G. 1991. Disturbance, equilibrium, and environmental variability:  what is 'natural' vegetation in a changing environment? Biological Conservation 58:1-18. Stanley, A.G., P.W. Dunwiddie, and T.N. Kaye. 2011. Restoring invaded Pacific Northwest prairies:  management recommendations from a region-wide experiment. Northwest Science 85:233-246.   217Statistics Canada (Dominion Bureau of Statistics). 1969. Canada Year Book 1969. Queen?s Printer, Ottawa, Canada. Statistics Canada. 2011. Canada Year Book 2011. Minister of Industry, Ottawa, Canada. Suttles, W.P. 1951. Economic life of the Coast Salish of Haro and Rosario Straits. PhD Thesis. University of Washington, Seattle, USA. Sutton, F. M. and J. W. Morgan. 2009. Functional traits and prior abundance explain native plant extirpation in a fragmented woodland landscape. Journal of Ecology 97:718-727. Swetnam, T.W., C.D. Allen, and J.L. Betancourt. 1999. Applied historical ecology: using the past to manage for the future. Ecological Applications 9:1189-1206. Szabo, P. 2010. Why history matters in ecology:  an interdisciplinary perspective. Environmental Conservation 37:380-387. Targulian, V. O. and S. V. Goryachkin. 2004. Soil memory:  types of record, carriers, hierarchy and diversity. Revista Mexicana de Ciencias Geologicas 21:1-8. Thimonier, A., J. L. Dupouey, F. Bost, and M. Becker. 1994. Simultaneous eutrophication and acidification of a forest ecosystem in north-east France. New Phytologist 126:533-539. Thoms, A.V. 1989. The Northern Roots of Hunter-Gatherer Intensification:  Camas and the Pacific Northwest. PhD Thesis. Washington State University, Pullman, USA. Tremlova, K. and Z. Munzbergova. 2007. Importance of species traits for species distribution in fragmented landscapes. Ecology 88:965-977. Trombulak, S.C. and C.A. Frissell. 2000. Review of ecological effects of roads on terrestrial and aquatic communities. Conservation Biology 14:18-30. Turner, N.C., and M.A.M. Bell. 1971. The ethnobotany of the Coast Salish Indians of Vancouver Island. Economic Botany 25:63-104.   218Turner, N.J. 1999. "Time to Burn":  Traditional us of fire to enhance resource production by aboriginal peoples in British Columbia. Pages 185-218 in R. Boyd, editor. Indians, Fire and the Land in the Pacific Northwest. Oregon State University Press, Corvallis, USA. Turner, N.J., and H.V. Kuhnlein. 1983. Camas (Camassia spp.) and riceroot (Fritillaria spp.):  Two Liliaceous "root" foods of the Northwest Coast Indians. Ecology of Food and Nutrition 13:199-219. Twiss, P. C., E. Suess, and R. M. Smith. 1969. Morphological classification of grass phytoliths. Soil Science Society of America Proceedings 33:109-115. USDA, NRCS. 2013. The PLANTS Database. National Plant Data Team, Greensboro, NC, USA. http://plants.usda.gov Vale, T.R. 2000. Pre-Columbian North America:  Pristine or Humanized - or both? Ecological Restoration 18:2-3. Valentine, K. W. G., P. N. Sprout, T. E. Baker, and L. M. Lavkulich, editors. 1978. The Soil Landscapes of British Columbia. The Resource Analysis Branch, Ministry of the Environment, Victoria, Canada. Vallet, J., H. Daniel, V. Beaujouan, F. Roze, and S. Pavoine. 2010. Using biological traits to assess how urbanization filters plant species of small woodlands. Applied Vegetation Science 13:412-424. Van Calster, H., L. Baeten, A. De Schrijver, L. De Keersmaeker, J. E. Rogister, K. Verheyen, and M. Hermy. 2007. Management driven changes (1967-2005) in soil acidity and the understorey plant community following conversion of a coppice-with-standards forest. Forest Ecology and Management 241:258-271.   219Vellend, M., A.D. Bjorkman, and A. McConchie. 2008. Environmentally biased fragmentation of oak savanna habitat on southeastern Vancouver Island, Canada. Biological Conservation 141:2576-2584. Vellend, M., C.D. Brown, H.M. Kharouba, J.L. McCune, and I.H. Myers-Smith. 2013. Historical ecology:  using unconventional data sources to test for impacts of global environmental change.  American Journal of Botany 100:1294-1305. Vellend, M., K. Verheyen, H. Jacquemyn, A. Kolb, H. Van Calster, G. Peterken, and M. Hermy. 2006. Extinction debt of forest plants persists for more than a century following habitat fragmentation. Ecology 87:542-548. Vellend, M., K. Verheyen, K. M. Flinn, H. Jacquemyn, A. Kolb, H. Van Calster, G. Peterken, B. J. Graae, J. Bellemare, O. Honnay, J. Brunet, M. Wulf, F. Gerhardt, and M. Hermy. 2007. Homogenization of forest plant communities and weakening of species-environment relationships via agricultural land use. Journal of Ecology 95:565-573. Verheyen, K., O. Honnay, G. Motzkin, M. Hermy, and D. R. Foster. 2003. Response of forest plant species to land-use change: a life-history trait-based approach. Journal of Ecology 91:563-577. Vitousek, P. M., C. M. D'Antonio, L. L. Loope, and R. Westbrooks. 1996. Biological invasions as global environmental change. American Scientist 84:468-478. Weaver, M., and M. Kellman. 1981. The effects of forest fragmentation on woodlot tree biotas in southern Ontario. Journal of Biogeography 8:199-210. Webb, C.O., and M.J. Donoghue. 2005. Phylomatic:  tree assembly for applied phylogenetics. Molecular Evolution Notes 5:181-183.   220Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant and Soil 199:213-227. Westoby, M., D. S. Falster, A. T. Moles, P. A. Vesk, and I. J. Wright. 2002. Plant ecological strategies:  some leading dimensions of variation between species. Annual Review of Ecology and Systematics 33:125-159. Whiston Spirn, A. 1996. Constructing Nature:  The Legac of Frederick Law Olmsted. Pages 91-113 in W. Cronon, editor. Uncommon Ground:  Rethinking the human place in nature. W. W. Norton & Company, New York, NY, USA. Whitlock, C. 1992. Vegetational and climatic history of the Pacific Northwest during the last 20,000 years ? implications for understanding present-day biodiversity. Northwest Environmental Journal 8:5-28. Whitlock, C. and M. A. Knox. 2002. Prehistoric burning in the Pacific Northwest:  human versus climatic influences. Pages 195-231 in T. R. Vale, editor. Fire, Native Peoples and the Natural Landscape. Island Press, Washington, USA. Whitney, G.G., and J. DeCant, J. 2005. Government land office survey and other early land surveys. Pages 147-172 in E. Egan and E. Howell, editors. The Historical Ecology Handbook. Island Press, Washington, USA. Wilcove, D. S., C. H. McLellan, and A. P. Dobson. 1986. Habitat fragmentation in the temperate zone. Pages 237-256 in M. E. Soule, editor. Conservation Biology: The science of scarcity and diversity. Sinauer Associates, Sunderland, MA, USA. Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying threats to imperiled species in the United States. Bioscience 48:607-615.   221Williams, J. W. and S. T. Jackson. 2007. Novel climates, no-analog communities, and ecological surprises. Frontiers in Ecology and the Environment 5:475-482. Williams, N. S. G., J. W. Morgan, M. J. McDonnell, and M. A. McCarthy. 2005. Plant traits and local extinctions in natural grasslands along an urban-rural gradient. Journal of Ecology 93:1203-1213.  Willis, K.J., M.B. Araujo, K.D. Bennett, B. Figueroa-Rangel, C.A. Froyd, and N. Myers, N. 2007. How can a knowledge of the past help to conserve the future?  Biodiversity conservation and the relevance of long-term ecological studies. Philosophical Transactions of the Royal Society B 362:175-186. Willis, K.J., and H.J.B. Birks. 2006. What is natural? The need for a long-term perspective in biodiversity conservation. Science 314:1261-1265. Willis, K.J., L. Gillson, and T.M. Brncic. 2004. How "virgin" is virgin rainforest? Science 304:402-403. With, K. A. and T. O. Crist. 1995. Critical thresholds in species' responses to landscape structure. Ecology 76:2446-2459. Zhang, Q. and R. J. Hebda. 2005. Abrupt climate change and variability in the past four millennia of the southern Vancouver Island, Canada. Geophysical Research Letters 32:1-4.   222Appendices Appendix A  Supplementary data for resurveys (Chapters 3 and 4)  A.1 Compilation of trait data (Chapter 3) In order to asses which potential drivers of change in plant communities have been the most important in the study area over the past four decades, I collected data on traits of understorey species based on these drivers (Table A.1).  Trait data was collected as follows:  I coded all species as ?exotic? or ?native? according to the species account on the Electronic Atlas of the Flora of British Columbia (E-Flora; Klinkenberg, 2013).     To assess whether the ability to tolerate disturbance was a key trait for succeeding in the highly human-dominated landscape of the Saanich peninsula, I assessed disturbance tolerance for each species from the ?Habitat/Range? description in E-flora.  If any of the following were encountered in the habitat description, the species was coded ?1?:  burns, continuously disturbed sites, disturbed areas/sites, disturbed communities, ditches, fencelines, fencerows, fields, fire-disturbed sites, lawns, pastures, railways, roadsides, waste places, early-seral/young-seral/early successional forests, garden, garden escape.  A few rare garden escapes not given habitat data in E-flora were coded as ?1?.  All others were coded as ?0? for disturbance tolerance.    Given the increase in fossil-fuel powered vehicles and increased shipping traffic to the nearby port of Vancouver, emissions of nitrogen oxides are thought to be on the rise in the region (Nasr et al., 2010), which could lead to eutrophication of terrestrial systems via atmospheric nitrogen deposition (Bobbink et al., 2010).  I assessed the preferred soil nutrient status of each species   223using the ?Ecology? section of E-Flora, which provides the ?modal nutrient regime class? for many plant species as calculated from data provided by the British Columbia Ministry of Forests and Range.  This value represents the average nutrient availability of all survey plots (out of over 30,000 across British Columbia) in which a species was present. Values range from ?1? (very poor, oligotrophic) to ?6? (saline, hypereutrophic).    Climatic warming is a potential driver of plant community changes (e.g. Damschen et al., 2010).   I therefore assessed the overall North American distribution of each species to identify species with affinities for warmer (more Southern) climates.  I coded the nature of the overall North American range of each species using the distribution maps provided by E-flora and the USDA Plants database (USDA, NRCS, 2011) in combination.  Species for which Vancouver Island represents the northern edge of their range were coded ?1?, whereas species having ranges in which Vancouver Island is central were coded ?2?.     Increased fragmentation of natural vegetation can favour species more able to disperse propagules successfully across inhospitable habitats (e.g. Marini et al., 2012).  In order to approximate the dispersal ability of plant species, I gathered data on seed weight and seed dispersal mechanism. I compiled values for seed weight from the Kew seed information database (Royal Botanic Gardens Kew, 2008), which reports the average weight of 1000 seeds in grams.  I compiled seed dispersal mechanism from the Kew database, or from a database compiled from the literature for plants of the Garry oak vegetation type (Bennett et al., 2013).   I coded dispersal mechanism in 3 categories, where ?0? = no assistance, ballistic dispersal, or ant dispersal, ?1? = dispersed by wind or water, and ?2? = dispersed by vertebrates.     224Due to scarce large predators and reduced hunting levels, populations of deer in the region are thought to have increased since the 1940s (Gonzales and Arcese, 2008).  To assess the preference of deer for each plant species, I recorded deer palatability from a study that ranked the palatability of plant species on southern Vancouver Island using stomach content analysis, observation, and plant quadrat surveys (Cowan, 1945).  Values range from 0 (never eaten) to 3 (highly palatable).    Increased fragmentation can lead to more high light, forest edge conditions, favouring shade intolerant species (e.g. Metzger, 2000).  Alternatively, maturation of young forests can reduce light levels in the understorey, favouring shade tolerant species (e.g. Weaver and Kellman, 1981).  In the absence of disturbances such as fire, many open vegetation types in this region are prone to invasion by shrub species (Fuchs, 2001).  Therefore, I assessed shade tolerance of each species from E-flora, or from the USDA Plants database where information was not available on E-Flora.  Values range from ?1? (very shade tolerant) to ?5? (very shade intolerant).   Each species was also designated as a shrub or herbaceous plant (including forbs, grasses, and ferns).   Finally, specific leaf area (SLA; leaf area per unit dry mass) has been found to correlate with gradients of nutrient and water availability, with high SLA being associated with fast-growing species in productive conditions, and low SLA being associated with more stress-tolerant species (Westoby, 1998).  I obtained specific leaf area in square millimeters per gram from a database of standardized measurements of plants from the region compiled by Dr. Will Cornwell.   225 Table A.1: Hypothesized drivers of plant community change, expected shifts in plant traits, and the percentage of species for which data were available for each trait  SPECIES TRAITS: DRIVERS: origin (native vs. exotic) disturbance tolerance nutrient regime geographic range seed weight dispersal mechanism deer palatability shade tolerance life form (herb vs. shrub) Specific leaf area (SLA) Disturbance, broadly defined  exotics ? ?        ? Warming climate     southern ?      ? Increased nitrogen deposition   high ?       ? Fragmentation of vegetation     ? wind, animal ?     Increased herbivory by deer       ?    Succession, infilling by shrubs        ? shrubs ?  PERCENT  OF SPECIES WITH DATA 100% 100% 71% 99% 78% 32% 33% 58% 100% 56%        226A.2 Indicator species analysis results (Chapter 3) Table A.2:  Results of indicator species analysis for all 101 understory species (or taxa) present in at least 5% of all plot-year combinations (18 of 368 total plots).  There were 184 plots in each year, 115 in the Douglas-fir vegetation type, 43 in the Garry oak vegetation type, and 26 in the Arbutus vegetation type.  Origin refers to native (N) or exotic (E).  If 2009 appears in the ?Year indicator? column, the species was significantly more frequent and/or abundant in 2009 than in 1968, and was deemed to be a ?winner?.  If 1968 appears in the ?Year indicator? column, the species was significantly more frequent and/or abundant in 1968 than in 2009, and was deemed to be a ?loser?.  Empty cells in the ?Year indicator? column indicate that there was no significant difference in frequency/abundance of the species between years OR that species did not occur in any plot in the given vegetation type.  Frequency refers to the number of plots in which the species was present in a given year.  Total Cover is the sum of the percent cover (expressed as midpoints of cover classes) of the species in all plots in which it was present in the given year.   All Plots (N=184)   Douglas-fir plots (N=115)   Garry oak plots (N=43)   Arbutus plots (N=26)   Species (or taxon) Origin Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Achillea millefolium N  31 24 13.9 9.6  0 0 0 0  29 24 13.7 9.6  2 0 0.2 0 Achlys triphylla N  74 72 538.6 226.7 1968 73 67 538.1 225  0 0 0 0  1 5 0.5 1.7 Adenocaulon bicolor N  47 55 36 33.6  37 46 33 31.1  3 1 0.7 0.1  7 8 2.3 2.4 Aira caryophyllea E  12 20 16 10.4  0 0 0 0  12 18 16 10.2  0 2 0 0.2 Aira praecox E  22 34 57.6 101.1  0 0 0 0  22 33 57.6 101  0 1 0 0.1 Amelanchier alnifolia N 2009 58 89 38.2 85.4 2009 36 62 20.4 44.7 2009 4 9 2 14.5  18 18 15.8 26.2 Anthoxanthum odoratum E 2009 2 36 0.6 377 2009 0 6 0 3 2009 2 26 0.6 370  0 4 0 4 Athyrium filix-femina N 2009 3 15 18 73.5  3 15 18 73.5  0 0 0 0  0 0 0 0 Bromus carinatus N 1968 33 13 50.2 14.1  0 0 0 0 1968 31 13 49.6 14.1  2 0 0.6 0 Bromus hordeaceus E  18 19 17.8 145.5  0 0 0 0 2009 18 19 17.8 145.5  0 0 0 0 Bromus rigidus/sterilis E  22 34 628.6 622.7  0 1 0 0.5  22 30 628.6 621.1  0 3 0 1.1 Bromus vulgaris N  69 69 40.5 58.5  38 42 18.2 27.8  5 5 2.1 12.5  26 22 20.2 18.2 Camassia leichtlinii/quamash N  35 40 289.8 183.6  0 0 0 0  33 36 289.6 182.4  2 4 0.2 1.2 Campanula scouleri N  16 24 24.9 8.4 2009 3 22 0.7 8.2  4 0 4 0 1968 9 2 20.2 0.2 Cardamine oligosperma N 2009 11 47 1.9 7.5  0 1 0 0.1 2009 8 34 1.6 5.8 2009 3 12 0.3 1.6 Carex inops N 1968 30 24 148.3 15.2  0 4 0 1.2 1968 25 15 146.6 9.5  5 5 1.7 4.5   227  All Plots (N=184)   Douglas-fir plots (N=115)   Garry oak plots (N=43)   Arbutus plots (N=26)   Species (or taxon) Origin Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Cerastium arvense N  29 24 12.9 8.4  0 0 0 0  29 24 12.9 8.4  0 0 0 0 Claytonia perfoliata N  20 41 65.9 53.5  0 0 0 0  20 35 65.9 51.7 2009 0 6 0 1.8 Clinopodium douglasii N  20 21 6.8 15.3  2 5 0.6 2.5  4 8 1.2 6.8  14 8 5 6 Collinsia parviflora N  20 25 25.3 5.3  0 0 0 0  20 25 25.3 5.3  0 0 0 0 Corallorhiza maculata N 2009 10 31 2.2 4.3  10 19 2.2 2.3  0 2 0 0.2 2009 0 10 0 1.8 Cynosurus echinatus E  13 14 23.8 54  0 0 0 0  13 14 23.8 54  0 0 0 0 Cytisus scoparius E 2009 29 52 165.4 324.2 2009 0 7 0 1.5 2009 29 40 165.4 316.2  0 5 0 6.5 Dactylis glomerata E 2009 9 23 10.1 105.6 2009 0 8 0 20.1  8 12 7.6 82  1 3 2.5 3.5 Daphne laureola E 2009 0 75 0 61.7 2009 0 46 0 18.2 2009 0 16 0 36.6 2009 0 13 0 6.9 Delphinium menziesii N  22 16 13 5.2  0 0 0 0  21 16 12.5 5.2  1 0 0.5 0 Dodecatheon hendersonii N  23 16 24.7 32.9  0 0 0 0  18 15 20.6 32.4  5 1 4.1 0.5 Elymus glaucus N  49 46 77.8 71.5  1 1 0.5 0.5  37 38 72.6 65.5  11 7 4.7 5.5 Epipactis helleborine E 2009 0 80 0 17.6 2009 0 76 0 16.8  0 1 0 0.1  0 3 0 0.7 Erythronium oreganum N 1968 53 44 83.5 22.4  7 5 4.7 1.3  25 25 52.3 16.1 1968 21 14 26.5 5 Festuca occidentalis/ idahoensis/roemeri N  55 48 31.9 22.4 2009 17 22 2.9 11 1968 20 8 20.8 4  18 18 8.2 7.4 Festuca subulata/subuliflora N  98 91 57 45.5  76 81 40.4 40.5  1 2 0.5 1 1968 21 8 16.1 4 Fragaria vesca N  13 11 5.7 5.1 2009 0 7 0 3.1  3 0 1.1 0  10 4 4.6 2 Fritillaria affinis N  15 9 5.1 1.3  0 0 0 0 1968 14 8 5 1.2  1 1 0.1 0.1 Galium aparine N  63 103 370.6 82.8 2009 3 38 1.5 23.8 1968 42 41 350.5 41  18 24 18.6 18 Galium triflorum N 2009 67 94 30.7 66.3 2009 63 86 28.7 63.9  0 2 0 0.6  4 6 2 1.8 Gaultheria shallon N  109 121 2615.7 2393.1  96 95 2550.5 2029.6  1 4 2.5 18.5 2009 12 22 62.7 345 Geranium molle E  17 20 57.5 18.8  0 0 0 0  17 18 57.5 18.6  0 2 0 0.2 Geranium robertianum E 2009 0 31 0 42.1 2009 0 26 0 25.1  0 1 0 0.5  0 4 0 16.5 Goodyera oblongifolia N  45 41 14.5 10.5  28 18 6 5.4  4 9 1.2 2.1  13 14 7.3 3 Hedera helix E 2009 6 102 1.8 647.3 2009 5 84 1.3 364.8  0 2 0 175 2009 1 16 0.5 107.5 Heuchera micrantha N  23 25 7.5 10.1 2009 0 7 0 2.7  15 11 5.9 4.7  8 7 1.6 2.7 Hieracium albiflorum N 1968 17 8 14.5 1.6  0 1 0 0.1  9 5 7.3 1.3 1968 8 2 7.2 0.2   228  All Plots (N=184)   Douglas-fir plots (N=115)   Garry oak plots (N=43)   Arbutus plots (N=26)   Species (or taxon) Origin Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Holodicus discolor N 1968 118 131 3703.4 2120.9  70 79 1478.9 865.2  22 28 866.5 440.7 1968 26 24 1358 815 Hypochaeris radicata E 2009 23 51 5.5 21.1 2009 0 12 0 2.8 2009 22 32 5 17.6  1 7 0.5 0.7 Ilex aquifolium E 2009 35 106 84.3 144.2 2009 32 89 83.2 132.5  0 3 0 1.5 2009 3 14 1.1 10.2 Mycelis muralis E 2009 42 120 62.8 60.4  39 90 61.7 51 2009 1 12 0.5 2.8 2009 2 18 0.6 6.6 Lathyrus nevadensis N  72 82 93.5 28.6  34 48 11.4 16.8  18 15 19 5.1 1968 20 19 63.1 6.7 Linnaea borealis N 1968 57 37 324.5 116.6 1968 53 35 283.9 113.6  0 0 0 0  4 2 40.6 3 Lomatium utriculatum N  15 14 41.7 8.6  0 0 0 0  15 14 41.7 8.6  0 0 0 0 Lonicera ciliosa N  84 106 191.9 81  59 73 42.7 48.9 2009 3 10 1.5 9 1968 22 23 147.7 23.1 Lonicera hispidula N 2009 38 99 164.2 198.1 2009 20 60 13.6 68.9 2009 7 15 9.1 34  11 24 141.5 95.2 Lotus micranthus N  12 16 5.2 3.6  0 0 0 0  12 15 5.2 3.5  0 1 0 0.1 Luzula multiflora N  18 27 7 13.5  0 4 0 0.8  15 18 6.7 10.2  3 5 0.3 2.5 Mahonia aquifolium/nervosa N  152 163 3617.9 2411.6 1968 112 113 3217.7 2070.5 2009 15 24 15.5 37.6  25 26 384.7 303.5 Melica subulata N  86 94 297.6 294.3  37 49 27.4 49.8  25 22 199.2 128  24 23 71 116.5 Mimulus alsinoides N  6 19 3.8 4.7  0 0 0 0  6 19 3.8 4.7  0 0 0 0 Moehringia macrophylla N  49 36 116.1 44.9  10 8 3.8 18.1  17 15 87.7 17.1 1968 22 13 24.6 9.7 Monotropa uniflora N 2009 15 42 3.5 12.6 2009 13 37 3.3 11.3  0 0 0 0  2 5 0.2 1.3 Montia parvifolia N  21 23 36.2 15.9  0 1 0 0.1  16 18 34.5 14.2  5 4 1.7 1.6 Myosotis discolor E 2009 4 27 1.2 7.9  0 0 0 0 2009 4 26 1.2 7.8  0 1 0 0.1 Nemophila parviflora N 2009 13 58 27 47.1 2009 0 15 0 5.1  13 32 27 33.3 2009 0 11 0 8.7 Oemleria cerasiformis N 2009 35 117 91.9 299.2 2009 33 84 91.7 181.6 2009 1 16 0.1 113.9 2009 1 17 0.1 3.7 Osmorhiza berteroi N  42 74 35.1 34.2 2009 16 47 3.2 26.7  9 11 8.1 2.7  17 16 23.8 4.8 Paxistima myrsinites N  18 25 255.7 20.1  3 10 2.7 8.2  2 6 1 5.8 1968 13 9 252 6.1 Philadelphus lewisii N  12 6 149.3 2.6 1968 12 6 149.3 2.6  0 0 0 0  0 0 0 0 Piperia elongata/ transversa/unalascensis N 1968 24 15 8.4 3.1  0 0 0 0  13 13 4.5 2.9 1968 11 2 3.9 0.2 Pentagramma triangularis N  5 14 2.1 5  0 0 0 0  5 13 2.1 4.9  0 1 0 0.1 Plectritis congesta N  17 18 22.1 10.2  0 0 0 0  15 17 21.5 9.7  2 1 0.6 0.5 Poa pratensis N  25 29 167.8 36.5  0 2 0 3  24 26 167.7 33  1 1 0.1 0.5   229  All Plots (N=184)   Douglas-fir plots (N=115)   Garry oak plots (N=43)   Arbutus plots (N=26)   Species (or taxon) Origin Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Polypodium glycyrrhiza N  46 79 796.7 666.7 2009 5 24 1.3 6  26 38 621.7 508.8  15 17 173.7 151.9 Polystichum munitum N  129 156 1633 1622.6  100 115 1617.7 1578.5  16 20 10.8 18 2009 13 21 4.5 26.1 Prunus laurocerasus E 2009 0 41 0 12.1 2009 0 37 0 11.7  0 3 0 0.3  0 1 0 0.1 Pteridium aquilinum N  78 76 183.9 100.2  75 70 168.3 95.2  0 0 0 0  3 6 15.6 5 Ranunculus occidentalis N  24 17 11.2 6.1  0 1 0 0.1  21 15 10.5 5.9  3 1 0.7 0.1 Rhamnus purshiana N  54 43 58 20.7 1968 51 40 57.3 17.6  1 2 0.1 2.6  2 1 0.6 0.5 Rosa gymnocarpa N  119 115 178.1 183.1  87 75 111 82  8 14 20.1 12.6  24 26 47 88.5 Rosa nutkana N 2009 5 17 4.5 34.6  3 3 1.5 5.5 2009 2 9 3 24.6  0 5 0 4.5 Rubus armeniacus E 2009 0 19 0 78.3 2009 0 10 0 57.7 2009 0 8 0 20.1  0 1 0 0.5 Rubus parviflorus N  22 26 15 34.2  22 24 15 33.2  0 0 0 0  0 2 0 1 Rubus spectabilis N  10 19 39.6 76.3  10 19 39.6 76.3  0 0 0 0  0 0 0 0 Rubus ursinus N 2009 132 145 291.1 802.2 2009 104 108 244.1 494 2009 4 12 2 90.1 2009 24 25 45 218.1 Rumex acetosella E 2009 17 35 4.5 33.5  0 0 0 0 2009 17 34 4.5 33  0 1 0 0.5 Sanicula crassicaulis N  60 92 91.1 91.1 2009 2 33 0.2 9.7  35 36 71.8 63.9  23 23 19.1 17.5 Sedum spathulifolium N  20 17 40.1 62.4  0 0 0 0  18 16 39.5 61.9  2 1 0.6 0.5 Maianthemum racemosum N  14 12 9.8 5.6  9 11 6.1 5.5  0 0 0 0  5 1 3.7 0.1 Sorbus aucuparia E 2009 0 25 0 5.3 2009 0 19 0 4.3  0 0 0 0 2009 0 6 0 1 Stellaria media E 2009 17 44 11.3 19.2  0 1 0 0.5 2009 17 37 11.3 16.9 2009 0 6 0 1.8 Symphoricarpos albus/hesperius N  135 157 1038.7 869.5  80 93 538.1 252.3  30 39 305.5 485.6  25 25 195.1 131.6 Tiarella trifoliata N 1968 57 40 282.8 25.2 1968 57 40 282.8 25.2  0 0 0 0  0 0 0 0 Trientalis borealis ssp. latifolia N  131 133 146.8 135.8  102 104 113.5 87.2  6 6 2.2 7  23 23 31.1 41.6 Trifolium oliganthum N  12 7 5.2 0.7  0 0 0 0  12 7 5.2 0.7  0 0 0 0 Trillium ovatum N 1968 58 46 62.4 9.4 1968 57 44 62.3 9.2  0 0 0 0  1 2 0.1 0.2 Triteleia hyacinthina or    Brodiaea coronaria N  18 13 7.8 4.5  0 0 0 0  18 13 7.8 4.5  0 0 0 0 Vaccinium parvifolium N 2009 78 118 82.6 170.8 2009 75 101 79.1 154.7  0 3 0 1.5 2009 3 14 3.5 14.6   230  All Plots (N=184)   Douglas-fir plots (N=115)   Garry oak plots (N=43)   Arbutus plots (N=26)   Species (or taxon) Origin Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Year indicator Frequency 1968 Frequency 2009 Total Cover 1968 Total Cover 2009 Veronica arvensis E 2009 8 29 3.2 5.7  0 0 0 0 2009 8 28 3.2 5.6  0 1 0 0.1 Vicia hirsuta E  17 24 10.5 6.4  0 3 0 0.3  15 18 7.9 5.4  2 3 2.6 0.7 Vicia sativa E  28 36 23.6 12.8  0 3 0 0.7  26 26 20.6 10.6  2 7 3 1.5 Vulpia bromoides E  20 17 59 51.1  0 0 0 0  20 17 59 51.1  0 0 0 0 Vulpia myuros E  6 14 20.7 44  0 0 0 0  6 14 20.7 44  0 0 0 0 Zigadenus venenosus N  10 8 3.4 2.8  0 0 0 0  8 8 2.8 2.8  2 0 0.6 0              231A.3 Comparison of phylogenetically corrected versus uncorrected tests of the relationship between plant traits and association of colonization, extirpation, and presence/absence with landscape context variables (Chapter 4)  Table A.3:  Comparison between phylogenetically corrected and uncorrected models testing the relationship between traits and the positive or negative association of (a) colonizations, (b) extirpations, (c) presence/absence in 1968, and (d) presence/absence in 2009 with landscape context variables.  For each landscape variable, the results of three models are shown: ?fisher/wilcox? refers to Fisher?s exact tests (for categorical traits) or Wilcoxon rank sum tests (for continuous traits), ?gls? refers to generalized least squares models, and ?pgls? refers to phylogenetically corrected generalized least squares models, with lambda set to its maximum likelihood value.  In each cell, ?s? refers to a significant (p<0.05) relationship in the direction predicted (see Chapter 4: Results), ?o? refers to a significant relationship opposite to the direction predicted, and blank cells indicate no significant relationship.  Grey shading indicates traits/landscape variables for which one or more models differed in terms of significance.  However, the models never differed in terms of the direction of the trend. (a) COLONIZATIONS Road density 1964 Nat. veg. area 1964 Shape index 1964 Road density 2005 Nat. veg. area 2005 Shape index 2005     TRAIT f isher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls Disturbance tolerance    s s  s s s    s s     Origin    s s  s s s    s s s    Form s s  s s  s s s s s s s s s s s s Lifespan    s s s s s s s s s s s s s   Shade tolerance    s s s s s s s s  s s s    Nutrient regime   o            o o o o Dispersal mechanism          s s s       Seed weight                   SLA    s   s s  s s s s s  s s s     232(b) EXTIRPATIONS Road density 1964 Nat. veg. area 1964 Shape index 1964 Road density 2005 Nat. veg. area 2005 Shape index 2005     TRAIT f isher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls Disturbance tolerance                 s  Origin   s                Form s s s s s  s s  s s s    s s  Lifespan s s s                Shade tolerance   s                Nutrient regime                   Dispersal mechanism           s      s  Seed weight                   SLA s s  s s                (c) PRESENCE/ABSENCE ?68 Road density 1964 Nat. veg. area 1964 Shape index 1964     TRAIT f isher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls Disturbance tolerance          Origin s s s       Form    s s s s s s Lifespan   s s s  s s  Shade tolerance       s s s Nutrient regime          Dispersal mechanism          Seed weight          SLA    s s s        233 (d) PRESENCE/ABSENCE ?09 Road density 1964 Nat. veg. area 1964 Shape index 1964 Road density 2005 Nat. veg. area 2005 Shape index 2005     TRAIT f isher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls fisher/wilcox gls pgls Disturbance tolerance       s s           Origin s s  s s s s s s s s s s s s    Form    s s s s s s s s s s s s s s s Lifespan  s  s s s s s s s s s s s s s   Shade tolerance s s s s  s s s s s s s s s s s s s Nutrient regime               o   o Dispersal mechanism              s    s Seed weight                   SLA    s s s s s s s s s s s  s s s   234Appendix B  Supplementary phytolith data (Chapters 5 and 6) B.1 Collection information and incineration data for plant tissue samples Table B.1:  I weighed each plant tissue sample before and after burning in the muffle furnace and recorded the percent of the initial weight remaining after burning.  Numbers in the ?after weight? column that are bold with an asterisk are samples where material was lost in the process; therefore the ?after weight? is in error.  Most plant tissue was a fine white ash when burning was complete, but some material remained blackened even after several hours in the muffle furnace.  The final column (?still black??) indicates this for each sample.  Collectors? initials are as follows:  JM=Jenny McCune, EM=Eric McLay, HR=Hans Roemer, CZ/RG=Carly Ziter and Rachel Germain (field technicians at the Cowichan Garry Oak Preserve). Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Abies grandis tree July 20, 2010. JM UBC campus twigs Sep-30 2010 1 0.5691 0.0096 1.69 8 500 no Abies grandis tree July 20, 2010. JM UBC campus needles Aug-16 2010 1 0.4799 0.0085 1.77 2 500 no Acer macrophyllum tree July 20, 2010. JM UBC campus seeds Jul-30 2010 1 0.3287 0.0175 5.32 2.5 500 no Acer macrophyllum tree July 20, 2010. JM UBC campus leaves Jul-30 2010 1 0.3282 0.0218 6.64 1.25 500 no Acer macrophyllum tree July 20, 2010. JM UBC campus twigs/petioles Oct-01 2010 1 0.215 0.0119 5.53 4.75 500 no Acer macrophyllum tree July 20, 2010. JM UBC campus leaves Aug-16 2010 1 0.37 0.0249 6.73 2.5 500 no Achlys triphylla herb May 14, 2011. JM Cowichan river trail flower heads Jun-12 2011 1 0.0787 0.006 7.62 3 500 yes Achlys triphylla herb May 14, 2011. JM Cowichan river trail leaves/stems Jun-12 2011 1 0.4694 0.0261 5.56 2 500 yes Achnatherum lemmonii grass May 18, 2009. JM releve plot 317 heads Mar-17 2011 1 0.1196 0.0057 4.77 5 500 yes Achnatherum lemmonii grass August 11, 2011. JM Hans Roemer's house heads Aug-17 2011 1 0.1375 0.0057 4.15 5 500 yes   235Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Achnatherum lemmonii grass August 11, 2011. JM Hans Roemer's house leaves/stems Aug-17 2011 1 0.4235 0.0091 2.15 5 500 yes Agrostis exarata grass June 19, 2011. JM Marine Dr. Vancouver heads Jul-01 2011 1 0.1131 0.0043 3.80 3 500 no Agrostis exarata grass June 19, 2011. JM Marine Dr. Vancouver leaves/stems Jul-01 2011 1 0.444 0.0209 4.71 3 500 yes Agrostis pallens grass August 11, 2011. JM Hans Roemer's house heads Aug-31 2011 1 0.0979 0.0042 4.29 5 500 no Agrostis pallens grass August 11, 2011. JM Hans Roemer's house leaves/stems Aug-31 2011 1 0.2393 0.0061* 2.55 5 500 yes Aira caryophyllea grass July 25, 2010. HR Saanich leaves/stems Sep-28 2010 2 0.1257 0.0062 4.93 15.75 500 yes Aira caryophyllea grass July 25, 2010. HR Saanich leaves/stems Sep-29 2010 1 0.1718 0.0063 3.67 7.75 500 yes Aira praecox grass July 25, 2010. HR Saanich leaves/stems Oct-22 2010 1 0.1681 0.005 2.97 7 500 no Aira praecox grass July 25, 2010. HR Saanich heads Oct-22 2010 1 0.076 0.0059 7.76 7 500 no Alnus rubra tree July 20, 2010. JM UBC campus twigs Sep-30 2010 1 0.3177 0.0075 2.36 6.75 500 no Alnus rubra tree July 20, 2010. JM UBC campus leaves Aug-19 2010 1 0.3073 0.0202 6.57 4.75 500 yes Alopecurus arundinaceus grass May 11, 2011. JM Cowichan heads Jun-15 2011 1 0.3807 0.0225 5.91 7.75 500 yes Alopecurus arundinaceus grass May 11, 2011. JM Cowichan leaves/stems Jun-15 2011 1 0.3403 0.0256 7.52 7.75 500 yes Anthoxanthum odoratum grass late June, 2010 CZ/RG Cowichan Preserve heads Jul-09 2010 1 0.1039 0.0027 2.60 1.75 500 no Anthoxanthum odoratum grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-21 2010 1 0.3027 0.0105 3.47 2 500 yes   236Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Anthoxanthum odoratum grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-29 2010 3 0.5891 0.0208 3.53 23 500 yes Arbutus menziesii tree late June, 2010 CZ/RG Cowichan Preserve leaves Jul-21 2010 1 0.6157 0.0143 2.32 0.25 500 no Arrhenatherum elatius grass May 20, 2011. JM Summit Park heads Jul-01 2011 1 0.1922 0.0116 6.04 2.5 500 no Arrhenatherum elatius grass May 20, 2011. JM Summit Park leaves/stems Jul-01 2011 1 0.3949 0.0255 6.46 2.5 500 yes Bromus carinatus grass July 25, 2010. HR Saanich leaves/stems Nov-17 2010 1 0.2592 0.0147 5.67 3.5 600 BLUE! Bromus carinatus grass July 25, 2010. HR Saanich heads Nov-17 2010 1 0.1052 0.0097 9.22 7.75 600 yes Bromus carinatus grass July 25, 2010. HR Saanich leaves/stems Nov-29 2010 1 0.312 0* 0.00 6 500 no Bromus carinatus grass July 25, 2010. HR Saanich heads Nov-29 2010 1 0.0905 0.0056 6.19 8.5 500 yes Bromus hordeaceus grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Sep-06 2010 2 0.1452 0.0077 5.30 8.5 500 no Bromus hordeaceus grass late June, 2010 CZ/RG Cowichan Preserve heads Jul-09 2010 1 0.2436 0.01 4.11 3 500 no Bromus sterilis grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Nov-11 2010 1 0.1139 0.0102 8.96 7 600 yes Bromus sterilis grass late June, 2010 CZ/RG Cowichan Preserve heads Jul-20 2010 1 0.2604 0.0121 4.65 2 500 no Bromus sterilis grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-21 2010 1 0.4924 0.0306 6.21 2 500 yes Bromus vulgaris grass July 25, 2010. HR Saanich fruits/glumes Sep-27 2010 1 0.2086 0.0119 5.70 7.25 500 no Bromus vulgaris grass July 25, 2010. HR Saanich leaves/stems Sep-27 2010 1 0.2144 0.0149 6.95 7.25 500 yes Camassia leightlinii herb August 30, 2010. EM Nursery-Cowichan bulb/roots Oct-04 2010 1 0.4069 0.0129 3.17 6.75 500 yes Camassia leightlinii herb August 30, 2010. EM Nursery-Cowichan bulb/roots Oct-04 2010 1 0.4201 0.015 3.57 6.75 500 yes   237Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Camassia quamash herb August 30, 2010. EM Nursery-Cowichan bulb/roots Oct-06 2010 1 0.4747 0.0076 1.60 6.25 500 no Camassia quamash herb August 30, 2010. EM Nursery-Cowichan bulb/roots Oct-06 2010 1 0.6706 0.0118 1.76 6.25 500 no Camassia spp. herb late June, 2010 CZ/RG Cowichan Preserve leaves/stems Nov-05 2010 1 0.1702 0.0134 7.87 7.5 500 no Camassia spp. herb late June, 2010 CZ/RG Cowichan Preserve flowers/fruit/seeds Jul-13 2010 1 0.3171 0.0165 5.20 5.5 500 no Camassia spp. herb late June, 2010 CZ/RG Cowichan Preserve stems Jul-13 2010 1 0.2175 0.0122 5.61 4.75 500 yes Camassia spp. herb late June, 2010 CZ/RG Cowichan Preserve flowers/fruit/seeds Jul-16 2010 1 0.4075 0.0224 5.50 7.25 500 no Camassia spp. herb late June, 2010 CZ/RG Cowichan Preserve flowers/fruit/seeds Jul-23 2010 1 0.8393 0.0376 4.48 5.25 500 no Carex inops sedge June 4th, 2009. JM plot 345 Mt. Douglas leaves/stems Mar-07 2011 1 0.0385 0.004 10.39 6 500 yes Carex inops sedge June 4th, 2009. JM plot 345 Mt. Douglas heads Mar-07 2011 1 0.0087 0.0015 17.24 6 500 yes Claytonia perfoliata herb May 19, 2011. JM Mt. Douglas leaves/stems Jun-03 2011 1 0.3159 0.0401 12.69 4 500 yes Claytonia perfoliata herb May 19, 2011. JM Mt. Douglas flowers/buds Jun-12 2011 1 0.1141 0.0111 9.73 2 500 yes Cornus nutallii tree August 30, 2010. EM Gabriola Island leaves Nov-05 2010 1 0.2606 0.0323 12.39 4.25 500 yes Cornus nutallii tree August 30, 2010. EM Gabriola Island flowers/fruit Feb-07 2010 1 0.2512 0.0134 5.33 5 500 yes Cornus nutallii tree August 30, 2010. EM Gabriola Island flowers/fruit Dec-10 2010 1 0.5518 0.0336 6.09 5 500 no Cornus nutallii tree August 30, 2010. EM Gabriola Island twigs Aug-25 2011 1 0.552 0.0306 5.54 4 500 yes Cynosurus echinatus grass July 25, 2010. HR Saanich leaves/stems Oct-08 2010 1 0.3129 0.0099 3.16 7.5 500 yes   238Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Cynosurus echinatus grass July 25, 2010. HR Saanich flowers/fruit/seeds Oct-08 2010 1 0.5191 0.0144* 2.77 7.5 500 no Cytisus scoparius shrub late June, 2010 CZ/RG Cowichan Preserve twigs Oct-01 2010 1 0.2232 0.0041 1.84 4.5 500 yes Cytisus scoparius shrub late June, 2010 CZ/RG Cowichan Preserve leaves Jul-20 2010 1 0.2668 0.0162 6.07 1 500 no Dactylis glomerata grass late June, 2010 JM Jericho, Vancouver heads Jul-20 2010 1 0.6616 0.0481 7.27 1.5 500 no Dactylis glomerata grass late June, 2010 JM Jericho, Vancouver leaves/stems Jul-28 2010 1 0.3679 0.0297 8.07 5.5 500 no Danthonia californica grass July 25, 2010. HR Saanich leaves/stems Oct-27 2010 1 0.2277 0.0093 4.08 7 500 yes Danthonia californica grass July 25, 2010. HR Saanich heads Oct-27 2010 1 0.1634 0.0072 4.41 7 500 no Digitaria sanguinalis grass August 6, 2011. JM UBC campus heads Aug-10 2011 1 0.3721 0.0258 6.93 6.25 500 yes Digitaria sanguinalis grass August 6, 2011. JM UBC campus leaves/stems Aug-10 2011 1 0.3042 0.0356 11.70 6.25 500 yes Elymus glaucus grass late June, 2010 CZ/RG Cowichan Preserve heads Jul-09 2010 1 0.3137 0.0149 4.75 2.75 500 no Elymus glaucus grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Nov-11 2010 1 0.1446 0.0095 6.57 3.5 600 no Elymus glaucus grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-13 2010 1 0.2114 0.0109 5.16 2.5 500 yes Elymus glaucus grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-20 2010 1 0.4238 0.0181 4.27 4.5 500 yes Elymus glaucus grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-27 2010 2 1.0341 0.0483 4.67 14 500 no Epipactis helleborine herb July 20, 2010. JM UBC campus leaves/stems Nov-03 2010 1 0.1839 0.0173 9.41 7 500 yes Epipactis helleborine herb July 20, 2010. JM UBC campus flowers Aug-19 2010 1 0.1862 0.0126 6.77 3 500 yes Epipactis helleborine herb July 20, 2010. JM UBC campus leaves/stems Aug-25 2010 1 0.179 0.0182 10.17 6 500 yes   239Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Festuca occidentalis grass July 25, 2010. HR Saanich heads Oct-13 2010 1 0.0681 0.0055 8.08 7 500 yes Festuca occidentalis grass July 25, 2010. HR Saanich leaves/stems Oct-13 2010 1 0.1444 0.0136 9.42 7 500 yes Festuca roemeri grass July 25, 2010. HR Saanich leaves/stems Jan-11 2011 1 0.2569 0.0066 2.57 6 500 yes Festuca roemeri grass July 25, 2010. HR Saanich heads Jan-11 2011 1 0.0839 0.0049 5.84 6 500 no Festuca rubra grass May 17, 2011. JM Providence Farm heads Jul-17 2011 1 0.0939 0.006 6.39 3 500 yes Festuca rubra grass May 17, 2011. JM Providence Farm leaves/stems Jul-17 2011 1 0.2359 0.0133 5.64 3 500 yes Festuca subulata grass July 25, 2010. HR Saanich heads Dec-01 2010 1 0.0777 0.0086 11.07 5.5 500 no Festuca subulata grass July 25, 2010. HR Saanich leaves/stems Dec-01 2010 1 0.2328 0.0274 11.77 7.25 500 yes Gaultheria shallon shrub May 14, 2011. JM Cowichan River Trail leaves Jun-03 2011 1 0.3636 0.0208 5.72 2.5 500 no Gaultheria shallon shrub May 14, 2011. JM Cowichan River Trail twigs Aug-25 2011 1 0.3632 0.0064 1.76 4 500 no Hedera helix shrub July 20, 2010. JM UBC campus branches Sep-03 2010 2 0.2398 0.0172 7.17 9.5 500 no Hedera helix shrub July 20, 2010. JM UBC campus petioles/stems Sep-03 2010 2 0.2116 0.0223 10.54 7.5 500 no Hedera helix shrub July 20, 2010. JM UBC campus leaves Aug-19 2010 1 0.2891 0.0225 7.78 3 500 yes Holcus lanatus grass late June, 2010 JM Jericho, Vancouver heads Jul-20 2010 1 0.4308 0.0217 5.04 0.75 500 no Holcus lanatus grass late June, 2010 JM Jericho, Vancouver leaves/stems Jul-20 2010 1 0.5105 0.0316 6.19 4.25 500 yes Holcus lanatus grass late June, 2010 JM Jericho, Vancouver leaves/stems Jul-28 2010 2 0.5366 0.0336 6.26 13 500 yes Holodiscus discolor shrub late June, 2010 CZ/RG Cowichan Preserve buds Jul-21 2010 1 0.302 0.0184 6.09 1.25 500 no   240Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Holodiscus discolor shrub late June, 2010 CZ/RG Cowichan Preserve leaves Jul-21 2010 1 0.3763 0.0299 7.95 0.75 500 no Holodiscus discolor shrub late June, 2010 CZ/RG Cowichan Preserve twigs Oct-25 2010 1 0.3628 0.0098 2.70 7.5 500 no Hordeum murinum grass May 20, 2011. JM Summit Park heads Jun-10 2011 1 0.3512 0.0171 4.87 4.5 500 yes Hordeum murinum grass May 20, 2011. JM Summit Park leaves/stems Jun-10 2011 1 0.43 0.023 5.35 4.5 500 yes Hypochaeris radicata herb July 20, 2010. JM UBC campus stems Sep-22 2010 2 0.1473 0.0143 9.71 17 500 yes Hypochaeris radicata herb July 20, 2010. JM UBC campus flower heads Sep-22 2010 2 0.1285 0.0141 10.97 11 500 no Hypochaeris radicata herb July 20, 2010. JM UBC campus leaves Sep-22 2010 1 0.1883 0.038 20.18 6 500 no Ilex aquifolium shrub July 20, 2010. JM UBC campus stems Sep-13 2010 2 0.2757 0.0139 5.04 10.25 500 no Ilex aquifolium shrub July 20, 2010. JM UBC campus leaves Aug-25 2010 1 0.342 0.0136 3.98 6 500 no Linnaea borealis herb August 1, 2010. JM Garibaldi stems, leaves, and flowers Oct-25 2010 1 0.095 0.0075 7.89 8.57.5 500 no Lolium perenne grass July 20, 2010. JM UBC campus leaves/stems Sep-13 2010 3 0.2624 0.0241 9.18 22.25 500 yes Lolium perenne grass July 20, 2010. JM UBC campus heads Aug-25 2010 1 0.1552 0.0104 6.70 5.25 500 no Lonicera hispidula shrub May 16, 2011. JM plot COW12 stems Jun-10 2011 1 0.417 0.0131 3.14 2 500 no Lonicera hispidula shrub May 16, 2011. JM plot COW12 leaves Jun-10 2011 1 0.1844 0.0144 7.81 2 500 no Luzula fastigiata herb May 14, 2011. JM Cowichan River Trail heads Jul-20 2011 1 0.0255 0.0021 8.24 2 500 no Luzula fastigiata herb May 14, 2011. JM Cowichan River Trail leaves/stems Jul-20 2011 1 0.1494 0.0144 9.64 4 500 yes Mahonia aquifolium shrub July 20, 2010. JM UBC campus stems Sep-08 2010 2 0.1807 0.008 4.43 11 500 no   241Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Mahonia aquifolium shrub July 20, 2010. JM UBC campus leaves Jul-21 2010 1 0.4808 0.019 3.95 0.5 500 no Mahonia nervosa shrub July 20, 2010. JM UBC campus twigs, fruit stalk Nov-09 2010 1 0.1548 0.0029 1.87 2.75 500 no Mahonia nervosa shrub July 20, 2010. JM UBC campus leaves Nov-09 2010 1 0.3197 0.008 2.50 3.5 500 no Mahonia nervosa shrub July 20, 2010. JM UBC campus berries Nov-09 2010 1 0.4031 0.0129 3.20 6.5 500 no Melica subulata grass late June, 2010 CZ/RG Cowichan Preserve leaves/stems Jul-01 2010 2 2.0919 0.105 5.02 13 500 yes Melica subulata grass late June, 2010 CZ/RG Cowichan Preserve bulb Jul-23 2010 1 0.9258 0.0144 1.56 3 500 no Melica subulata grass late June, 2010 CZ/RG Cowichan Preserve heads Jun-29 2010 1 0.2854 0.0138 4.84 5 500 no Oemleria cerasiformis shrub late June, 2010 CZ/RG Cowichan Preserve leaves Jul-20 2010 1 0.4977 0.0434 8.72 1.5 500 no Oemleria cerasiformis shrub late June, 2010 CZ/RG Cowichan Preserve twigs Oct-20 2010 1 0.2609 0.0066 2.53 6.25 500 no Oemleria cerasiformis shrub late June, 2010 CZ/RG Cowichan Preserve berries  Oct-20 2010 1 0.331 0.0167 5.05 6.25 500 yes Oemleria cerasiformis shrub late June, 2010 CZ/RG Cowichan Preserve fruits Jul-21 2010 2 0.6787 0.0297 4.38 10.5 500 no Paxistima myrsinites shrub August 1, 2010. JM Garibaldi leaves Sep-17 2010 1 0.131 0.006 4.58 1 500 no Paxistima myrsinites shrub August 1, 2010. JM Garibaldi small twigs Sep-20 2010 1 0.0725 0.003 4.14 1.25 500 no Phalaris arundinacea grass June 19, 2011. JM Marine Dr. Vancouver heads Jul-17 2011 1 0.2042 0.9306 455.73 3 500 yes Phalaris arundinacea grass June 19, 2011. JM Marine Dr. Vancouver leaves/stems Jul-17 2011 1 0.4664 0.0181* 3.88 3 500 yes Pinus contorta tree August 10, 2011. JM UBC campus needles Aug-25 2011 1 0.4118 0.0145 3.52 1.5 500 no Pinus contorta tree August 10, 2011. JM UBC campus twigs Aug-25 2011 1 0.3921 0.0128 3.26 1.5 500 no   242Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Plantago lanceolata herb July 20, 2010. JM UBC campus flowers/fruit  Aug-19 2010 1 0.1372 0.0074* 5.39 3 500 no Plantago lanceolata herb July 20, 2010. JM UBC campus leaves/stems Aug-19 2010 1 0.2614 0.0304 11.63 4.5 500 no Poa bulbosa grass summer 2009 JM Saanich leaves/stems Mar-16 2011 1 0.0812 0.0038 4.68 7 500 yes Poa bulbosa grass summer 2009 JM Saanich heads Mar-16 2011 1 0.0492 0.002 4.07 7 500 no Poa pratensis grass late June, 2010 JM Jericho, Vancouver leaves/stems Jul-30 2010 2 0.2467 0.0209 8.47 14.5 500 no Poa pratensis grass late June, 2010 JM Jericho, Vancouver heads Jul-15 2010 1 0.2076 0.0133 6.41 3.5 500 no Polypodium glycyrrhiza herb May 17, 2011. JM Providence Farm leaves/stems Jun-19 2011 1 0.2694 0.0102 3.79 1.5 500 yes Polystichum munitum herb July 20, 2010. JM UBC campus leaves with spores Nov-24 2010 1 0.1477 -0.0041* -2.78 3.3 500 no Polystichum munitum herb July 20, 2010. JM UBC campus stems Nov-24 2010 1 0.138 0.0063 4.57 5.5 500 yes Prunus laurocerasus shrub July 20, 2010. JM UBC campus stems Sep-08 2010 1 0.3675 0.0142 3.86 5.75 500 no Prunus laurocerasus shrub July 20, 2010. JM UBC campus leaves Jul-21 2010 1 0.8056 0.051 6.33 0.25 500 no Pseudotsuga menziesii tree July 20, 2010. JM UBC campus needles Aug-12 2010 1 0.292 0.0075 2.57 1.5 500 yes Pseudotsuga menziesii tree July 20, 2010. JM UBC campus cones Aug-16 2010 1 0.8803 0.008 0.91 2.5 500 no Pseudotsuga menziesii tree July 20, 2010. JM UBC campus cones Aug-19 2010 1 0.5933 -0.0072* -1.21 3 500 no Pseudotsuga menziesii tree July 20, 2010. JM UBC campus twigs Aug-19 2010 1 0.4381 0.0125 2.85 3.25 500 no Pseudotsuga menziesii tree July 20, 2010. JM UBC campus needles Feb-07 2011 1 0.3378 0.0092 2.72 1.5 500 no   243Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Pseudotsuga menziesii tree July 20, 2010. JM UBC campus needles Feb-07 2011 1 0.2568 0.007 2.73 1.5 500 no Pteridium aquilinum herb July 20, 2010. JM UBC campus stems Oct-16 2010 2 0.2424 0.008 3.30 12.5 500 yes Pteridium aquilinum herb July 20, 2010. JM UBC campus leaves Aug-19 2010 1 0.1359 0.0075 5.52 3.25 500 no Pteridium aquilinum herb May 14, 2011. JM Cowichan River Trail leaves Jun-19 2011 1 0.263 0.0156 5.93 1.25 500 no Pteridium aquilinum herb May 14, 2011. JM Cowichan River Trail stems Jun-19 2011 1 0.4372 0.0088 2.01 1.5 500 no Quercus garryana tree late June, 2010 CZ/RG Cowichan Preserve leaves  Jul-01 2010 1 1.5199 0.0692 4.55 1.5 500 no Quercus garryana tree August 24, 2010. JM Somenos acorns Nov-05 2010 1 0.3547 0.0085 2.40 3 500 no Quercus garryana tree late June, 2010 CZ/RG Cowichan Preserve twigs Sep-07 2010 1 0.3521 0.0214 6.08 5 500 no Rosa gymnocarpa shrub July 20, 2010. JM UBC campus stems/branches Sep-07 2010 1 0.2694 0.0041 1.52 2 500 no Rosa gymnocarpa shrub July 20, 2010. JM UBC campus leaves + small stems Aug-12 2010 1 0.154 0.0133 8.64 1 500 no Rubus discolor shrub July 20, 2010. JM UBC campus stems Nov-03 2010 1 0.2447 0.0131 5.35 7 500 no Rubus discolor shrub July 20, 2010. JM UBC campus stems Sep-08 2010 2 0.2772 0.015 5.41 11 500 no Rubus discolor shrub July 20, 2010. JM UBC campus berries/flowers Jul-21 2010 1 0.7244 0.0251 3.46 0.5 500 no Rubus discolor shrub July 20, 2010. JM UBC campus leaves Jul-21 2010 1 0.468 0.0311 6.65 0.25 500 no Rubus ursinus shrub July 20, 2010. JM UBC campus stems Sep-07 2010 1 0.189 0.0136 7.20 7 500 no Rubus ursinus shrub July 20, 2010. JM UBC campus leaves Sep-06 2010 1 0.2424 0.0165 6.81 2.25 500 yes Sanicula crassicaulis herb May 18, 2011. JM plot 288 Bear Hill leaves/stems Jun-12 2011 1 0.35 0.0571 16.31 2.5 500 yes   244Scientific name FORM Collection Date Collector(s) Location Plant Part Date burnt Year burnt # days before weight (g) after weight (g) % remaining total hours temperature (degrees C) still black? Sanicula crassicaulis herb May 17, 2011. JM Providence Farm leaves/stems Jun-12 2011 1 0.1863 0.0216 11.59 0.75 500 no Symphoricarpos albus shrub July 20, 2010. JM UBC campus fruits Sep-13 2010 1 0.2308 0.0086 3.73 3 500 no Symphoricarpos albus shrub July 20, 2010. JM UBC campus stems Aug-16 2010 1 0.2695 0.013 4.82 2.75 500 no Symphoricarpos albus shrub July 20, 2010. JM UBC campus leaves Aug-16 2010 1 0.1724 0.0136 7.89 7.25 500 yes Thuja plicata tree July 20, 2010. JM UBC campus bark Sep-03 2010 1 0.2098 0.0045 2.14 4.5 500 no Thuja plicata tree July 20, 2010. JM UBC campus twigs Sep-03 2010 1 0.2777 0.0101 3.64 3.5 500 no Thuja plicata tree July 20, 2010. JM UBC campus leaves Sep-03 2010 1 0.1733 0.009 5.19 4.5 500 no Trientalis borealis herb July 20, 2010. JM Garibaldi leaves/stems Sep-20 2010 1 0.1908 0.0185 9.70 4 500 yes Trifolium dubium herb late June, 2010 JM Jericho, Vancouver whole plant Jul-23 2010 1 0.3339 0.0211 6.32 2.25 500 no Tsuga heterophylla tree July 20, 2010. JM UBC campus twigs Oct-01 2010 1 0.3083 0.0056 1.82 9 500 no Tsuga heterophylla tree July 20, 2010. JM UBC campus needles Jul-21 2010 1 0.5675 0.014 2.47 0.25 500 no Tsuga heterophylla tree July 20, 2010. JM UBC campus cones Jul-21 2010 1 0.6836 0.0211 3.09 1 500 no Vulpia bromoides grass July 25, 2010. HR Saanich leaves/stems Dec-08 2010 1 0.1874 0.005 2.67 7.5 500 yes Vulpia bromoides grass July 25, 2010. HR Saanich heads Dec-08 2010 1 0.1082 0.0039 3.60 7.5 500 no Vulpia myuros grass July 25, 2010. HR Saanich heads Oct-15 2010 1 0.1847 0.0057 3.09 7 500 yes Vulpia myuros grass July 25, 2010. HR Saanich leaves/stems Oct-15 2010 3 0.2653 0.0121 4.56 21.8 500 yes    245B.2 Detailed phytolith extraction protocol   I used a modified version of Pearsall?s (2000) phytolith extraction procedure.  I created a datasheet so that each step of the protocol could be checked off when completed, and data on each sample could be kept in one place.  This is extremely useful when multiple samples are in the lab in various stages of completion.  The finalized protocol datasheet is shown in Fig. B.1. Here I comment on some of the things I learned during the development of this protocol.  The ?pellet? at the bottom of the centrifuge tube following centrifugation is more or less consolidated and sticky depending on what steps of the protocol have been competed.  I tested the supernatant of each step to determine where phytoliths might be escaping while pouring off the supernatant (see Appendix B.3).  I found that losses were higher following the hydrogen peroxide treatment, and therefore began removing supernatants by pipette (rather than simply pouring the supernatant off) after that step.  This adds additional time for pipetting, and may not be necessary considering the amount of phytoliths lost is still quite low compared to the millions of phytoliths found per gram of soil; however it gave me peace of mind.  The loss of phytoliths in the supernatant following organic removal is also the reason for increasing the centrifuge speed and time after this step.  During the acid treatment to remove carbonates, Pearsall (2000, p. 425) suggests bubbles should appear.  The reaction for my samples was hardly ever visible, even using strong acids.  I determined that the carbonate content of my soils is likely quite low.  However, I maintained the carbonate removal step, using 10% hydrochloric acid in a hot water bath for one hour, because I   246noticed that if this step was not completed, the hydrogen peroxide reaction tended to be more violent.  When adding hydrogen peroxide to samples from near the surface, it is wise to keep them out of the hot water bath at first, and ensure the reaction is not going to overflow the beaker.  Organic removal can require many hours, and some of my samples required 12 hours or more of hydrogen peroxide treatment, while the least organic required only two hours or less.  It is also quite difficult to be sure the organic material has been completely removed, as it can be tricky distinguishing between the small fizzy bubbles due to boiling versus reaction bubbles, particularly near the end of the reaction.  If the reaction is not fully completed, there can be visible eruptions of material from the pellet at the bottom of the centrifuge tube following the first and even the second centrifuge.  I suspect these ejections could permit escape of phytoliths into the supernatant, therefore it is important to ensure the organic removal reaction is completely finished before continuing.  For this reason, I started diluting the hydrogen peroxide with distilled water when I thought the reaction was completed and letting the sample sit overnight in order to ensure the reaction was finished.  I called the weight of the sample following removal of organic material the ?acid insoluble fraction? (AIF) after Albert et al. (1999), who gave the term to the percentage of the original soil sample remaining after acid and hydrogen peroxide treatment.  My sedimentation procedure was not very high-tech, as I did not have access to the handy ?fractionator? described by Pearsall (2000).  I washed the sample into a 600mL beaker and added   247distilled water to 11cm in height.  I then stirred the solution and allowed it to sit for 9 hours and 20 minutes.  I removed the supernatant with a large turkey baster, being careful to leave at least the bottom 1cm of solution in the beaker.  I checked the supernatant from this procedure for three samples, and none had any escaped phytoliths.  The flotation step can be quite lengthy due to centrifuge time.  My assistant, Alejandra Canela, pointed out that we could complete the flotation step repeats first, and THEN centrifuge all six tubes for 10min at once.  This of course saves a LOT of time and I should have thought of it much sooner!  A note on sodium polytungstate recycling Given that the sodium polytungstate (SPT) solution is quite dilute by the time the phytolith flotation procedure is complete, I found that I could filter this solution using gravity alone.  I used Whatman no. 50 filter paper, which filters out particles greater than about 2.7 microns.  It goes through the filter paper quite slowly, but it is easy to keep adding more solution as it filters through while waiting for the centrifuge to complete a cycle, etc.  I checked the filtered solution periodically for phytoliths and did not find any.  There was some clay remaining in the filtered solution, however.  I let this settle to the bottom for several days and then poured off the top, clay-free part for evaporation. I saved the highly clay-contaminated solution for chemical waste disposal.  Once I had filtered the solution, I slowly evaporated it over low heat until it was back near the specific gravity required. It is quite a dark yellow colour when near this 2.3 g/mL density.      248Chad Yost (phytolith researcher with PaleoInstitute, Colorado) first heats the very dilute SPT for about 1 hour and then allows the liquid to cool completely.   He has found that the clay particles will start sticking together and precipitate out of the solution. He then centrifuges the dilute SPT for about 20 minutes to further remove the remaining clay fraction before evaporating back to 2.3 g/mL.  Chad notes that recycled SPT tends to be more viscous than a fresh batch of SPT, and that the surface tension of the solution may also increase. Therefore, recycled SPT may not produce as ?clean? of a phytolith separation as fresh SPT would. To try and counteract increasing viscosity and surface tension, Chad mixes in some new SPT back to the recycled batch of SPT before re-using (Chad Yost, pers. comm.).   I found that adding some fresh SPT powder was often required in order to get the solution back to the exact 2.3g/mL density required.              249   250   251Figure B.1: (previous 2 pages) Final phytolith extraction protocol used to extract phytoliths from soil samples.  The box labeled ?check? was used to select a step for checking the supernatant for escaped phytoliths.  I rolled a 6-sided die to determine the number, which corresponds with the number in square brackets (e.g. a role of 3 indicated to check the supernatant following bleach treatment).                       252B.3 Quantification of phytolith loss during extraction  I was concerned about how many phytoliths might be ?escaping? with the supernatant poured off during the various steps of the phytolith extraction procedure (see Appendix B.2).  Therefore, I tested a number of supernatants from the different steps of the protocol for phytoliths.  I poured the supernatant into a beaker, covered it with parafilm, and left it to settle overnight.  Then I used a small pipette to extract 3-4 drops from the bottom of the beaker, where any visible material had settled.  I mounted these drops on a microscope slide and examined it under the microscope at 200x magnification to count any phytoliths present.  I usually checked two slides for each supernatant sample.  Table B.2 includes all extractions, including some from before I began removing supernatants by pipette following the organic removal step.             253Table B.2:  Results of quantification of phytoliths found in the supernatant from various stages in the phytolith extraction protocol Protocol Step (see Fig. B.1) Description Number of samples checked Minimum number of phytoliths observed Maximum number of phytoliths observed Mean number of phytoliths observed 5 first rinse of soil with H2O 12 0 1 0.21 7-9 rinse following acid treatment 14 0 46 3 11-13 rinse following bleach treatment 22 0 40 4 17-19 rinse following hydrogen peroxide  56 0 2,010 64 21-22 rinse following clay removal 14 0 14 3 24 supernatant removed after sedimentation 3 0 0 0 25-26 rinse following sedimentation 5 0 111 45 30 supernatant removed after flotation 27 0 52 10 31-32 rinse following flotation 2 1 2 1 32 rinse of remaining soil following flotation 4 0 0 0 NA check of remaining soil 3 0 3 1               254B.4 Some unknown phytolith morphotypes observed during phytolith counts  Phytoliths categorized as ?other? represented approximately 2% of counted phytoliths from all surface and core samples from Vancouver Island (Chapter 6).  Most ?other? phytoliths were hairbases (66% of ?other? phytoliths).  A few were tracheid phytoliths (5%), and Carex-type phytoliths (1%).  About 28% of ?other? phytoliths were unknown phytoliths, which I had not seen during the compilation of the reference collection (see Chapter 5), or could not definitively match with phytoliths in the reference collection.  Figure B.2 shows some of these unknown phytoliths (or potential phytoliths).                 255    256Figure B.2:  (previous page) Photos of ?unknown? phytoliths or potential phytoliths.  The number in brackets in the following descriptions is the total number of times that morphotype was observed (out of approximately 1,500 ?other? phytoliths observed in total):  (a)-(c) a type of hairbase perhaps, always brown (91); (d) and (e) elongate with pits (139); (f)-(h) sinuate elongate phytolith with scrobiculate centre (11); (i) and (j)  unknown (3); (k) short cell (type of bilobate?) (4); (l) and (m) ?bottlecaps? (3); (n) and (o) cylindrical polylobate scrobiculate (2); (p) unknown (1); (q) ?perforated? (19); (r) unknown (1); (s) potential cross type? (1); (t) unknown (1); (u) unknown (1); (v) sclereid (see Piperno, 1988, p. 54) (1); (w) Pinus-type? (1); (x) ?saddle rondel?? (1).                   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0165697/manifest

Comment

Related Items