UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Hydrogeomorphic disturbance, landscape development and riparian vegetation dynamics of an alluvial, temperate… Little, Patrick James 2011

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

Item Metadata

Download

Media
24-ubc_2012_spring_little_patrick.pdf [ 80.24MB ]
Metadata
JSON: 24-1.0072413.json
JSON-LD: 24-1.0072413-ld.json
RDF/XML (Pretty): 24-1.0072413-rdf.xml
RDF/JSON: 24-1.0072413-rdf.json
Turtle: 24-1.0072413-turtle.txt
N-Triples: 24-1.0072413-rdf-ntriples.txt
Original Record: 24-1.0072413-source.json
Full Text
24-1.0072413-fulltext.txt
Citation
24-1.0072413.ris

Full Text

Hydrogeomorphic disturbance, landscape development and riparian vegetation dynamics of an alluvial, temperate rainforest in the Carmanah River valley, British Columbia, Canada  by  Patrick James Little BSc, The University of British Columbia, 2004 BEDS, Dalhousie University, 2006   A thesis submitted in partial fulfillment of the requirements for the degree of   Master of Science  in  The Faculty of Graduate Studies (Forestry)  The University of British Columbia (Vancouver)  November 2011 © Patrick James Little 2011   ii Abstract	
   The alluvial forest of the Carmanah River valley on the west coast of Vancouver Island, British Columbia, was studied to examine the role of hydrogeomorphic disturbance in perpetuating the shifting-mosaic of habitats within this diverse ecosystem. Field-based research was complemented by a landscape-scale analysis that examined changes in the extent of specific forest types using a 70-year aerial photographic record.  Thirty-eight plots containing 4509 trees were sampled for forest structure, composition, age, understory composition, and elevation above the contemporary channel. These field data, including a vegetation chronosequence spanning over 500 years, were used to examine vegetation dynamics. Over the past century, Carmanah River has eroded nearly 30% of the alluvial forest in this study area – 65% over the past 500 years.  High magnitude floods result in diminished floodplain forest area by converting forests to channel.  This results in a subsequent course of vegetation succession and geomorphic development. Fluvial deposits are colonized by a high density of Alnus rubra accompanied by a subcanopy of Picea sitchensis individuals. As Alnus die off after 60-100 years, Picea increasingly dominates the canopy while Tsuga heterophylla regenerate within the understory.  The original cohort of Picea dies off after 300-500 years, which allows Tsuga to dominate old growth terrace forests.  Picea or Alnus do not tend to regenerate under these dense canopies and without disturbance Tsuga may remain dominant indefinitely. Understory composition was related to landform age, however species distributions at low elevation floodplain sites were also driven by elevation above thalweg and flood frequency. Light availability was also a significant factor in driving community composition. It appears that understory dynamics were linked to overstory succession and geomorphic development processes, which alter environmental conditions at the understory level. That is, species distributions are driven by dynamic environmental filters, which change as a result of biogeomorphic succession. Mature forest patches tended to persist longer than young forests. The landscape composition reflects a balance between episodes of hydrogeomorphic disturbance and periods of successional development.  Increased hydrogeomorphic disturbance rates due to climate change have the capacity to alter the landscape composition resulting in diminished mature forests.   iii Table	
  of	
  Contents	
   Abstract ....................................................................................................................................... ii	
   Table of Contents ....................................................................................................................... iii	
   List of Tables ............................................................................................................................... v	
   List of Figures ............................................................................................................................ vi	
   Acknowledgements .................................................................................................................. viii	
   Chapter 1: Introduction ................................................................................................................... 1	
   1.1 Introduction ........................................................................................................................... 1	
   1.2 The riparian system ............................................................................................................... 2	
   1.3 The complex relationship between hydrology, geomorphology, and vegetation ................. 4	
   1.4 Climate - a top down control on hydrology, geomorphology, and vegetation dynamics ..... 7	
   1.5 Study area ............................................................................................................................ 10	
   1.5 Research objectives ............................................................................................................. 14	
   Chapter 2: Channel and landscape dynamics in the alluvial forest mosaic of the Carmanah River valley, British Columbia, Canada. ................................................................................................ 16	
   2.1 Introduction ......................................................................................................................... 16	
   2.2 Methods ............................................................................................................................... 19	
   2.3 Results ................................................................................................................................. 30	
   2.4 Discussion ........................................................................................................................... 42	
   2.5 Conclusions ......................................................................................................................... 46	
   Chapter 3: Development of canopy and understory vegetation in the alluvial rainforest of the Carmanah River valley, British Columbia, Canada. ..................................................................... 47	
   3.1 Introduction ......................................................................................................................... 47	
   3.2 Methods ............................................................................................................................... 52	
   3.3 Results ................................................................................................................................. 61	
   3.4 Discussion ........................................................................................................................... 87	
     iv 3.5 Conclusions ......................................................................................................................... 97	
   Chapter 4: Conclusions ................................................................................................................. 99	
   4.1 General conclusions ............................................................................................................ 99	
   4.2 Climate change and forest management within the riparian mosaic ................................. 100	
   4.3 Future research .................................................................................................................. 104	
   References ................................................................................................................................... 106	
   Appendix A – Plot overstory composition and structure statistics. ............................................ 112	
   Appendix B – Forest type plates ................................................................................................. 116	
       v List	
  of	
  Tables	
  	
   Table 2.1– Aerial photograph and satellite image sources ........................................................... 23	
   Table 2.2 – Characteristics used for remote classification of forest patch types .......................... 24	
   Table 2.3 - Hydrologic characteristics of watersheds in the Carmanah Valley region. ................ 28	
   Table 2.4 – Landscape mosaic statistics, organized by patch type. .............................................. 36	
   Table 3.1 – Plot environmental statistics. ..................................................................................... 66	
   Table 3.2 – Overstory characteristics by forest type. .................................................................... 67	
   Table 3.3- Partial Constrained Correspondence Analysis (CCA) inertia and P-values. ............... 81	
   Table 4.1 – Succession rate vs. erosion rate for each forest type. .............................................. 102	
       vi List	
  of	
  Figures	
   Figure 1.1 – Carmanah watershed location map. .......................................................................... 11	
   Figure 1.2 – Site photos of Carmanah River and the Carmanah alluvial forest. .......................... 12	
   Figure 1.3 – Carmanah alluvial forest study area map. ................................................................ 13	
   Figure 2.1 – Aerial photographic record of the Carmanah Valley. .............................................. 25	
   Figure 2.2 – Forest patch delineation example. ............................................................................ 26	
   Figure 2.3 – Regional average yearly peak flows. ........................................................................ 29	
   Figure 2.4 - Channel avulsion of Carmanah River near Grunt’s Grove. ...................................... 31	
   Figure 2.5– Lateral channel migration of Carmanah River near Three Sisters. ........................... 31	
   Figure 2.6 - Channel configuration of Carmanah River since 1937. ............................................ 32	
   Figure 2.7 – Map of Carmanah River channel position over the past 500 years. ......................... 33	
   Figure 2.8 – Map of hydrogeomorphic disturbance frequency over the past century. ................. 34	
   Figure 2.9 - Carmanah Valley landscape mosaic maps. ............................................................... 37	
   Figure 2.10 – Riverside view of forest patches near Carmanah River. ........................................ 38	
   Figure 2.11 – Cumulative forest erosion by patch type. ............................................................... 39	
   Figure 2.12 – Areal extent of forest types over time. ................................................................... 40	
   Figure 2.13 – Areal change of forest types. .................................................................................. 41	
   Figure 2.14 – Floodplain and channel area as a function of peak flow. ....................................... 41	
   Figure 3.1 – Plot locations map. ................................................................................................... 59	
   Figure 3.2 – Establishment lag of Alnus rubra, Picea sitchensis, and Tsuga heterophylla.. ........ 60	
   Figure 3.3 – Schematic sectional diagram of transect near Grunt’s Grove. ................................. 68	
   Figure 3.4 – Basal area as a function of landform age. ................................................................. 69	
   Figure 3.5 – Relative dominance as a function of landform age. ................................................. 70	
   Figure 3.6 – Species density as a function of landform age. ........................................................ 71	
     vii Figure 3.7 – Canopy openness as a function of landform age. ..................................................... 72	
   Figure 3.8 – Tree diameter as a function of landform age. ........................................................... 73	
   Figure 3.9 – Non-metric Multidimensional Scaling ordination of overstory composition and structure. ........................................................................................................................................ 74	
   Figure 3.10 – Elevation as a function of landform age. ................................................................ 75	
   Figure 3.11 – Overstory composition and structure as a function of landform age. ..................... 76	
   Figure 3.12 – Overstory composition and structure as a function of elevation above contemporary thalweg. .................................................................................................................. 77	
   Figure 3.13 – Percent cover of common understory species as a function of landform age. ....... 82	
   Figure 3.14 – Percent cover of five ubiquitous understory species across landform ages. .......... 83	
   Figure 3.15 – Non-metric Multidimensional Scaling ordination – understory shrubs, herbs, ferns, bryophytes. .................................................................................................................................... 84	
   Figure 3.16 – Understory community composition as a function of landform age. ..................... 85	
   Figure 3.17 – Understory community composition as a function of elevation above thalweg. ... 85	
   Figure 3.18a and b – Comparison between elevation and age as drivers for understory community composition change on lower floodplain sites (elevation above thalweg < 3 m). ..... 86	
   Figure 3.19a and b – Comparison between elevation and age as drivers for understory community composition change on upper terrace sites (elevation above thalweg ≥ 3 m). .......... 86	
   Figure 4.1 – Conceptual model of the Carmanah Valley alluvial forest. ................................... 103	
      viii Acknowledgements	
   I would like to thank my supervisors, Dr. John Richardson and Dr. Younes Alila for their mentoring of my scientific education. Dr. Richardson for taking me on midway through my time in graduate school and lending invaluable support, advice, and weekly meetings that kept me on track.  And Dr. Younes Alila, for taking me on initially, supporting me and giving me the freedom to develop my own research project. I would also like to thank my committee members, Dr. Lori Daniels and Dr. Steve Mitchell, for their academic support, training, equipment lending, and helpful discussions. And, thanks to Dr. Richard Guthrie and the BC Ministry of Environment for sharing of data and for providing me with an invaluable field assistant. I could not have completed this research without my two fantastic field assistants, Sophie Harrison and Silke Lutzmann. I thank them for their excellent field effort, enthusiasm, assistance in the tree ring lab and friendship throughout a summer in the forest.  Their hard work during long days made this project possible.  I would also like to thank to friends and colleagues in the Aquatic Ecosystems Lab and the Tree Ring Lab for their help, advice, and friendship. I am very grateful for funding provided by the Pacific Institute for Climate Solutions (PICS), for the financial assistance provided by UBC Faculty of Forestry scholarships, and especially for funding from the Natural Sciences and Engineering Research Council of Canada (NSERC), including NSERC’s Canada Graduate Scholar program and Younes Alila’s NSERC Grant RGPIN 194388-11. Finally, thank you all my friends and family and especially Jamie Leathem for encouragement and support throughout.   1 Chapter	
  1: Introduction	
   1.1	
  Introduction	
   This study examines the development patterns of a dynamic riparian forest ecosystem and the history of hydrogeomorphic disturbance processes that have structured it.  The Carmanah watershed is a pristine system, which has not been exposed to forest harvesting activities. Within this watershed, the alluvial forest of the Carmanah River valley is an example of the hundreds of isolated ecosystems that develop on the floodplains of alluvial rivers within the temperate rainforest of the Pacific Coastal Ecoregion of North America, an area that extends 2000 kilometers from the northern coast of California up to Alaska. Not unlike the forests that occupy the hillslopes in this vast area, these riverine forests are subject to many natural disturbance processes including wind-throw, dwarf mistletoe outbreaks, and fire.  All of these processes may drive landscape-level forest dynamics and control vegetation composition to some extent, however, within alluvial forests, hydrogeomorphic disturbance often has the largest impact in structuring forest and landscape dynamics. This disturbance type can be briefly described as the movement of the river channel across the landscape, and it is this continuous geomorphic reworking of the floodplain over a timescale of hundreds of years that allows Sitka spruce (Picea sitchensis (Bong.) Carrière) forests to perpetuate along river corridors in the Pacific Coastal Ecoregion.  This species is semi-shade tolerant, and thus requires disturbance to re-establish. The environmental conditions created by the piecewise erosion of forests and deposition of alluvial substrates allow individuals and populations of this species to establish, grow, and in the case of the Carmanah Valley, to become the tallest trees in Canada.  A thorough understanding of the disturbance regime that governs this forest type, as well as the succession and development patterns of flora exposed to this regime, is fundamental in predicting how these   2 forests may develop as the global climate and local hydroclimate changes in the future. Research, such as this, on a pristine and highly dynamic riparian ecosystem is extremely valuable as a reference for restoration of riparian forests that have been impacted by dams, channelization, species invasions and other forces that affect similar ecosystems worldwide. 1.2	
  The	
  riparian	
  system	
   Riparian forests are among the most diverse, productive, and spatially heterogeneous ecosystems in the landscape (Naiman et al. 2005).  These alluvial forests grow as a mosaic of patches on a variety of fluvial landforms and the heterogeneity observed in these systems is attributed to a range of hydrogeomorphic disturbance processes that operate on near-stream vegetation including hydraulic shear stress (scouring), prolonged hydrologic inundation, sediment deposition, erosion, and channel migration (Hupp and Ostercamp 1996, Naiman and Décamps 1997, Naiman et al. 2010).  The development and succession of alluvial forests is highly linked to processes of landscape evolution, which are ultimately controlled by weather events and resulting floods (Latterell et al. 2006, Whited et al. 2007, Perona et al. 2009). The riparian forest is an ecotone, which lies at the interface of aquatic and terrestrial systems and is unique in the broader landscape as a transitional zone that affects and is affected by processes from both domains (Naiman and Décamps 1997).  Hydrologic systems, in general, reflect the recent climate as well as the regional geography.  Furthermore, riparian ecosystems are, definitively, those portions of the landscape affected by historical and contemporary hydrologic forces.  For this reason, these forests represent a unique part of the landscape where historical climatic patterns may be reflected in vegetation patterns (Naiman et al. 2005).  The width of the riparian zone will depend on the size of the stream, the position of the stream within the drainage network, the hydrologic regime, and the local geomorphology  (Naiman and   3 Décamps 1997).   In small, steep headwater streams the riparian area may be extremely narrow and any flood disturbance here may be severe, mechanical and of short duration (Swanson et al. 1998, Honda 2008).  Conversely, in large streams the riparian area may be recognizable as a wide band of vegetation occupying a well-developed and geomorphologically complex floodplain that is regularly inundated by lower intensity floods of long duration (Naiman and Décamps 1997).  In general the riparian area may be thought of as an ecosystem that exists at the aquatic-terrestrial interface, however the exact defining width of this zone may differ by researchers or managers.  Depending on the definition, part or all of the riparian forest may be influenced by seasonal flooding and may contribute organic matter such as wood or leaves to the stream.  In this thesis I define the riparian forest as the broad valley-bottom forest that occupies the portion of the landscape that is of fluvial origin and includes, but is not limited to, surfaces that are regularly inundated by floodwaters. Within the riparian area, forests of different age, structure and composition grow as a patchwork or mosaic, supporting a great range of habitats that sustain biodiversity in the riparian ecotone.  The diversity of patches is maintained by a naturally fluctuating hydrologic regime whereby temporary surges in river flow (floods) result in the piecewise erosion of portions of forest, providing opportunity for rejuvenation (Whited et. al. 2007, Naiman et al. 2008).  In this dynamic riparian landscape, river channels regularly shift laterally due to geomorphic processes such as lateral migration, channel avulsion, and channel switching, as well as vertically due to channel incision, channel avulsion and sediment wedge formation (Beechie et al. 2006).  Lateral migration is the process by which the river gradually erodes the banks of the channel (usually on the outer edge of a meander bend) causing the channel to move across the floodplain at rates of centimeters to tens of meters per year.  Channel avulsion is the process whereby a channel is rapidly abandoned and a new channel is formed in an area of steeper slope.  Regardless of the   4 specific process, as the channel moves across the floodplain, some geomorphic landforms are destroyed with erosion and others are created with deposition, providing opportunity for vegetation to establish on freshly scoured or newly deposited substrates (Balian and Naiman 2005, Van Pelt et al. 2006, Polzin and Rood 2006, Stolnack and Naiman 2010).  In this way a shifting mosaic of forest patches is maintained by the disturbance regime of the channel, giving rise to a diverse-valley bottom landscape. Within a single floodplain, forest patches cover a gradient of species compositions, forest structures, successional stages as well as ecological and geomorphic functions. The extent of different riparian patch types is thought to remain steady at large spatial and temporal scales, although locally the spatial distribution and extent may change dramatically over time scales of less than a century (Kalliola and Puhakka 1988, Latterell et al. 2006, Whithed et al. 2007, Naiman et al. 2010). 1.3	
  The	
  complex	
  relationship	
  between	
  hydrology,	
  geomorphology,	
  and	
   vegetation	
   The disturbance processes enforced by hydrologic regimes and geomorphological change have been shown to drive riparian vegetation patterns (e.g. Kalliola and Puhakka 1988, Latterell et al. 2006, Whited et al. 2007).  However, the distribution of riparian vegetation has also been shown to influence fluvial geomorphological processes and, over time, change the degree of hydrologic influence on the vegetation itself.  Feedbacks between landform biological composition and  geomorphic development result in a self-organizing alluvial forest-channel system (Francis et al. 2009).  This complex relationship results in a dynamic and diverse environment that must be characterized in a dynamic framework.   5 1.3.1	
  Hydrologic	
  and	
  geomorphologic	
  drivers	
  of	
  vegetation	
  distribution	
   Distinct floristic assemblages often occur on different geomorphic landforms such as point bars, channel shelves, floodplains, lower terraces, and upper terraces.  Relationships between floristic assemblages and fluvial landforms have been rigorously documented in the eastern United States (Hupp 1982, 1983, Hupp and Ostercamp 1985, 1996), and studies in the Pacific Coastal Ecoregion and elsewhere in the world have also described such patterns (e.g. Fonda 1974, Shin and Nakamura 2005, Van Pelt et al. 2006, Naiman et al. 2010).  Species and assemblages may be stratified on different fluvial landforms because they are adapted to a particular hydrologic disturbance regime associated with the landform or because they are a relic relating to a specific patch history. Several adaptations allow riparian species to thrive under a frequent disturbance regime in the riparian environment. These include the production of large numbers of wind- or water- dispersed seeds that colonize bare alluvial surfaces, resprouting after fluvial breakage and burial of the stem or roots, the ability to withstand anoxic conditions produced by long periods of flood inundation, and shade tolerance that allow proliferation on previously-colonized and infrequently-disturbed landforms (Naiman and Décamps 1997).  Many near-stream species rely on one or several of these life history attributes.  Furthermore, in riparian areas subject to seasonal drought, species may also have to rely on traits related to drought tolerance to successfully establish.  For example Salix, a common floodplain genus in rivers around the world owes its success to several of the above listed evolutionary adaptations. However, in several river systems of southern Europe, recruitment of Salix alba is declining due to increased drought stress during low flow periods (Barsoum 2002, Little et al. in prep). Many studies have related distinct riparian assemblages to their elevation above the river channel, attributing the relationship to the vegetation’s adaptive response to the hydrologic   6 regime of that landform (e.g. Teversham and Slaymaker 1976, Hall and Harcomb 1998, Chapin et al. 2002). However, by focussing on the elevational gradient without considering that this same range of landforms also represents a temporal gradient, (i.e. a function of time since substrate deposition), much of this research has not addressed the dynamic nature of riparian vegetation communities.  The elevational gradient is a measurable attribute that indirectly affects ecological diversity through mechanisms related to the frequency and duration of hydrologic inundation, drought stress tolerance and resistance to mechanical disruption, while the temporal gradient controls diversity related to successional processes.  In many river systems the joint process of fluvial landform development and vegetation succession, biogeomorphic succession, may also result in correlations between elevation or landform type and floristic assemblages, explaining observed species stratifications (Van Pelt et al. 2006, Corenblit et al. 2007). Vegetation composition may be controlled by adaptations to a regular hydrologic disturbance regime, by channel migration and geomorphic change or the interaction of these processes. 1.3.2	
  The	
  influence	
  of	
  vegetation	
  on	
  hydrology	
  and	
  geomorphology	
   Riparian vegetation may play an active geomorphological role in the maintenance and creation of the alluvial floodplains that it occupies (Gurnell 1997, Corenblit et al. 2007, 2008, 2009).  For example, near-stream forests contribute to the creation of their own substrates during times of overbank flow by slowing highly turbid water and, thus, accelerating sediment deposition and increasing floodplain accretion (Hickin 1984, Huang and Nanson 1997, Allmendinger et al. 2005, Gurnell and Petts 2006, Corenblit et al. 2008).  The hydraulic friction imposed by riparian vegetation can cause the velocity of overbank flows to drop sufficiently that large volumes of sediment are deposited on the floodplain.  This process may result in several cm/year of above-bank vertical accretion, depending on the vegetation type and hydrologic conditions (Almendinger et al. 2005). This sediment builds floodplain soils and may influence   7 species composition and successional trajectories (Francis 2006, Van Pelt et al. 2006).  Over time, this vertical floodplain growth results in sufficient elevation that the landform may be disconnected from future flood events, resulting in a radically different disturbance regime and a shift from allogenic (hydrologic) to autogenic drivers of floristic dynamics (Francis 2006). The presence of large wood in the channel system represents another way vegetation impacts geomorphic conditions.  Live trees on channel bars and dead trees in the channel influence the spatial distribution of hydrogeomorphic impacts and future germination sites. Large wood collects in log jams, which initiate channel modification processes such as gravel bar accretion, bedload transport and channel movement including dramatic channel avulsions (Hickin 1984, Abbe and Montgomery 1996, Hogan et al. 1998, Latterell and Naiman 2007, Naiman et al. 2010).  The sediment accretion behind logs in the channel may result in microsites with fine sediment texture and relatively high moisture-holding capacity that result in recruitment of certain species over others, contributing to diversity at the scale of a single bar (Barsoum 2002, Robertson and Augspurger 1999).  In addition, riparian vegetation directly influences channel form and hydraulic geometry by providing bank strength and resistance to channel widening (Bird 1993, Allmendinger et al. 2005, Eaton and Giles 2009).  Through these interactive processes riparian forests contribute to geomorphological change and have been referred to as ecosystem engineers (Gurnell and Petts 2006, Corenblit et al. 2007). 1.4	
  Climate	
  -­‐	
  a	
  top	
  down	
  control	
  on	
  hydrology,	
  geomorphology,	
  and	
  vegetation	
   dynamics	
    The hydrogeomorphic processes that shape the fluvial landscape and control vegetation establishment and dynamics generally occur during high-flow hydrologic conditions.  These are stochastic occurrences that arise as a result of a single weather event or a series of events.  In the   8 case of the Pacific Coastal Ecoregion most streams are pluvial (rainfall dominated) or pluvio- nival (mixed rain and snowmelt).  Under a pluvial regime, flood events capable of affecting floodplains and riparian forests occur as a result of extreme rainstorms, rather than as a result of rapid warming as may be the case in snow-dominated hydroclimates.  Although many studies have linked floristic composition to flood return periods, few studies have related changes in the riparian vegetation mosaic to climate characteristics.  However, two studies have extended the scope of hydrology-vegetation relationships to look at the effects of future climate scenarios or historical climate cycles (Primack 2000, Whited et al. 2007). At the reach scale, Primack (2000) calculated how a change in the duration of flood inundation, as predicted by climate change scenarios, would affect floodplain vegetation composition of a stream in the eastern United States.  In this ecosystem, flood inundation duration is thought to be a major control on riparian composition.  However, this study rests on the assumption that future flow regimes will operate within the same geomorphic parameters, which would be an invalid assumption in a natural, unconstrained, dynamic channel environment.  This is because intense floods, predicted by climate change scenarios, would likely modify the fluvial landforms on which riparian vegetation occurs. Therefore, their elevation relative to the channel would change, thereby changing the hydrologic inundation regime experienced by vegetation patches.  However, with this assumption in mind it was found that areal extent of vegetation classes would change by up to 27% under climate change scenarios. That is, because riparian compositions were presumably adapted to a particular inundation regime, a change in the duration of annual flood inundation would result in 27% of the riparian area mismatched with suitable hydrologic conditions.  This study illustrates the spatial scale of the effect climate change may have on this system and others.   9 A more robust study, at the scale of a 6 km long river corridor, examined the distribution and extent of floodplain habitat types in relation to historical hydrologic disturbance and regional climate over a 60-year period (Whited et al. 2007).  It was found that warming and cooling phases of the Pacific Decadal Oscillation resulted in different frequencies and magnitudes of critical overbank flows which were responsible for channel migration and floodplain forest erosion.  This, in turn, was reflected in the landscape as the relative abundance of each forest type changed in association with the magnitude of recent hydrologic disturbance.  Although short-term changes were observed during this 60-year timeframe, over sufficiently large temporal and spatial scales the abundance and extent of different patch types was thought to have remained in roughly constant proportions – a steady-state mosaic. Climate change and the resulting non-stationarity of disturbance regimes may result in a restructuring of this landscape mosaic, precluding the idea of a steady state on a temporal scale that is relevant for most tree species and many ecological processes.  For example, an increase in the occurrence of landscape-reconfiguring floods could result in a long-term decrease in mature forest terrace area.  This could occur if the frequency of high impact storms is increased such that the erosion rate of forest patches is greater than the rate of biogeomorphic succession from exposed floodplains to mature terrace forests. This thesis examines these two processes, biogeomorphic succession and landscape reconfiguration, to better understand the development of the floodplain forest mosaic. It is an analysis of the temporal and spatial dynamics of fluvial landforms / vegetation patches and an investigation of the joined processes of landform development and riparian forest succession. The overall objective of this study is to examine the role of the hydrogeomorphic disturbance regime on the composition and structure of the riparian landscape to better understand the possible effects of increased storm frequencies as predicted by climate change scenarios.   10 1.5	
  Study	
  area	
  	
   The Carmanah River Valley is located on the west coast of Vancouver Island in the Carmanah Walbran Provincial Park of British Columbia, Canada (48o 40’ N, 124o 41’W) (Figure 1.1). The park encompasses both the Carmanah and Walbran River basins and lies within the Coastal Western Hemlock Biogeoclimatic Zone.  The Carmanah watershed is a pristine wilderness area, unaffected by logging and road construction. It covers an area of 67 km2, mostly composed of steep forested valleys.  Small, but developed, floodplains occur along the lower length of the river and the most extensive alluvial floodplain area occurs well above the mouth of the stream. This area, where my study site is situated, is part of a hanging valley, downstream of which the river becomes dramatically steeper.  The alluvial study area is a 400-800 m wide, 3 km stretch of valley-bottom forest.  In this section, the Carmanah River is a meandering gravel-bed stream, with a 40-90 m wide active channel characterized by riffles, pools and glides (Figure 1.2). Downstream of the study area the river morphology becomes a cascade, step-pool and then bedrock canyon system crossing a set of waterfalls before reaching the Pacific Ocean.  We divided the 3 km study area into six 500 m sections for ease of communication when referencing specific reaches along the valley (Figure 1.3). The climate of this area is dominated by moisture laden air masses from the Pacific Ocean driven by westerly winds and locally influenced by steep mountainous topography. Catchment elevation ranges from 0 to 1050 meters above sea level. Orographic forcing of heavily saturated air masses results in high annual precipitation.  Annual precipitation values from the nearby well-gauged Carnation Creek, ~40 km away, are 2000-5000 mm across the watershed. Three-quarters of this precipitation falls during the fall and winter seasons (October to March); however, soils remain moist year round (Hartman and Scrivener 1990).  The hydrologic regime of these small Pacific coastal catchments is rainfall dominated with snowfall   11 occurring infrequently, and that which does fall rarely stays on the ground for extended periods (Fannin et al. 2000).  Streamflow is highly responsive to precipitation intensity (Hetherington 1988) and is related to hourly precipitation (Hetherington 1982).   Figure 1.1 – Carmanah watershed location map. The Carmanah River is located on the west coast of Vancouver Island off the south west coast of British Columbia, Canada.  The watershed boundary is shown in red, the study area boundary is shown in yellow, and the fluvial network is shown in blue. Vancouver Island North America   12  Figure 1.2 – Site photos of Carmanah River and the Carmanah alluvial forest. A) Carmanah River with alluvial forest mosaic. B) Mature Picea sitchensis individuals. C) Log jam on Carmanah River. D) Aggrading reach of Carmanah River. a b c d   13  Figure 1.3 – Carmanah alluvial forest study area map. The light area in this 2007 satellite image denotes the extent of the 400-800 m wide, 3 km stretch of alluvial forest that comprises the study area.  In this section, Carmanah River is a meandering gravel bed stream, with a 40-90 m wide active channel characterized by riffles, pools and glides. The 3 km study area was divided into six 500 m sections for ease of communication when referencing specific reaches along the valley.   14 1.5	
  Research	
  objectives	
   This research is guided by several interrelated objectives that investigate processes which structure riparian forests at multiple scales. Chapter 2 examines historical channel movement and resulting landscape dynamics at the scale of the entire floodplain along the 3 km long study reach. This information is then related to historical regional hydrologic records.  Specific objectives are stated briefly here and elaborated in Chapter 2: 1) To examine patterns of channel movement to quantify the magnitude and extent of hydrogeomorphic disturbance that has impacted the Carmanah alluvial forest in the past centuries.  2) To analyze the historical change in areal cover of several riparian forest types. 3) To investigate the relationship between the regional historical hydrologic record and changes in landscape composition. Chapter 3 examines the development of riparian forests at the patch scale in relation to biogeomorphic succession and hydrologic influence.  In this chapter I use a space for time substitution to quantify change in forest canopy composition and structure as a function of landform age.  I also examine the environmental conditions that drive understory composition and dynamics.  Specific objectives are stated briefly here and elaborated in Chapter 3: (1) To examine the driving forces that result in the large diversity of observed canopy composition and structure. (2) To examine evidence of alternative successional pathways that may affect forest composition.  (3) To examine the driving forces of understory composition. (4) To assess the applicability of a conceptual model of alluvial forest development to the Carmanah River watershed in order to refine the model in terms of basin size, hydroclimate, and other watershed characteristics that affect forest dynamics and composition.   15 A concluding Chapter 4 integrates the findings of chapters 2 and 3 and qualitatively assesses the impacts that climate change could have on the distribution and abundance of forest types and habitats in the Carmanah Valley and similar watersheds. It was my intention to investigate the Carmanah River alluvial forest as a model typical of smaller rivers along the western slope of the Pacific Coastal Ecoregion.  This research should serve to increase understanding of processes that govern the dynamics of forests in similar-sized riparian systems.  It is my hope that the descriptive and quantitative information presented in this thesis may be used as a reference for restoration efforts in similar ecosystems that have been degraded by logging, road construction, dams, channel stabilization, placer mining, agriculture or other activities. Furthermore, the overall aim of this study is to understand cycles of forest development at the landscape and patch scales in order to understand how a change in the disturbance regime may affect these forests in light of climate change.  	
     16 Chapter	
  2: Channel	
  and	
  landscape	
  dynamics	
  in	
  the	
  alluvial	
  forest	
  mosaic	
   of	
  the	
  Carmanah	
  River	
  valley,	
  British	
  Columbia,	
  Canada	
   2.1	
  Introduction	
   Floodplain forests of natural, dynamic rivers exist as a patchwork or mosaic of different forest ages and compositions. The diversity seen in these alluvial ecosystems is controlled by the interplay between biological succession and physical hydrogeomorphic disturbance (Geerling et al. 2006, Latterell et al. 2006).  The mosaic’s development and resilience results from feedbacks between physical and biological processes, which vary in accordance with physical parameters that affect the hydrogeomorphic disturbance regime and biological characteristics (Tabacchi et al. 2009).  The result is that the configuration of the landscape mosaic reflects the historical severity and frequency of hydrologic disturbance and the resilience and successional rates and trajectories of floodplain forests (Whited et al. 2007). In natural alluvial river systems, the riparian environment is a geomorphologically dynamic landscape where channels are regularly shifting laterally due to processes such as channel migration and avulsion, as well as vertically due to channel incision, avulsion and sediment wedge formation (Beechie et al. 2006).  Lateral migration is the process by which the river gradually erodes the outer banks of the channel causing the river to move across the floodplain at rates of centimeters to tens of meters per year (Latterell et al. 2006).  Channel avulsion is the process whereby a channel is rapidly abandoned and a new channel formed in an area of steeper slope, often precipitated by sediment wedge formation.  Regardless of the specific geomorphic process, as the channel moves across the floodplain, some areas of forest are impacted by erosion, and at the same time vegetation establishes opportunistically on the substrates of freshly scoured or newly deposited surfaces (Balian and Naiman 2005, Van Pelt et   17 al. 2006, Polzin and Rood 2006, Stolnack and Naiman 2010).  In this way, a shifting mosaic of forest patches is maintained by the repeated disturbance of the channel’s movement. The distribution and extent of riparian forest patch types is of interest to problems of landscape ecology, as this dynamic mosaic of patches supports a diversity of habitats that sustain biodiversity, ecosystem functions, and geomorphological integrity. Within a single floodplain, the diversity of forest types covers a gradient of species compositions, forest structures, as well as ecological and geomorphic functions.  For example, in the Pacific Coastal Ecoregion, mature terraces support an abundance of large mature conifers and are a primary source of large wood capable of initiating log jams that control river geomorphology (Fetherston et al. 1995, Latterell and Naiman 2007).  In contrast, recently formed floodplains are characterized by dense stands of deciduous pioneer tree species such as red alder (Alnus rubra Bong.), whose leaves are a primary energy source for stream ecosystems and have been shown to be linked with increased energy subsidies to downstream aquatic communities in the form of invertebrate exports (Wipfli and Musslewhite 2004, Richardson et al. 2005).  Riparian forest composition can affect aquatic community composition and it has been demonstrated that riparian forest type will influence benthic invertebrate community structure (Hernandez et al. 2005).  Furthermore, differences in type, age, density and composition of forests may result in varying degrees of resistance to erosion during flood events and can result in differences in channel geometry (Bird 1993, Anderson et al. 2004, Allmendinger et al. 2005, Beschta and Ripple 2006).  Thus, it is important to many aspects of stream ecology and geomorphology to understand the dynamics of riparian patch distributions. The areal extent of different riparian patch types is thought to remain steady at large spatial and temporal scales, however, locally, over time scales of less than a century, the spatial distribution and extent may change as a result of many factors including changes in upstream   18 sediment flux, river modification, and natural or anthropogenic climate fluctuations (Kalliola and Puhakka 1988, Latterell et al. 2006, Whited et al. 2007, Naiman et al. 2010).  Recently, great effort has been put into monitoring the current and historical biologic and hydrogeomorphic conditions of landscape mosaics along large rivers in Europe and elsewhere as a baseline that may be used for future restoration efforts (Geerling et al. 2006, Van Looy et al. 2008, Gonzalez et al. 2010).  Similar studies in more natural systems in North America have focused on large rivers to define processes and conditions of healthy riparian mosaics (Latterell et al. 2006, Whited et al. 2010.)  However, few projects have examined natural conditions within smaller river systems.  This research is necessary for understanding how processes of riparian landscape development transfer across river systems of varying sizes, for defining restoration goals in small and medium size rivers and for informing forest harvest practices that seek to mimic natural disturbance regimes.  Furthermore, this basic research is needed to define reference conditions that may serve to gauge the effects of climate change on riparian forests in which composition is linked to climate fluctuations (Primack 2000, Whited et al. 2007). Our overall objective was to examine historical channel movement and landscape dynamics within the Carmanah River valley bottom.  Specific inter-related objectives were: 1) To examine patterns of lateral channel migration and avulsion and to quantify the magnitude and extent of hydrogeomorphic disturbance over the past centuries; We hypothesize that the Carmanah floodplain experiences lower rates of hydrogeomorphic disturbance, which result in a relatively greater extent of mature forests, compared to larger rivers; 2) To analyze the change in time of areal cover of a range of riparian patch types to assess differences in the frequency and extent of hydrogeomorphic disturbance across forest types. We hypothesize that the riparian system is a steady-state mosaic, within which the relative change in area of developing forests is greater than that of older forest types; 3) To investigate the extent to which the regional historical   19 hydrologic record is linked to changes in landscape composition. Specifically, we hypothesize that in this rain-dominated watershed a negative correlation exists between recent winter storm flow and the areal extent of floodplain forests. 2.2	
  Methods	
   2.2.1	
  Site	
   See Chapter One for site description. 2.2.2	
  Air	
  photo	
  analysis	
   A 70-year chronology of air photos and high-resolution satellite images from 1937, 1952, 1963, 1965, 1969, 1987, 2001, 2007 provided an extensive dataset for spatial and temporal analysis of patch dynamics at the landscape scale (Table 2.1 and Figure 2.1). This dataset was used to examine geomorphic evolution and vegetation succession over the past century. The 2001 image was previously orthorectified by the British Columbia Ministry of Environment and was therefore used as a base for georectification of images from other flight years. Air photos were rectified in ArcGIS using recognizable individual trees, clusters of trees, and stable geomorphic features, such as old landslide scars and mountain ridges, as base points. We have examined the temporal and spatial patterns of the landscape mosaic in the valley bottom area of the Carmanah watershed. The extent of our study area is the alluvial forest area that has grown on landforms that have been formed by fluvial processes and is referred to as the historical floodplain. Using a digital elevation model, the extent of the alluvial forest was delineated using slope as an indicator for the floodplain edge. A boundary was drawn where the relatively flat valley bottom and the steeper hillslope met. This historical floodplain area includes both the active, regularly flooded contemporary floodplain as well as terraces that do not   20 experience flood inundation but are periodically eroded by the channel providing a supply of organic inputs such as large wood to the river system. Within this historical floodplain area, patch types based on forest types and their associated fluvial landforms (Latterell et al. 2006, Van Pelt et al. 2006) were delineated in ArcGIS. The nine patch types were: primary channels, secondary channels, abandoned channels, pioneer bars, developing floodplains, established floodplains, transitional fluvial terraces, mature fluvial terraces, and old-growth fluvial terraces (Latterell et al. 2006, Van Pelt et al. 2006). Patches were delineated at a fixed scale of 1:3000 for each air photo / satellite image. Vegetation textures, as well as spatial and temporal proximities to other patches, were used to classify forest areas in the delineation process (Table 2.2 and Figure 2.2). We separated the Mature Terrace category used in Latterell et al. 2006, into Mature Terraces and Old Growth Terraces.  In our study Mature Terraces were those forests that contained Picea sitchensis individuals of the pioneer cohort, aged 250-600 years, while Old Growth terraces were landforms 400+ years of age, devoid of mature Picea from the original cohort (Chapter 3). The total area of each patch type was calculated for each flight year.  Areas of each forest type were amalgamated into four broad categories, (Channel, pioneer bar and developing floodplain, established floodplain and transitional terrace, and mature and old growth terrace), and presented for each air photo year.  This amalgamation was done for ease of interpretation and to decrease error that may result from uncertainty in identifying very similar forest types. Vegetation textures in the 1937 air photo along with empirical field data of forest composition, ages, and terrace boundaries collected during July and August 2010 were used to identify and infer historical channel coverage during ~1850-1936 and ~1500-1850.  These inferences are based on a vegetation chronosequence that relates landform age to forest composition and structure (Chapter 3).  The results of this chronosequence indicate that Picea   21 sitchensis individuals only establish in areas of freshly deposited channel substrate, i.e. areas that have been exposed to hydrogeomorphic disturbance. Furthermore, this species commonly lives up to 500 years. Thus, areas that were impacted by the river channel at some point between ~1500-1850 will now contain mature Picea sitchensis individuals aged 150-500 years (Chapter 3).  The areas that were channel during ~1850-1936 were delineated as those areas visible as young forests (developing or established floodplains established 10-100 years ago according to our chronosequence) in the 1937 air photo. Channel sinuosity was calculated by first drawing a centerline along primary and secondary channels in the GIS for each air photo year.  The length of this line (channel length) was then divided by the length of the entire study area (valley length) to yield a metric of channel sinuosity.  Mean channel width was calculated as the total area of channel divide by the channel length. A spatial analysis of landscape dynamics was performed to illustrate the extent and frequency of the recent hydrogeomorphic disturbance regime. For each flight year an area of ‘change’ (CH) was defined according to: !" = !! +   !! Where A1 = the area that had been recently colonized by vegetation since the last air photo year (developing floodplain in the case of the 1937 air photo), and A2 = the area that was vegetation in the previous flight year but is now channel. This process was re-iterated such that a value, from one to seven, was calculated for over the entire landscape, indicating the minimum number of times change was recorded in each area of the landscape over the past century. Change in landscape configuration was calculated for total channel, floodplain, and terrace areas.  These broad landscape categories were used in order to assess the changes   22 between geomorphically distinct units (i.e. channel vs. floodplain vs. terrace), which are less prone to uncertainty in aerial identification than are individual forest type categories (eg. pioneer bar vs. developing floodplain vs. established floodplain etc.) The erosion rate and patch half-life was calculated for each forest type based on changes in landscape composition since 1937.  The area of each forest patch identified in 1937 was tracked over the subsequent 70 years and the fraction remaining in each air photo year was calculated relative to the total area that existed in 1937.  Exponential curves were then fitted to plots of fraction remaining over time and values of half-life, (time it takes for half the forest area to be eroded), and decay constants were calculated based on these exponential functions.     23 Table 2.1– Aerial photograph and satellite image sources Year Image source Scale/resolution Flight Number 1937 aerial photography 1:10,000 Fed 5675, 5679 1952 aerial photography 1:65,000 BC1444 1963 aerial photography 1:32,000 BC5087 1969 aerial photography 1:32,000 BC5367, BC5365 1987 aerial photography 1:70,000 BC87024 2001 Orthographic photo mosaic unknown - 2007 satellite image 2.5 m -     24 Table 2.2 – Characteristics used for remote classification of forest patch types (based on Latterell et al. 2006). See Chapter 3 for biological and geomorphological characteristics of patch types. Patch Type Visual Characteristics Primary Channel Main channel. Secondary Channel Smaller channel, which branches off from main channel. Abandoned Channel Former site of a secondary channel, colonized by sparse vegetation. Pioneer Bar A recently colonized lateral bar or island, beside or within a primary channel. Developing Floodplain An evenly and finely textured surface of vegetation, slightly coarser in texture than a Pioneer Bar. Established Floodplain An even, coarse textured surface of vegetation, older and coarser in texture than a Developing Floodplain. With large ‘billowing’ crowns that indicate mature deciduous tree presence. Transitional Terrace An evenly coarse textured surface of vegetation, older and coarser in texture than an Established Floodplain. Crowns are spaced further apart than in an Established Floodplain. Deciduous crowns are absent. Mature Terrace An uneven and very coarse textured surface of vegetation, older and coarser in texture than a Transitional Terrace. Large gaps are present and crowns are very large and spaced further apart, with smaller, closer spaced crowns in interstitial spaces. Old Growth Terrace Evenly textured surface, medium to large size crowns, with gaps present.   25  Figure 2.1 – Aerial photographic record of the Carmanah Valley. Aerial photographs are cropped to the boundary of the Carmanah alluvial forest study area. 500 750 1000 1500 20002500 125 1937 1952 1963 2007200119871969   26    Figure 2.2 – Forest patch delineation example. Delineation of forest patches is based on vegetation texture as well as relative position of the channel in time and space. This image displays a reach (Grunt’s Grove) of the Carmanah valley bottom in 1937.  Patch polygons were drawn and superimposed over an aerial photograph at a 1:3000 scale.    2.2.3	
  Hydrologic	
  analysis	
   To investigate the relationship between hydrologic disturbance and landscape dynamics, a metric of hydrologic intensity was calculated for each year.  The Carmanah River does not have a long-term hydrologic record, however it is very close to long term Water Survey of Canada gauging stations on the San Juan River (~25 km southeast), Carnation Creek (~35 km northwest), and Sarita River (~30 km northwest) (Table 2.3).  Records from these sites were analyzed to determine the relative hydrologic intensity in the region for each year of record. Bankfull discharge is a flow value that corresponds to the onset of floodplain inundation and several channel modification processes that would be the beginning of a hydrogeomorphic disturbance event (Andrews and Nankervis 1995). Therefore, a theoretical value of bankfull discharge (Qbf) was calculated for each of the known watersheds using:   27 !!" = 2.6986  ×!"#$!.!"#$ where Qbf is in m3/s and Area is watershed area in km2 Equation 2.1 is an empirically-based regional estimate based on data from several watersheds smaller than 100 km2 within the Pacific Coastal Ecoregion. The R2 value for this equation is 0.896 (Brayshaw 2011). The return period of this theoretical Qbf for each watershed differed substantially, and was likely an underestimate of the true Qbf for watersheds larger than 100 km2 (Table 2.3). So, the discharge at the 1.5 year recurrence interval (Q1.5) was used as a scale invariant estimate of a critical flow value close to true Qbf.  Q1.5 would be a reasonable threshold value that should be related to processes of overbank flow and would be consistent with other studies (Whited et al. 2007). Q1.5 should be large enough to initiate channel maintenance, bedload scouring, bank erosion, and to control floodplain development (Wolman and Miller 1960, Leopold 1992, Andrews and Nankervis 1995, Whited et al. 2007). Return Periods (RP) were calculated for each year of record for each watershed using Aquatic Informatics’ Aquarius software. The return periods were based on a Pearson Type III fitted distribution, which was used instead of actual return period in order to obtain RP values that would be comparable across the three watersheds with very different record lengths. The Pearson Type III distribution was then used to calculate Q1.5 for each watershed. We used a standardized regional average of the peak flow for each year as a Hydrologic Index (HI) for the region. The yearly maximum daily discharge (Qmax) was divided by the Q1.5 in order to standardize it between watersheds of different size. The mean of this value across the three gauged watersheds was used to create a regional average Hydrologic Index for each year according to:   28 !" = !!"#!!.! !"#$%" + !!"#!!.! !"#  !"#$ + !!"#!!.! !!"#!$%&# 3  Figure 2.3 shows HI for all years of record as well as the years of aerial photographs.  For each period of time between air photo years the HI during the year before the air photo was taken was used as a measure of the recent hydrologic disturbance. HI was correlated with the time series of spatial patch data to investigate the relationship between climate forced hydrologic disturbance and vegetation dynamics.  Specifically, relationships between HI and total channel, floodplain, and terrace areas were examined.  These broad landscape categories were used in order to assess the changes between geomorphically distinct units (i.e. channel vs. floodplain vs. terrace), which are less prone to uncertainty in aerial identification than are individual forest type categories (eg. pioneer bar vs. developing floodplain vs. established floodplain etc.)  A Pearson’s product-moment correlation analysis that related the extent of areal landscape change to indices of hydrologic disturbance severity was performed for each period of time between air photo flight years. We used P = 0.05 as a threshold for statistical significance.  Table 2.3 - Hydrologic characteristics of watersheds in the Carmanah Valley region.  San Juan Sarita Carnation Carmanah Watershed Area (km2) 580 162 10 67 Approx distance from Carmanah valley (km) 25 38 40 - Hydrologic Regime Pluvial Pluvial Pluvial Pluvial Period of Record 1959-2008 1948-2008 1973-2008 - # Years of Record 46 56 36 - Theoretical Qbf (m3/s) 200 85 12.9 47 Return Period (years) of theoretical Qbf < 1.00 < 1.00 1.65 - Q1.5 (m3/s) 536 232 11.4 -    29  Figure 2.3 – Regional average yearly peak flows. Arrows show when air photos were taken. The flow values are the average Peak Flow (Qmax) divided by Q1.5 (1.5 year return interval flow) averaged over three nearby watersheds.  2.2.4	
  Sources	
  of	
  uncertainty	
   This research is not immune to uncertainty, and this should be discussed.  Some level of uncertainty always accompanies aerial photo analysis.  For example, errors in the georectification of air photos are not uncommon due to low resolution of images and unclear base points. However, these errors are likely to be less than 5-10 m at any given point and should not affect broad scale conclusions on channel or landscape dynamics.  There is also considerable uncertainty in any sort of air photo interpretation.  There are inaccuracies in defining forest types using visual cues, especially when the forest is categorized into many, similar forest types. This could affect overall conclusions on landscape configurations, which is why we amalgamated forest types into categories when presenting overall conclusions of changes in landscape composition.  The misclassification error is substantially reduced if overall conclusions and 0 0.5 1.0 1.5 2.0 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Q m ax /Q І͘Њ Year Annual  Peak  Flow  (regional  average) 1937 1952 1963 1969 1987 2001 2007   30 comparisons are drawn between young forests vs. established/transient vs. mature/old forests rather than pioneer vs. developing vs. established vs. transient vs. mature vs. old growth.  2.3	
  Results	
  	
   2.3.1	
  Channel	
  migration	
   Over the past century, Carmanah River has eroded an area of 47.9 ha, or 28.7% of the historical floodplain.  Channel movement, including channel avulsions (Figure 2.4) and lateral migration (Figure 2.5), have shaped the floodplain geomorphology and imprinted a signature on the riparian landscape, visible as an the arrangement of forest types of varying age, composition and structure. The mean channel area was 18.8 ha or 11.2% of the entire floodplain, however, this figure ranged substantially from 17.0 ha in 1937 to 22.0 ha in 2007.  During the period of record sinuosity decreased substantially from a value of 1.73 in 1937 to 1.36 in 2007 (Figure 2.6).  The 2007 channel was the widest and straightest since 1937. Although in any given year the channel is in the process of eroding an average of only 11% of the valley floor, the proportion of mature terrace on the floodplain indicates that over the past 500 years the continuous movement of the channel has modified and reworked approximately 65% of the valley bottom (Figure 2.7).  This channel migration has acted as a substantial disturbance agent to the surrounding forest.  Riparian areas have periodically been disturbed, re-vegetated and disturbed again. Figure 2.8 displays the frequency of disturbance within the study area. While the channel has eroded ~30% of the historical floodplain in the past century, only 4.4 ha or 2.6% of the floodplain has continuously remained as a river channel during this time.   31   Figure 2.4 - Channel avulsion of Carmanah River near Grunt’s Grove. Aerial photography of the Carmanah valley floodplain illustrates channel avulsion as the meander to the west of the current main channel becomes abandoned over time.   Figure 2.5– Lateral channel migration of Carmanah River near Three Sisters. Aerial photography overlayed with channel locations through time illustrates the gradual process of lateral channel migration as a meanders gradually move downstream.   0 50 100 150 20025 Meters 1937  channel 1952  channel 1963  channel 1969  channel 1987  channel 2001  channel 2007  channel   32  Figure 2.6 - Channel configuration of Carmanah River since 1937. Several meanders have been cut off since 1937.  The 2007 channel was at its widest and straightest.    33  Figure 2.7 – Map of Carmanah River channel position over the past 500 years. Channel configurations since 1937 are based on visible channel area in georectified air photos and satellite images. Channel extent prior to 1937 is inferred based on vegetation textures of the 1937 air photo which provide an approximate time since hydrogeomorphic disturbance / channel presence (Chapter 3). The historical floodplain area is delineated by the gray outline.    34  Figure 2.8 – Map of hydrogeomorphic disturbance frequency over the past century.  The frequency of hydrogeomorphic disturbance is displayed as the number of changes in an area over the past century. Change is defined as 1) area that has recently become colonized by vegetation (release of hydrogeomorphic disturbance), plus 2) the area that was previously vegetation but has since become channel since the previous air photo (initiation of hydrogeomorphic disturbance). The historical floodplain area is delineated by the gray outline. The white area with black outline delineates areas that have remained as main channel since 1937.   35 2.3.2	
  The	
  dynamic	
  riparian	
  landscape	
  mosaic	
   The continuous reworking of the riparian landscape by the Carmanah River has resulted in a mosaic of forest patches each progressing at various stages of succession (Figure 2.9). These patches range in age, forest structure and composition (Chapter 3) and are visually identifiable, aerially by vegetation textures or on the ground in terms of composition and structure (Figure 2.10).  Patch areas ranged from less than 0.04 ha to 11.5 ha, with young patches generally smaller than old.  Older patches persisted longer than young patches on a per total area basis.  Based on the erosion rate of patches that existed in 1937, the half-life of pioneer bars was 24 years while the half-life of mature terraces was 357 years (Table 2.4, Figure 2.11).  However, in terms of absolute values, older forests are eroded as frequently or more frequently as younger forests. For instance, 6.4 ha (13%) of mature terrace forest area that existed in 1937 was eroded over the next 70 years, whereas only 4.9 ha (58%) of the original developing floodplain was eroded.  However, over this time period several hectares of developing floodplain forest were initiated and subsequently eroded. The areal cover of forest types has changed locally in many areas, however, over the scale of the entire historical floodplain and over the 70-year record, the areal extent of each forest type is proportionally fairly stable. However, there has been a slight decrease in older forest area and a slight increase in channel and developing forest area (Figure 2.12).  Our data series is too short to establish whether this trend may indicate that this landscape is in fact in a state of flux.    36 Table 2.4 – Landscape mosaic statistics, organized by patch type. Mean proportion of the floodplain area, mean area, standard deviation, minimum area and maximum area over the 70-year period of record are reported for each patch type. Decay statistics based on the erosion of patches that existed in the 1937 air photo are also reported.  For example, for the Pioneer Bar patch type, Fraction Remaining refers to the fraction of the original Pioneer Bar area that existed in 1937 that has not been eroded by the channel. Similarly, Area Eroded, refers the area of the original 1937 Pioneer bar area that has been eroded.  Half life is the time it takes for half of the total patch area to be eroded and is based on decay curves shown in Figure 2.11, as are Erosion rate K and R2. Patch Type Mean Proportion (%) Mean (ha) S.D. (ha) Min. (ha) Max. (ha) Fraction Remaining (%) Area Eroded (ha) Half Life (yrs) Erosion Rate K R2 Channel 11.2% 18.8 1.9 17.0 22.0 25%  Pioneer Bar 2.9% 4.9 2.9 1.1 8.7 17% 0.9 24 -0.028 0.59 Developing Floodplain 5.9% 9.9 0.9 8.5 11.1 42% 4.9 47 -0.015 0.71 Abandoned Channel 0.6% 1.1 1.0 0.0 2.8 55% 0.8 76 -0.0091 0.77 Established Floodplain 4.9% 8.1 1.5 6.0 11.1 56% 4.9 78 -0.0089 0.83 Transitional Terrace 9.1% 15.1 0.7 14.2 16.2 63% 5.3 97 -0.0071 0.91 Mature Terrace 29.0% 48.4 1.6 46.2 50.9 87% 6.4 357 -0.0019 0.99 Old Growth Terrace 36.2% 60.5 0.5 59.7 61.2 97% 2.0 1516 -0.00046 0.98   37  Figure 2.9 - Carmanah Valley landscape mosaic maps. The landscape mosaic is shown for each flight year, 1937-2007. 38   Figure 2.10 – Riverside view of forest patches near Carmanah River. Photo taken August 2010. Patch creation often takes place on the point bar of a meander bend – as the river erodes forests along the outer bank, the point bar is colonized.  Cohorts that have established more recently are seen in the foreground, while older patches with very different composition and forest structure are further away from the contemporary channel. Labels refer to approximate stand establishment dates based on dendrochronological sampling.  39   Figure 2.11 – Cumulative forest erosion by patch type. The proportion of area remaining since 1937 for each forest type is approximated using an exponential model. !"#"$%&'&()*+ ,-"#"&'./.0 ! !"# $ ! $! %! &! '! #! (! )! Fr ac tio n Re m ai ni ng Time (years) Developing Floodplain !"#"$%&'&/01+ ,-"#"&'2002 ! !"# $ ! $! %! &! '! #! (! )! Fr ac tio n Re m ai ni ng Time (years) Pioneer Bar !"#"$%&'&&03+ ,-"#"&'0)&( ! !"# $ ! $! %! &! '! #! (! )! Fr ac tio n Re m ai ni ng Time (years) Established Floodplain !"#"$%&'&&.(+ ,-"#"&'3&.2 ! !"# $ ! $! %! &! '! #! (! )! Fr ac tio n Re m ai ni ng Time (years) Transitional Terrace !"#"$%&'&&&2+ ,-"#"&'3.0/ ! !"# $ ! $! %! &! '! #! (! )! Fr ac tio n Re m ai ni ng Time (years) Old Growth Terrace !"#"$%&'&&(3+ ,-"#"&'30.* ! !"# $ ! $! %! &! '! #! (! )! Fr ac tio n Re m ai ni ng Time (years) Mature Terrace 40   Figure 2.12 – Areal extent of forest types over time.  2.3.3	
  Hydrologic	
  flow	
  –	
  the	
  driver	
  of	
  landscape	
  dynamics	
  along	
  the	
  river	
  corridor.	
   Although the riparian landscape mosaic is relatively stable over the long term, patch areas fluctuate dramatically at the decadal scale (Figure 2.13).  Channel area and developing forest area, (pioneer bars and developing floodplains), fluctuated the most substantially from year to year followed by established floodplains/terraces, (established floodplains and transitional terraces), and the most stable mature and old growth terraces.  The magnitude of winter peak flows of the previous year was significantly negatively correlated with the areal extent of floodplain forests (pioneer bars, developing floodplains, and established floodplains), (Pearson’s product-moment correlation P = 0.014) and is suggestively, although not significantly, positively correlated with channel area, (Pearson’s product-moment correlation P = 0.409) (Figure 2.14).  It appears that high-magnitude floods diminish the extent of floodplain forests by converting some forests to channel, while years of low-magnitude peak flows result in areas of channel being colonized by vegetation. 0 5 10 15 20 25 30 35 105 110 115 1937 1952 1963 1969 1987 2001 2007 To ta l  P at ch   A re a   (h a) Year Channel Pioneer  &  Developing   Floodplain Established  Floodplain   /  Transitional  Terrace Mature  &  Old  Growth   Terrace 100 41    Figure 2.13 – Areal change of forest types. The change in area from one flight year to the next is more pronounced in developing forests than established forests or mature/old-growth terraces. Developing forest area exhibits dramatic increases and decreases from one flight year to the next, which are negatively correlated with change in channel area.   Figure 2.14 – Floodplain and channel area as a function of peak flow. A significant negative correlation exists between the areal extent of floodplain forest and the regional peak flow magnitude of the previous winter, and an insignificant positive correlation between channel area and regional peak flow magnitude is suggested. Floodplain forests include pioneer bars, abandoned channels, developing and established floodplains. 0.8 1.0 1.2 1.4 1.6 1.8 16 18 20 22 24 26 Regional Average Qmax/Q1.5 (year -1) A re a (h a) Floodplain Forest Area Channel Area 42  2.4	
  Discussion	
   2.4.1	
  Channel	
  and	
  landscape	
  dynamics	
  	
   The observed channel migrations and reconfigurations are the product of natural fluvial processes, which have had considerable impact on landscape dynamics of the valley bottom forest.  These processes include the gradual downstream drift of meanders caused by lateral erosion at the outer bank of a meander bend, as well as dramatic channel avulsions likely initiated by large wood jams which force the channel through forested areas (Bird 1993, Latterell et al. 2006).  From a vegetation dynamics perspective, as channel configuration changes, hydrogeomorphic disturbance is initiated as the channel floods forested land, and later released, as the channel narrows or migrates away from the forest. This process may be brief (hours or days) or it may take several years to change course once a path is established through a previously forested area.  As channel expansion occurs, an area of riparian forest undergoes complete disturbance, while channel contraction results in an area of channel that becomes available to primary succession processes.  Channel avulsion and lateral migration result simultaneously in some area of forest experiencing erosion while another area experiences a release of hydrologic inundation as a reach is abandoned and made available to stand initiation processes. The mosaic configuration of the valley bottom forest is perpetuated by these disturbance processes and can be described as a cycle of change operating at the scale of decades to centuries (Latterell et al. 2006).  The erosion of mature and old growth terrace forests results in deposits of large wood within the channel system.  Large conifers endemic to these older forests act as key pieces that form the structural core of log jams, while smaller trees contributed by younger patch types fill the interstitial spaces of jams eventually resulting in walls of wood, cobbles and boulders several meters in height (Abbe and Montgomery 1996).  Large jams may hold back 43  wedges of sediment several meters thick and upwards of 100 m in length, eventually resulting in sufficiently lowering the channel gradient such that alternate paths are eroded through forest (Hogan, 2010 pers. comm.).  As new channel paths are established, stand initiation processes take place at the site of the abandoned channel. Smaller jams may not result in channel avulsion, however they may still contribute to patch initiation as pioneer bars form in the low energy wake behind them (Hickin 1984, Fetherston et al. 1995).  Pioneer bars may develop into floodplains and eventually terrace forests as riparian plants further alter the energy environment during floods, slowing sediment and propagule-laden water and forcing the deposition of sand, silt and seeds (Gurnell & Petts 2006, Corenblit et al. 2009).  These fluvial processes combined with biological succession processes result in forest patches occupying a temporal and elevational gradient of community compositions and structures. The heterogeneity of the mosaic is maintained as this cycle is interrupted at various stages of development each time the active channel moves into a forested area.  This heterogeneity results in ecological diversity in riparian and aquatic environments.  Different forest types provide preferred habitats to a range of terrestrial mammals, birds and insects at the same time as they influence aquatic habitat.  For example, the prolific Rubus spectabilis of developing floodplains provide nesting habitat for several bird species, while the abundance of Alnus rubra in this forest type also provides a large nutrient input into the stream and may be considered a primary building block of the aquatic foodchain (Wipfli and Musslewhite 2004). In this way a cyclical and self-perpetuating process of erosion and creation along the valley bottom manifests as a mosaic of forest patches which make up one of the most diverse and productive landscape types on the planet. Young patches are generally closer to the channel and are subject to a high frequency of disturbance.  Although young patches are generally smaller than old, much of this range is likely due to forest development processes that blur boundaries between patches of similar ages.  For 44  example, it is not difficult to distinguish patches of age 30 and 100 years based on aerial vegetation texture.  However, due to forest development processes, the same patches at age 330 and 400 years will have very similar forest structures and compositions and are therefore very difficult to distinguish (Chapter 3).  For this reason, older forest patches are larger in area and generally more heterogeneous in age. 2.4.2	
  Generality	
  of	
  findings	
   Our findings support a conceptual model that describes a cycle of patch development and destruction in unconfined alluvial forests of the Pacific Coastal Ecoregion (Latterell et al. 2006, Naiman et al. 2010).  The scale and hydro-climate of the Carmanah River are very different from the large, snow and glacier-fed rivers that have dominated research in this field; however, our observations suggest similar patterns of landscape development and vegetation dynamics. Our results support the existence of similar climate-driven hydrogeomorphic biophysical feedbacks operating across scales and hydro-climates (Fetherston et al. 1995, Gurnell et al. 2001, Whited et al. 2007). We found that quantitative metrics describing processes operating in large pluvial-nival rivers are comparable in the relatively small, pluvial Carmanah River.  Landform half-life was generally longer although at a similar scale to that on the Queets River of coastal Washington, except for the mature terrace category.  Latterell et al. (2006) reported a mature terrace half life of 401 years on the Queets, while Carmanah has a mature terrace half life of only 357 years. However, if our mature terrace category was amalgamated with the old growth category (half life = 1500 years), as it presumably was in the Queets study, the overall half life of older terraces would be much higher than that on larger rivers.  The longer half-life of patches and lower overall erosion rate of Carmanah River is likely a product of the reduced erosive power of this 45  smaller stream. Thus, the intensity of the riparian disturbance regime is decreased in this smaller watershed and the cycle of change is operating at a slightly slower rate.  This slower rate of change has resulted in a slightly different proportional configuration of the landscape mosaic.  The ratio of developing to established/transient to mature/old forests in the Queets river valley bottom, (1100 km2 basin area), is reported as 15:15:70, while the ratio at Carmanah Valley is approximately 11:15:74.  Presumably, the greater erosive power of the larger Queets River has resulted in a larger ratio of young to old forests. 2.4.3	
  Hydrology,	
  climate	
  and	
  landscape	
  dynamics	
  	
   The difference in landscape composition between the Queets and Carmanah valley raises interesting questions in light of climate change.  We attributed the greater proportion of young forests in the Queets to the greater watershed size, discharge and resulting erosive power of the Queets River.  That is, the greater rate of forest erosion in this system likely results in greater area for young forests to establish.  It is predicted that the frequency of large storms may increase as a result of climate change therefore more erosion can be expected.  Thus, it is possible that the relative proportion of developing to mature forests in the Carmanah valley may increase. However, further research is necessary, as the relative forest composition likely depends not only on the mean discharge of the system but also of the inter-year variability of peak flows and rate of biogeomorphic succession.  The succession rate determines the time needed to develop resilience to flood disturbance and the inter-year variability determines the development time between large and small peak flow years.   However, a regime of larger peak flows would likely restructure the system. The change in forest composition in relation to climate was examined in the snow dominated Flathead River (Whited et al. 2007).  In this watershed 30 times larger in basin area 46  and 3 times larger in floodplain width, 70% of the floodplain was restructured over the past century as opposed to 30% in the Carmanah.  In this system it was found that the Pacific Decadal Oscillation cycles affected the hydrologic regime of the river which in turn structured the relative balance of developing vs. mature forest.  Specifically, the cool phase of the PDO from 1947 to 1976 produced a peak flow regime that caused extensive habitat restructuring, resulting in a substantial decrease in mature forest area and a significant increase in the aerial extent of channel area (Whited et al. 2007).  Thus, it is reasonable to expect that future climate change and the resulting alteration of the storm regime at Carmanah will result in a restructuring of the floodplain forests here.  This restructuring could result in decreased recruitment rates of developing forests into mature forests, thus decreasing the abundance of large Picea sitchensis individuals for which the area is known. 2.5	
  Conclusions	
   The Carmanah valley bottom landscape is a shifting mosaic of forest patches.  The size, distribution, and relative proportion of forest types have changed in response to channel movement.  The areal extent of young floodplain forests is inversely related to a regional average of peak flow from the previous winter, indicating that large peak flows erode large areas of the floodplain forest while small peak flow years allow for vegetation establishment of the channel. The smaller scale of hydrogeomorphic disturbance in the Carmanah watershed has resulted in a larger proportion of mature terrace to young floodplain area relative to larger rivers governed by similar processes.  This difference has manifested the abundance of large Picea sitchensis individuals for which the Carmanah Walbran Provincial Park was founded.  The uniquely natural condition of the Carmanah River valley offers a template for rehabilitation of similar floodplain forests.  The mosaic here reflects the particular history of this river system, however the processes are transferable to other alluvial forests. 47  Chapter	
  3: Development	
  of	
  canopy	
  and	
  understory	
  vegetation	
  in	
  the	
   alluvial	
  rainforest	
  of	
  the	
  Carmanah	
  River	
  valley,	
  British	
  Columbia,	
   Canada	
   3.1	
  Introduction	
   The biologically diverse riparian landscape along an alluvial river is best described as a dynamic mosaic of forest patches.  Within this mosaic, a range of overstory forest types and understory floristic assemblages occur on fluvial landforms such as channel shelves, gravel bars, floodplains, lower terraces, and upper terraces (Hupp and Ostercamp 1985, 1996).  In geomorphically active river systems, this mosaic may be a reflection the history of river channel migration across the valley bottom as well as the contemporary hydrologic regime (Kalliola and Puhakka 1988, Van Pelt et al. 2006, Whited et al. 2007, Naiman et al. 2010, Gonzales et al. 2010, Chapter 2).  From the perspective of the riparian forest, river channel migration may simultaneously result in the disturbance of one area of forest and the provision of fresh substrate, available for vegetation colonization, elsewhere on the floodplain.  This process is referred to as hydrogeomorphic disturbance. As a patch of forest develops, post disturbance, it may be subjected to periodic, geomorphically non-destructive flooding as part of the contemporary flow regime. This process referred to as hydrologic inundation. Thus, the contemporary hydrologic regime may influence the maintenance of a vegetation community by periodically wetting soils, recharging groundwater or excluding non-adapted species during prolonged inundation or prolonged low flow conditions.  Environmental conditions change rapidly along the elevation gradient near river channels and plant stratification often reflects this gradient due to competitive advantage of one species over another resulting from specific adaptations associated with the gradient.  For example, certain species of trees, herbs, shrubs and bryophytes are adapted to low oxygen 48  conditions induced by prolonged periods underwater while others may be well-adapted to mechanical stress caused by the hydraulic force of floodwaters. In areas of frequent hydrogeomorphic activity, the joint processes of fluvial landform development and biological succession, called biogeomorphic succession, may also explain correlations between landform type and floristic assemblages (Van Pelt et al. 2006, Corenblit et al. 2007).  Young sites may be occupied by pioneer assemblages and replaced by late-seral stages as stand maturation and landform accretion occurs over tens or hundreds of years.  These successional processes result in changing local light availability, soil moisture, soil temperature and other environmental parameters (Fonda 1978).  Many combined processes manifest in the diversity seen in the riparian mosaic.  That is, species distributions and forest structure may be driven by constraints and advantages imposed by the contemporary hydrologic regime, as well as broad scale environmental filters, successional processes and dynamic geomorphological conditions.  The extent to which riparian vegetation is controlled by the hydrologic inundation regime, the hydrogeomorphic disturbance history, or the prevailing regional climate may depend on local system characteristics that affect the relative strength of each of these drivers. Interactions between the hydrogeomorphic disturbance history and the contemporary hydrologic regime result in a diverse riparian mosaic in which patches may develop along multiple successional trajectories.  These trajectories depend on the degree of hydrologic influence on the individual patch during different stages of geomorphic and biological development (Van Pelt et al. 2006).  For example, riparian forests may be influenced by a single environmental attribute (e.g., soil drainage capacity), which manifests as a diverse mosaic due to successional processes.  In the case of a Finnish alpine stream, the interaction between this single environmental variable and the process of biological succession resulted in 13 distinct assemblages or riparian types (Kalliola and Puhaka 1988).  In this example, a single 49  environmental driver (e.g., wet or dry) resulted in a complex mosaic of numerous patch types due to successional processes.  For this reason, riparian assemblages along geomorphically active rivers may be described as a function of the length of time since hydrogeomorphic disturbance. In contrast, several studies exploring the relationship between hydrology and vegetation have not accounted for the dynamic nature of vegetation communities or the dynamic geomorphic environment that these communities occupy (Teversham and Slaymaker 1976, Barnes 1978, Bedinger 1978, Bren and Gibbs 1986, Hughes 1990, Hall and Harcomb 1998, Sakai et al. 1999, Chapin et al. 2002, Friedman et al. 2006, Honda 2008).  These studies focussed on the elevational gradient of fluvial landforms without consideration that this range of landforms is also located along a temporal gradient.  The elevational gradient may drive diversity related to the frequency and duration of hydrologic inundation, however, the temporal gradient drives diversity related to successional processes. Studies that have related riparian assemblages to their elevation above the river channel without regard to geomorphological or biological dynamics of the system are incomplete.  This literature attributes the relationships between elevation and species composition to flows of specific return periods referred to as riparian maintenance flows (Chapin et al. 2002).  River systems that are both geomorphically constrained (i.e., not dynamic), and characterized hydrologically by a long period of glacially or nivally driven floodplain inundation, which acts as a strong and predictable environmental filter, could host riparian communities that are well characterized without reference to succession.  That is, if the geomorphic constraints are static and the constraints imposed by hydrologic inundation are powerful and predictable, then a static vegetation assemblage may occur and the element of time since disturbance may not be relevant. However, in geomorphically dynamic river systems, episodes of hydrogeomorphic disturbance 50  structure the landscape mosaic as periodic floods change relative landform elevations and interact with successional processes manifesting ecological complexity. In most river systems, the riparian environment is a geomorphically dynamic landscape where channels are shifting laterally (channel migration, channel avulsion) as well as vertically (channel avulsion, channel incision, and sediment wedge formation).  In streams of the Pacific Coastal Ecoregion, these geomorphic processes are often controlled, highly influenced or accelerated by log jam formation and disintegration (Abbe 2000, Brummer et al. 2006, Naiman et al. 2010).  After a log jam forms, a sediment wedge several meters deep and hundreds of meters in length may accrete over a relatively short time period (Little et al. 2010).  This sediment wedge may grow both vertically and laterally to a point that the river may be forced onto the floodplain temporarily (overbank flow) or permanently (channel avulsion).  Temporary overbank flow may result in increased hydrologic disturbance farther into the forest in the process known as chute erosion (Bird 1993), while channel avulsion will leave behind a large sediment wedge (Hogan et al. 1998).  Both these processes result in the formation of a new patch within the landscape mosaic. In some cases of channel avulsion the sediment wedge may be significantly higher in elevation relative to the new path of the river, thus colonizing vegetation will develop in dry conditions without frequent flood inundation.  In other cases, abandoned channels may be regularly inundated by silt-laden floodwaters and vegetation that develops here will be exposed to wetter conditions with more rapid inorganic soil development.  Different initial conditions may favour different communities and thus these patches may develop under alternate pathways of succession (Van Pelt et al. 2006, Naiman et al. 2010).  Thus, riparian patch composition is function of the how current hydrologic regime operates on the patch, as well as the geomorphic history of the patch. 51  The dynamics and ecology of floodplain forests of large, gravel-bed rivers in the Pacific Coastal Ecoregion have been well studied (Fonda 1974, Van Pelt et al. 2006, Naiman et al. 2010).  Research from the Queets River in Washington has led to the development of a conceptual model that describes the biogeomorphic succession of floodplain trees but does not examine understory composition (Van Pelt et al. 2006, Naiman et al. 2010).  This model describes the development of forest structure along multiple pathways toward a single old- growth fluvial terrace forest type.  Forest structure and composition are thought to depend on stand age and the geomorphic history of the landform on which a stand develops, whereby alternate trajectories of succession are determined by initial geomorphic conditions and subsequent shifts in the fluvial disturbance regimes (Van Pelt et al. 2006).  We are interested in extending the scope of this model to include the dynamics of understory species as most riparian understory research to date has focused on geomorphic position and hydrologic inundation as drivers of composition without regard to succession.  Additionally, many of the hydrologic and geomorphic controls that affect riparian forest composition and development may depend on river size and, thus, watershed scale. The Carmanah River valley is a pristine ecosystem that presents the opportunity to assess the generality of riparian vegetation dynamics literature across scales and hydroclimates. The Carmanah is a small pluvial river and likely much flashier than larger snow dominated systems such as the Queets River; thus, different species assemblages and processes may be observed. We used a vegetation chronosequence spanning more than five centuries to study the development and succession of riparian forests of the Carmanah River.  We examined both canopy and understory composition in relation to fluvial landform age and elevation above the river thalweg in order to elucidate the processes that drive forest composition and structure. Our research questions were (1) Is there evidence that canopy composition and structure is driven by 52  the contemporary hydrologic regime or is it purely a function of the time since colonization and subsequent stand dynamics? (2) Do different successional pathways exist and if so are they related to the contemporary hydrologic regime? (3) Is understory composition driven by the frequency and duration of hydrologic inundation or does composition change through time, reflecting overstory successional processes? (4) How applicable is the conceptual model of large river floodplain forests proposed by Van Pelt et al. (2006) to the Carmanah River watershed and how do specific differences in basin size, hydroclimate, and other watershed characteristics affect forest dynamics and composition? 3.2	
  Methods	
   3.2.1	
  Site	
   See Chapter One for site description. 3.2.2	
  Field	
  methods	
   Plot establishment Seven forest patch types, based on previous literature, were identified using aerial photography: pioneer bar, developing floodplain, established floodplain, transitional fluvial terrace, and mature fluvial terrace, old growth fluvial terrace (Van Pelt et al. 2006, Latterell et al. 2006, Naiman et al. 2010).  We also included a floodplain subcategory called abandoned channel.  See Chapter One for details of aerial photographic interpretation. A random, stratified subset of patches was chosen to represent the range of forest ages and types that exist within the landscape mosaic.  Thirty-eight plots were established within alluvial forests of the three-kilometer long study area along the Carmanah River corridor (Figure 3.1).  We sampled between 3 to 7 patches within each forest type (median = 5).  Rectangular plots were used instead of square or circular plots in order to capture the heterogeneity within 53  patches and to fit plots within the generally long and narrow forest patches (Van Pelt et al. 2006, Gonzalez et al. 2010).  Plots were sized according to the height of the canopy trees where older, taller patches with greater internal heterogeneity were sampled using larger plots.  Plot dimensions were established such that plot length ≥ the height of dominant canopy trees and width was one-third the length.  Plots were aligned to fit within narrow patches and were approximately parallel to the contemporary channel.  In uncommon situations where a plot of these narrow dimensions was too long for the patch being sampled, a wider and shorter plot was used.  In most plots, especially younger forests where tree height was small, the dimensions were often much longer than the dominant tree height. Plot length ranged from 4 m to 100 m and plot area ranged from 16 m2 to 3267 m2. Sampling of trees, understory and landform characteristics Within plot boundaries, the diameter at breast height (DBH) of all trees > 3 cm measured at 130 cm above the ground was recorded and species noted.  In young plots, < 30 years old, all trees < 3 cm DBH but taller than 130 cm were also counted and species noted.  In our youngest plot (age = 3 y), in which almost all trees were less than 130 cm tall, trees below this height were counted and species noted.  Plots were divided into three, square sub-plots and at the centre of each of these a 2 m x 2 m quadrat was established. Within each quadrat, percent cover values of all plants (including bryophytes) and lichens were estimated. The same researcher estimated % cover for all quadrats, enhancing consistency.  Within each quadrat four other procedures were undertaken. The depth to fluvial substrate (pure sand, gravel or cobbles) was measured; a hemispherical photograph of the above canopy was taken; the elevation relative to other quadrats was measured; and, the elevation above the channel thalweg relative to the center quadrat was measured using an Impulse Laser rangefinder (Laser Technologies, Centennial, Colorado, USA) and a measuring rod.  At the same time, we measured distance from the channel and recorded a 54  rough topographic profile from the patch to the channel. To assess the light available to understory plants, percent canopy openness was calculated from the hemispherical canopy photographs taken 1 m above ground and analyzed using Gap Light Analyzer 2.0. Apart from the plot measurements we also established a transect that crossed several forest patches as well as the channel. This ran perpendicular to the channel, and we measured location (horizontal and relative elevation), height, diameter, and species of each tree (DBH > 3 cm) within an 8 meter band and recorded a topographic profile along the transect.  Tree heights were recorded using the Impulse Laser rangefinder and the elevation profile was recorded using this tool in conjunction with a measuring rod. 3.2.2	
  Dendrochronology	
  and	
  landform	
  dating	
   The ages of 246 trees were determined using tree cores (n = 215) and disks (n = 31). Three to ten trees in the largest diameter class were sampled in each patch in order to find the oldest individual in the stand. Height of all trees sampled was also measured using an Impulse Laser rangefinder. Trees were cored as close to the ground as possible and multiple attempts were made to reach pith. However, for trees larger than ~120 cm DBH it was often not possible to reach the pith due to insufficient length of our tree boring equipment.  Samples were sanded and a binocular microscope was used to determine the number of rings from bark to pith.  In situations, (n = 39), where the core did not reach pith the number of rings from bark to the last visible ring was counted and the length of the core was measured. Total number of rings was estimated according to:   55  !! = !! + !! where Rt is the total number of rings to an estimated pith, Rc is the number of rings counted on the core, and Re is the estimated number of rings before reaching a geometric centre of the tree. !! = ! − ! /! where r is the radius of the tree at core height, l is the length of the core, and w is the average ring width over the 20 cm of the core closest to the pith. Using samples of known age we determined that this method was accurate to within 15% of true age, 80% of the time. The maximum error recorded was 32%. All tree establishment dates were calculated by adding the total number of rings to a coring height correction, which was implemented using a height – age regression equation, which was parameterized for each species.  In most patches, (n = 34 of 38), the minimum landform age was determined as the age of the oldest tree sampled on that landform (Sakai et al. 1999, Malik 2005, Van Pelt et al. 2006). Red alder, (Alnus rubra, hereafter referred to as Alnus) generally colonize bare gravel within one to ten years of deposition and in most patches, (14 out of 19), Sitka spruce (Picea sitchensis (Bong.) Carrière), hereafter referred to as Picea) established within 0 to 10 years after Alnus (Figure 3.2) (Van Pelt et al. 2006).  Therefor, the oldest individual of either of these species should provide an appropriate estimate of stand age (Van Pelt et al. 2006). Five plots were in stands that did not contain any live Alnus or Picea individuals. These sites were characterized by an abundance of large fallen wood, large diameter stumps, snags and decaying wood, and an uneven diameter and height distribution.  This evidence indicates that these were true old-growth stands in which the pioneer cohort of Picea and Alnus likely had been replaced and that the landforms occupied by these stands were older than the oldest live tree in the stand.  In these cases we used various techniques to estimate landform age.  At site 3804 we counted rings on a fallen Picea individual that had been cut for trail maintenance at 2 m above 56  the root collar.  At site 3710 we used an age-DBH regression specific to western hemlock (Tsuga heterophylla (Raf.) Sarg., hereafter referred to as Tsuga) to determine an approximate age of the largest Tsuga snag.  This method dated the snag and thus the landform as 12 years older than a neighbouring patch, on a similar terrace level, which contained veteran Picea individuals.  In 3 other patches we used environmental cues to estimate a minimum landform age.  In these sites maximum Tsuga ages were between 200-300 years, however adjacent terraces with far less decaying wood, less diversity of tree diameters and vertical canopy positions, and lower landform elevation were found to be 200+ years older.  Therefor, in order to establish a minimum landform age we added 200 to 400 years to the age of the oldest living Tsuga individual.  200 years was added to two sites which were at similar terrace elevations to adjacent ~500 year old landforms, while a conservative 400 years was added to the terrace which was ~20 m above an adjacent 420 year old terrace.  Although imprecise, these dates are useful in characterising long-term processes operating in these very old forests, as exact ages are less important than the approximate age rank relative to other plots.  We were unable to use cross- dating techniques on tree cores from dead trees in these sites due to accelerated rot and due to the complacency of these riparian species to broad-scale environmental patterns. 3.2.3	
  Data	
  analysis	
   Overstory and understory composition and structure Overstory and understory plot data were each synthesized into several metrics to describe tree and understory composition and structure.  Stem counts and DBH measurements were used to calculate stem density, mean DBH, maximum DBH, standard deviation of DBH, and basal area for each plot and for each species within each plot.  Understory composition was averaged over the three quadrats in each plot to yield an average percent cover value for each species within each plot.  Initial air photo classification of forest types and landform ages were used to 57  define more objective boundaries for forest types. Age cut-off values for each forest type were based on researchers’ interpretations of visible dissimilarities that were recognizable in the field as well as in aerial photos.  These dissimilarities are based on visual cues that indicate processes such as understory emergence, canopy recruitment, and senescence of certain species. We recognize that forest types span gradients and thus the classification into types is arbitrary, however classification is useful for illuminating broad scale patterns and for further analysis. Minimum, mean and maximum values of the above metrics were calculated for each forest type. Several of metrics (DBH, basal area, density, relative dominance) were plotted against landform age and R was used to draw spline curves that illustrate trends across landform ages. Spline curves are not intended to be used for predictive purposes as variability around curves is generally high and curves do not follow mathematical distributions.  Percent cover of several common understory species that displayed a clear relationship with landform age were plotted and smoothed spline curves were drawn to represent trends. Again variability around these curves is very high and they are intended to illustrate trends among certain species. We also plotted environmental parameters against landform age to illustrate trends of environmental dynamism.  Elevation above thalweg, light availability or canopy openness, and depth to fluvial substrate were plotted as a function of landform age. Ordination – Non-metric Multi-Dimensional Scaling (NMDS) and Partial Constrained Correspondence Analysis (Partial CCA) NMDS ordination was performed on two separate matrices – understory composition and forest canopy structure and composition using VEGAN 1.15-4 (Oksanen et al. 2009).  The canopy composition and structure NMDS was based on 16 stand-level variables: basal area, mean DBH, standard deviation of DBH, and density for each of the four tree species commonly present in this ecosystem - Alnus, Picea, Tsuga, and Pacific Silver Fir (Abies amabilis (Douglas 58  ex Louden) Douglas ex Forbes, hereafter referred to as Abies). The understory NMDS was based on the plot average percent cover of each of 88 species.  The function Envfit in VEGAN 1.15-4 was used to overlay relative directions and magnitudes of correlation between environmental variables (landform age, depth to fluvial substrate, elevation above thalweg, and canopy openness) and site position in NMDS space onto NMDS ordinations (Oksanen et al. 2009). NMDS axis 1 scores for both trees and plants were used as a metric of community composition in further analyses and were plotted as a function of landform age and landform elevation above thalweg. Elevation above thalweg and landform age were found to be highly correlated.  Thus, to ascertain the relative importance of each variable in explaining the variability of tree and understory composition among sites, partial ordination using Constrained Correspondence Analysis was performed using VEGAN 1.15-4 (Oksanen et al. 2009.)  We used combinations of landform age, elevation above thalweg, canopy openness (light), and depth to fluvial substrate to constrain the understory matrix, however, we excluded canopy openness as an environmental constraint with our tree matrix. P-values of a pseudo-ANOVA permutation test, which assesses the explanatory strength of the environmental variables, were calculated for several combinations of environmental variables included in the constrained ordination. Several iterations of partial direct gradient ordinations were implemented to assess the relative strength of the environmental variables that may be driving community composition.  59   Figure 3.1 – Plot locations map. Thirty-eight rectangular plots of varying size were located at the study area. Patches were first delineated in ArcMap and each patch was assigned a number and a forest type. Using these numbers a random stratified set of patches was selected such that each forest type was well represented.  60  Figure 3.2 – Establishment lag of Alnus rubra, Picea sitchensis, and Tsuga heterophylla. The oldest maximum age of each species is subtracted from the age of the oldest tree found within the stand. Alnus is typically the first species to colonize fluvial substrates. Picea generally colonizes soon after, usually within 1-3 years although sometimes as late as 15 or 25 years.  3.2.4	
  Sources	
  of	
  uncertainty	
   This research is not immune to uncertainty, and this should be discussed.  We encountered uncertainty in landform dating and dendrochronology.  First, large trees are inherently difficult to age. Because our borers were too short to reach pith in most of our largest 0 20 40 60 80 100 120 0 5 10 15 20 25 30 Oldest Individual (years) E st ab lis hm en t l ag  (Y ea rs ) Alnus Picea Tsuga 61  Picea individuals and we were forced to estimate the number of “missing rings” based on the growth history of each tree, which may introduce error. However, the exact age of trees does not substantially affect overall conclusions as we are examining a chronosequence of over 500 years. Thus, the variability that even 50-100 years of error introduces, especially in older forests, is not substantial when looking at long-term trends of development.  We also encountered uncertainty when dating extremely old landforms that were devoid of pioneer Picea individuals. In these cases we relied on various techniques to assess a minimum landform age. In some cases we simply estimated the approximate age based on landform and forest structure characteristics. This introduces large amounts of uncertainty, although we are confident that approximate ages are appropriate in elucidating long-term development patterns.  However, we are assuming that these sites contained Picea individuals at one point in time, which cannot be proven.  The alternate hypothesis would have to be that these sites represent an alternative pathway of succession that did not include Picea and therefore these sites are much younger than estimated. However, given the data and patterns presented in this thesis, this explanation is more unlikely than the conclusion that these sites are simply older than those sites that contain Picea.  3.3	
  Results	
   Plots were sampled within a collection of 38 forest patches that spanned a large gradient of ages, elevations above the channel, and other environmental parameters (Table 3.1). Overstory composition and structure varied tremendously across this gradient.  Stand characteristics ranged from our youngest stand of 3 year-old seedlings with a density of ~500,000 trees/ha up to 400-700 year old stands that contain trees of DBH as large as 312 cm and stand densities as low as of 300 trees/ha (Appendix A). 62  The range of stand ages is due to varying lengths of time since hydrogeomorphic disturbance took place at each site.  Carmanah River is consistently reworking the alluvial floodplain during yearly flood events. These periodic floods result in the river moving across the valley bottom in stages at a rate of tens and hundreds of years (Chapter 2).  As the channel migrates, it leaves behind a freshly deposited substrate for trees to colonize, and over time a mosaic of patches, each with distinct origins grow beside one another (Chapter 2.)  Figure 3.3 illustrates the range of forest types and the spatial relationship between these patches and the contemporary position of the channel along a transect that crosses the Carmanah River valley bottom. 3.3.1	
  Overstory	
  composition	
  and	
  structure	
  	
   Plots were categorized into forest types that generally follow an age gradient.  Results revealed an increasing trend of mean DBH, maximum DBH, basal area, and canopy height with age (Table 3.2, Appendix B).  However, old-growth terraces had lower tree size values than the younger mature terraces.  This is due to fewer large, mature Picea individuals in this forest stage. Similarly, stem density followed a decreasing trend from the youngest plots up to mature terraces but was higher in old growth terraces. We also included a subcategory of floodplain, abandoned channels, which were of similar age to developing and established floodplains. The geomorphic history of this forest type differed from other floodplain forests as these patches were located within depressions left by abandoned secondary channels. These sites were generally wetter, and closer to water table, than other floodplain sites.  Abandoned channel forests had lower stem density, mean and maximum DBH, and basal area values than floodplain forests of similar ages. The basal area of each tree species was plotted as a function of landform age to illustrate trends in forest canopy dominance across the range of forest ages (Figure 3.4).  Young forests (< 50 y) were dominated by Alnus with many Picea and rare Tsuga individuals (Figure 3.5).  In 63  plots of ~50 years of age, Picea and Alnus were similar in height (~20 m) and were co-dominant in the canopy.  After 60-70 years, the abundance of Alnus decreased, however many individual alders persist until 100 years or older.  The oldest Alnus individual we recorded was 119 years old.  In forests older than 70 years, Picea dominated the canopy and those stands of 200-300 years contained the greatest basal area of this species, 150-200 m2 / ha.  Tsuga saplings and subcanopy individuals also contributed to total basal area in stands of between 50-200 years old. The largest trees in the study area were Picea individuals of 300 years or older.  We commonly recorded heights of 70-85 m although the record for the tallest Picea individual in the Carmanah River valley and the world is 95 m (Peterson et al. 1997).  The largest Picea DBH we recorded was 312 cm.  The great girth of these trees makes them impossible to age using increment borers only.  Although we could not confirm the ages of the largest trees our oldest individual was estimated at 595 years of age, while the oldest confirmed Picea that we found was ~500 years old (480 rings counted at a chainsaw cut 2 m above the root bole).  After 300 years, stand-level Picea basal area decreased as fewer mature individuals persisted above this age (Van Pelt et al. 2006).  We observed almost no regeneration of this species within old forests. In stands of 400-500 years, Tsuga dominated the canopy and subcanopy.  These old terraces also contained a substantial component of Abies. Although we observed increases and decreases in the basal area and relative dominance of each species across the gradient of forest ages, our three most common species were present in large numbers even on our youngest site.  Initial seedling densities were extremely high (> 70,000 trees/ha) for Alnus, Picea, and Tsuga on our youngest, 3-year-old, gravel bar site, however the youngest landform that Abies occurred on was 38 years old (Figure 3.6).  Although Tsuga occurred in abundance on our youngest sites, this species did not dominate in developing forests except on decaying large wood that had been deposited during flood events.  We 64  observed a rapid decrease in density for all species from age 0 to 50.  Canopy openness also decreased with age over the first 50 years to a minimum value of 4% openness and increases after this period towards a stable but variable value of 6-8% in older forests (Figure 3.7).  From 50 - 100 years, Alnus densities continue to decrease with landform age, while Picea densities decrease less rapidly over this range.  From 100-400 years Picea density decreases while average Tsuga density is varied around a stable mean value of ~1000 trees/ha. Mean DBH increased steadily and was consistently greater than the standard deviation of DBH in forests of age 0 to 150-200 years (Figure 3.8).  This was due to the single cohort of Picea and Alnus individuals in forests of this age.  After 150-200 years, standard deviation surpassed mean DBH, due to an abundance of Tsuga regeneration at this stage. In forests older than 400 years, concurrent with the near absence of the Picea cohort, both standard deviation and mean DBH decreased as smaller diameter Tsuga dominate both the overstory and understory. Non-metric Multidimensional Scaling (NMDS) ordination In general, the variability of overstory composition and structure is well described as a function of age.  Thus, most of the variation in forest structure and composition is explained by the first NMDS axis, which is highly related to both landform age and landform elevation above thalweg (Figure 3.9).  These two environmental parameters are highly correlated (Figure 3.10). We used NMDS1 scores as a metric of overstory composition and structure to assess the relationship between overstory composition and each of these parameters.  Plotting NMDS1 against landform age and elevation reveals that landform age is a stronger driver of forest composition than is elevation above thalweg (Figure 3.11 and 3.12). Further analysis using partial Constrained Correspondence Analysis revealed that when age is accounted for, the variance explained by elevation was not significant. 65  The variation in forest composition and structure along the second NMDS axis was not strongly related to any of our measured environmental parameters.  However, abandoned channel sites, were generally slightly lower on the NMDS2 gradient relative to sites of similar ages. These sites were on or near backchannels that would be regularly inundated by floodwaters and generally had high Rubus spectabilis cover.  66  Table 3.1 – Plot environmental statistics. Several environmental parameters were calculated for each plot.  Sites are numbered using a four-digit field code.  Area refers to plot area, not patch area. Forest type abbreviations are as follows: PB = Pioneer Bar, DF = Developing Floodplain, AC = Abandoned Channel, EF = Established Floodplain, TT= Transitional Terrace, MT = Mature Terrace, OGT = Old Growth Terrace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able 3.2 – Overstory characteristics by forest type. Statistics were calculated for each forest type. n refers to the number of plots of each forest type.  !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !&$# !"# !$% !"#$%%&'()&'*$'+',- . /. /01 203 /0, 104 101 /01 505 ,0, /. 305 20. 50/ 22.4.. 5436 45,,11 / 1 . 5 / 27 / 1 4 2/ 2 /, '(#)*+,)-,$ ./001 2344 22111 . 1 4 5 . 30 . 1 / 33 3 ./ 6"7&$+*"879&#*"* :0511 .444 4.2/11 3 1 . 4 1 2 3 1 . 3 1 4 ;*)<$+9&8&,=>9?(($ 30:/4 1 53111 1 1 3 3 1 4 1 1 3 1 1 1 8%9%:#;"$<'=:##>;:)"$'*$'+',- .4 ,4 /0, 203 .03 705 101 /2 405 40. ,07 2504 2, // .,21 23./ ,... 21 . 2, /7 // 41 3 4 21 ., 2, ,7 '(#)*+,)-,$ 31/3 25. 343: 3: 3. .4 .0 .. @1 : @ 0 .@ 3. @3 6"7&$+*"879&#*"* 3231 43: 41:2 : . 33 30 5 .0 / . 0 31 3 41 ;*)<$+9&8&,=>9?(($ 2@3 1 .::2 . 1 4 : . 33 3 1 . 3 1 . ?()$>#$%>'@A)$$%:'*$'+'.- .7 6, /01 203 /0. 23 10. /6 403 .04 303 /2 27 /. 744 .11 2436 2/ 6 26 /3 /1 .1 7 3 5 22 7 24 '(#)*+,)-,$ /.2 1 544 2 1 3@ ./ .1 .5 0 / 33 0 1 3. 6"7&$+*"879&#*"* .52 /5 /44 0 . 32 3/ . 41 : @ 0 4 1 0 ;*)<$+9&8&,=>9?(($ .1 1 /5 3 1 @ @ @ @ 1 1 1 1 1 1 BCD)(:"CA%>'=:##>;:)"$'*$'+'6- 35 215 .01 /01 402 2/ /0. .6 307 30/ 605 .1 /, .6 575 .11 2/36 /, /1 .7 34 4. 2,/ 2, 5 /6 3, .7 2/3 '(#)*+,)-,$ 351 1 42/ .5 1 @4 @4 43 /: 0 2 5 35 1 /0 6"7&$+*"879&#*"* @53 .5. 034 .: 3: 40 :. 4: 3/. 3: 0 @. 42 3. 51 ;*)<$+9&8&,=>9?(($ .04 1 244 33 1 35 4@ 3@ /: 33 : 32 0 1 .4 '-"&*+'A$-"("* ./ 1 :2 0 1 .4 3@ 0 .4 / / / 1 1 3 E&)$C"D"#$):'E%&&)@%'*$'+'3- 2/5 /23 .04 /0, 404 /. ,06 44 602 30/ 210/ ,7 ,/ 36 621 411 216, .6 /3 46 23. 2.5 /17 42 /3 45 262 6. /25 '(#)*+,)-,$ 02 1 453 2 1 4/ 44 / /3 : 3 5 . 1 3. 6"7&$+*"879&#*"* 35. 3:/ .43 05 @4 334 3/4 345 3:5 4: 4. @/ 3@1 /3 305 ;*)<$+9&8&,=>9?(($ @.1 3:3 035 .1 5 43 31@ :1 .10 .@ 3@ @@ .0 5 /2 '-"&*+'A$-"("* 33 1 ./ : 1 3@ 3: @ 40 33 : 32 1 1 3 F)DG&%'E%&&)@%'*$'+'6- ..1 ,5, 40/ /07 22 4. 24 64 30. ,0/ 60. 64 ,5 7. ,/1 2// 6,7 4, /, 5. /32 /2. .2/ ,3 44 73 272 21, /26 6"7&$+*"879&#*"* /3 3. 3@/ 3:/ 50 .34 .:3 .34 43. 22 .5 330 315 42 303 ;*)<$+9&8&,=>9?(($ @4@ :@ :@0 4. 3: /: 3:4 333 ..@ 4: .@ @2 :: .: 33/ '-"&*+'A$-"("* 41 1 33: 3: 1 :3 @2 0 31. 3@ : 41 3 1 / H:>'I&#JDA'E%&&)@%'*$'+',- .67 363 70/ .0/ // .1 2. ,6 ,0. .03 605 ,4 ,2 ,5 222, .56 /2.7 27 6 /7 223 7/ 24, /. 27 /5 37 .1 222 6"7&$+*"879&#*"* 34 1 @@ 3 1 4 4 . @ 3 1 3 1 1 1 ;*)<$+9&8&,=>9?(($ 52/ 415 .15. .1 : 43 310 0. 3@@ .@ 3: .0 // .5 25 '-"&*+'A$-"("* 3.3 1 ./0 .0 1 00 0: 3: 3@/ 4@ : 03 34 1 4. ;9$(B&<+CAD E(&FG+'-=F&+ '<&+C?&$,*D +HIJ+C7AD+C8,&&*+K+9$DJ&"<98+CAD+L>&##&**+CMDN()F"$(+O)-*8G+C7AD !$%"A)A+ O8G+H&FG+ I$*$(+',&$ +CA.K9$DHIJ+C7ADHIJ+C7AD P$#QR=,A+ S$#=>?+ !$%G+;,&&+ O8&A+H&#*"8? !&$#H&>89+8=+ 68   Figure 3.3 – Schematic sectional diagram of transect near Grunt’s Grove.  Tree heights, diameters, species and locations were measured along an 8 m wide transect that runs through different forest types and ages. Heights and locations are drawn to scale along this topographic profile. Stand age is used as a minimum estimate of the time since the channel last ran through that location and deposited the fluvial substrate on which the stand has grown.  This section of the Carmanah River valley bottom has experienced frequent channel avulsion (channel switching) over the past several hundred years.  Channel Pioneer  Bar Established  Floodplain Transitional  Terrace Mature  Terrace Old  Growth  Terrace 1200-1300 ~1593 ~1816~19351920-1950 1800-1850 2010 2000-07 1450-1550 60 50 40 30 0 m 10 20 70 Picea  sitchensisAlnus  rubra Abies  amabilisTsuga  heterophylla Snag  or  stump 69   Figure 3.4 – Basal area as a function of landform age. Smoothed spline curves illustrate changes in overstory structure and composition as a function of landform age. Year 0 refers to the first year of colonization on a freshly deposited fluvial landform. Young forests are dominated by Alnus with a component of Picea in the understory and subcanopy. Basal area increases rapidly over the first 250 years and Picea dominates riparian forests of 60-400 years. After 250-350 years basal area decreases due to fewer large Picea individuals. The oldest forests are dominated by Tsuga with a component of Abies interspersed in the canopy.  0 100 200 300 400 500 600 700 0 50 10 0 15 0 20 0 Landform  Age  (years) B as al   A re a   (m 2 /h a) Red  Alder  (Alnus  rubra) Sitka  Spruce  (Picea  sitchensis) Western  Hemlock  (Tsuga  heterophylla) Pacific  Silver  Fir  (Abies  amabilis) Channel Pioneer  Bar Developing  Floodplain Established  Floodplain Transitional  Terrace Mature  Terrace Old  Growth  Terrace 70   Figure 3.5 – Relative dominance as a function of landform age. Smoothed spline curves illustrate trends of each species’ basal area as a percentage of total basal area.    0 100 200 300 400 500 600 700 0 .2 .4 .6 .8 1 Landform  Age  (Years) R el at iv e   D om in an ce   (B as al   A re a   sp p   /  B as al   A re a   To ta l) Alnus Picea Tsuga Abies 71   Figure 3.6 – Species density as a function of landform age. Hand drawn curves illustrate trends of each species density across landform ages.  High initial densities of Alnus, Picea and Tsuga decrease rapidly on sites younger than 50 years old. Although Tsuga densities were very high on the youngest (age = 3 years) site, only 1% of this density exists in 10-20 year old sites. The density of Tsuga stays relatively constant over of age 50-700 while Alnus and Picea densities continue to decrease with age. Abies are rare for the first several hundred years but increase in density in mature forests.  0 100 200 300 400 500 600 700 10 10 0 10 00 10 00 0 10 00 00 Landform  Age D en si ty   (t re es /h a) Tsuga Picea Alnus Abies 72   Figure 3.7 – Canopy openness as a function of landform age. Canopy openness decreases rapidly over the first 50 years and then increases to a level of 6-8% openness that persists into older forests.  The line drawn is a smoothed spline curve shown to illustrate this trend.  0 50 100 150 200 5 10 15 20 Landform  Age  (years) C an op y   O pe nn es s   (% ) 73   Figure 3.8 – Tree diameter as a function of landform age. Smoothed spline curves illustrate trends in diameter across landform ages.  Circles represent the mean diameter of all stems >3 cm DBH for each site, while squares represent the standard deviation (SD) of DBH.  Over the first 150 years the mean is greater than the SD, reflecting the single cohort of Picea and Alnus.  After 150 years, SD is greater than mean, resulting from a high density of Tsuga saplings and subcanopy individuals.  After 300-400 years, mature Picea individuals are rare and both standard deviation and mean DBH decrease.  0 100 200 300 400 500 600 700 0 20 40 60 80 Landform  Age D ia m et er   (c m ) Mean Standard  Deviation 74   Figure 3.9 – Non-metric Multidimensional Scaling ordination of overstory composition and structure. The ordination is based on four structural variables (Basal Area, Density, mean DBH, and the standard deviation of DBH) for four common tree species – Alnus, Picea, Tsuga and Abies. (16 variables total). Each plot is positioned in NMDS space and color coded according to forest type.  Vectors of environmental variables illustrate general relationships between plots and depth (to fluvial substrate), elevation (above thalweg), and age (of landform). The length of the arrow indicates the degree of correlation between the environmental gradient and relative position of sites on the ordination. Variability in forest structure along the NMDS1 axis is driven mainly by an environmental gradient related to age and elevation. Abandoned  Channel Developing  Floodplain Established  Floodplain Transitional  Terrace Mature  Terrace Old  Growth  Terrace Pioneer  Bar -­1.5 -­1.0 -­0.5 0.0 0.5 1.0 -­1 .0 -­0 .5 0. 0 0. 5 1. 0 NMDS1 N M D S 2 age ele vat ion depth 75   Figure 3.10 – Elevation as a function of landform age. Elevation above contemporary thalweg is correlated with landform age.  0 100 200 300 400 500 600 700 1 2 3 4 5 6 10 11 22 23 Landform  Age  (years) E le va tio n   ab ov e   co nt em po ra ry   th al w eg   (m ) 76   Figure 3.11 – Overstory composition and structure as a function of landform age. NMDS1 scores are used as a metric of overstory composition and structure. This metric is based on a combination of basal area, stem density, mean DBH, and the standard deviation of DBH for the four common tree species.  0 100 200 300 400 500 600 700 -­1 .0 -­0 .5 0. 0 0. 5 1. 0 Forest  structure  as  a  function  of  Age Landform  Age  (years) N M D S   a xi s   1   sc or e   (fo re st   s tru ct ur e) 77   Figure 3.12 – Overstory composition and structure as a function of elevation above contemporary thalweg. NMDS1 scores are used as a metric of overstory composition and structure. This metric is based on a combination of basal area, stem density, mean DBH, and the standard deviation of DBH for the four common tree species.   	
   1 2 3 4 5 6 10 11 22 23 -­1 .0 -­0 .5 0. 0 0. 5 1. 0 Forest  structure  as  a  function  of  Elevation Elevation  above  contemporary  thalweg  (m) N M D S   a xi s   1   sc or e   (fo re st   s tru ct ur e) 78  3.3.2	
  Understory	
  vegetation	
  composition	
   Eighty-eight taxa of shrubs, herbs, ferns, grasses, tree seedlings, bryophytes and lichens were identified in understory quadrats within our 38 forest plots.  We identified most to the species level although grasses and a single bryophyte taxa were identified only to family level. Of these 88 taxa, 38 species were present on at least 5 sites. Individual species percent cover may reflect a response to successional processes relating to forest composition, adaptations related to frequency of hydrologic inundation, light availability, microsite differences such as soil moisture and nutrient availability or a combination of all of these (Figure 3.13).  Several species were present only at the youngest or oldest sites while others followed trends of increasing and decreasing percent cover over the gradient of forest ages (Figure 3.14).  Other species did not respond predictably to any of the four environmental factors that we measured.  For example, the percent cover of one of the most common species, Rubus spectabilis, did not show a strong relationship to landform age, elevation above thalweg, light availability or depth to fluvial substrate.  To elucidate patterns in community composition related to environmental drivers we used NMDS ordination and Partial Constrained Correspondence Analysis to assess the variation across sites.  NMDS ordination revealed that the primary axis of compositional variation, NMDS1, was related to landform age and elevation above the thalweg, while the second NMDS axis was related to light availability and less strongly to forest floor depth (Figure 3.15). According to the ranked scaling techniques used by NMDS we interpret that landform age was the most significant variable in determining community composition. Partial Constrained Correspondence Analysis indicated that the majority of the compositional variation accounted for by environmental parameters was related to landform age 79  and percent canopy openness (light availability). These two factors accounted for 19% of the total inertia (variance) within the dataset as a whole, while depth to fluvial substrate and elevation above thalweg were minor factors in explaining compositional variation (Table 3.2). Partial CCA demonstrates that on a community level, after landform age has been taken into account, elevation above thalweg explained very little of the compositional variation across sites. However, with elevation above thalweg accounted for, landform age remained a statistically significant environmental driver. Partial CCA also suggested that light availability may actually be a stronger driver than age, however we believe that the influence of an outlier site, in which canopy openness was double that of any other site, may invoke considerable bias in this statistic.  This possible bias is eliminated using a rank-based ordination technique such as NMDS and therefore we believe that NMDS is a more reliable in this circumstance due to its ability to deal with nonlinearities. Furthermore, canopy openness is governed by overstory structure, which is a function of landform age (Figure 3.7).  That is, light availability may be a mechanism that directly affects understory composition but is itself governed by landform age. Thus, we assert that landform age is the primary driver influencing understory community composition, although light availability should be considered a significant environmental factor that directly influences species distributions and changes over time with overstory succession. Because landform elevation above thalweg is highly correlated with landform age, (Figure 3.11), we were interested in disentangling the relative strength of these drivers. We have demonstrated that the dataset, examined as a whole, revealed landform age as a stronger driver than elevation.  However, we also assessed a subset of sites, to investigate the extent that elevation may drive community composition under certain conditions.  For these ends we used 80  NMDS1 scores as a metric that describes understory compositional variation related to age and elevation. Species composition changed rapidly across sites of ages 0-80, beyond which NMDS1 scores increased with gradual slope (Figure 3.16).  Similarly, species composition changed rapidly over an elevational gradient up to 3 m above thalweg, while sites higher than this did not vary significantly with elevation (Figure 3.17).  This elevation threshold roughly corresponded to the bank height of many terrace sites.  We analyzed a subset of our sites that included only those plots < 3 m above thalweg. Within this subset, vegetation composition showed a linear increase with elevation, while the relationship between composition and age was more complex and less strong (Figure 3.18a and 3.18b).  We also analyzed upper terrace sites (elevation above thalweg > 3 m) and found that within this subset, community composition had a very weak relationship with elevation (Figure 3.19a).  However, on these higher elevation sites a clear linear relationship between landform age and understory composition is present (Figure 3.19b).  81  Table 3.3- Partial Constrained Correspondence Analysis (CCA) inertia and P-values.  P-values from a pseudo-ANOVA permutation test, which assesses the explanatory strength of the environmental variables, are indicated for each combination of variables. Several iterations of partial direct gradient ordinations were implemented to assess the relative strength of four environmental variables that may be driving the variation in plant community composition. Variables that explain a significant portion of the compositional variation after other variables have been accounted for are indicated by bold. Constraints included in the model % inertia explained by constraint(s) Landform Age (A) Elevation above Thalweg (E) Light (Canopy openness) (L) Depth to fluvial substrate (D) A 7 0.01 - - - E  5 - 0.06 - - L 11 - - 0.01 - D 4 - - - 0.05 A+E 10 0.01 0.44 - - A+L 19 0.01 - 0.01 - A+D 10 0.01 - - 0.17 A+E+L 20 0.01 0.45 0.02 - A+L+D 22 0.01 - 0.01 0.04 E+L+D 20 - 0.05 0.01 0.05 A+E+L+D 24 0.01 0.35 0.01 0.03    82   Figure 3.13 – Percent cover of common understory species as a function of landform age. Species which occurred in at least five sites are plotted here against landform age. Several species show a clear relationship with landform age while others only occur in early successional and lower elevation or late successional and higher elevation forests. Other species occur with no pattern related to forest age and are likely responding to microsite conditions such as nutrient availability or soil moisture levels. Landform  age  (years) P er ce nt   C ov er 0 20 40 60 80 100 0 200 400 600 ARDI ATFI 0 200 400 600 BLSP BOEL 0 200 400 600 CLAD CLDE 0 200 400 600 CLSI DIHO DREX FEST GAAP GASH HYSP ISMY KIPR 0 20 40 60 80 100 LAMU 0 20 40 60 80 100 MADI PISI PLIN PLPO PLUN POMU RHGL RHLO RIBR RUPA RUSP SARA SCBO STAM STCO 0 20 40 60 80 100 TITR 0 20 40 60 80 100 0 200 400 600 TSHE 0 200 400 600 VAPA VIAD 0 200 400 600 VIGL Tree  seedling Trautvetteria   caroliniensis Tolmiea menziesii Tsuga heterophylla Vaccinium parvifolium Viola   adunca Viola   glabella Rubus parviflorus ibes bracteosum Rubus spectabilis Sambucus racemosa capania bolanderi Streptopus amplexifolius tachys cooleyae iarella   trifoliata Maianthemum dilatatum icea sitchensis Plagiomnium insigne Plagiochila porelloides Plagiothecium undulatum Polystichum munitum Rhizomnium glabrescens Rhytidiadelphus loreus Herb,  Grass Moss,  Liverwort,  Lichen Shrub,  Tree  seedling Fern Dryopteris   expansa estuca sp. Galium aparine Gaultheria shallon Hylocomium splendens Isothecium myosuroides Kindbergia praelonga actuca muralis runcus dioicus Athyrium filix-­femina Blechnum spicant oykinia elata Cladonia sp. Climacium dendroides Claytonia sibirica Disporum hookeri 83   Figure 3.14 – Percent cover of five ubiquitous understory species across landform ages. These extremely common species showed consistent relationships between abundance and landform age. Spline curves represent percent cover across landform ages of Maianthemum dilatatum (False Lily-of-the-Valley), Polystichum munitum (Sword Fern), Blechnum spicant (Deer Fern), Kindbergia praelonga (Slender Beaked Moss), and Hylocomium splendens (Step moss).  The smaller graphs show the large amount of variation around these curves. 0 10 20 30 40 50 50 60 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 200 400 600 Landform  Age  (years)   (%   C ov er )   (%   C ov er )   (%   C ov er ) 0 200 400 600 0 200 400 600 0 200 400 600 Landform  Age  (years) 0 200 400 600 Landform  Age  (years) 0 100 200 300 400 500 600 700 0 10 20 30 40 50 Landform  Age  (years)   (%   C ov er ) Polystichum  munitum Maianthemum  dilatatumKindbergia  praelongaHylocomium  splendens Blechnum  spicant Channel Pioneer  Bar Developing  Floodplain Established  Floodplain Transitional  Terrace Mature  Terrace Old  Growth  Terrace 84    Figure 3.15 – Non-metric Multidimensional Scaling ordination – understory shrubs, herbs, ferns, bryophytes. Plots are ordinated based on species compositions (% cover). Vectors of environmental attributes are added to show general relationships between plots and light (% canopy openness), depth (depth to fluvial substrate), elevation (above thalweg), and age (of landform). Variability in species composition along the NMDS1 axis, appears to be driven by an environmental gradient related to age and elevation. There is also considerable variation along the NMDS2 axis which is related to light availability.  Abandoned  Channel Developing  Floodplain Established  Floodplain Transitional  Terrace Mature  Terrace Old  Growth  Terrace Pioneer  Bar lig ht age elevation depth -­1.5 -­1.0 -­0.5 0.0 0.5 1.0 -­1 .0 -­0 .5 0. 0 0. 5 NMDS1 N M D S 2 85   Figure 3.16 – Understory community composition as a function of landform age. NMDS1 scores are used as a metric of understory composition based on percent cover of all understory species. Dark green circles indicate sites located > 3 m above thalweg while light green circles are those sites located < 3 m above thalweg. Circle size is based on the elevation above thalweg, where larger circles indicate higher elevation sites.   Figure 3.17 – Understory community composition as a function of elevation above thalweg. NMDS1 scores are used as a metric of understory composition based on percent cover of all understory species. Dark green circles indicate sites located higher than 3 meters above thalweg while light green circles are those sites located < 3 m above thalweg.  Circle size is based on the landform age, where larger circles indicate older sites.  Below 3 m, understory composition changes rapidly with elevation however, compositional variation of sites > 3 m above thalweg is not as large. 0 100 200 300 400 500 600 700 -1 .5 -1 .0 -0 .5 0. 0 0. 5 1. 0 Understory Composition = f(Age) circle size = elevation above thalweg Landform Age (years) N M D S  a xi s 1 sc or e (U nd er st or y C om po si tio n) 5 -­1 .5 -­1 .0 -­0 .5 0. 0 0. 5 1. 0 Understory  Composition  =  f(Elevation  above  Thalweg) circle  size  =  Landform  Age   Elevation  above  contemporary  thalweg  (m) N M D S   a xi s   1   sc or e   (U nd er st or y   C om po si tio n) 4321 106 232211 86   Figure 3.18a and b – Comparison between elevation and age as drivers for understory community composition change on lower floodplain sites (elevation above thalweg < 3 m). Elevation and understory composition display a strong linear relationship on low elevation sites. The relationship between landform age and understory composition is more complex.  In 3.18a, circle size is based on the landform age, where larger circles indicate older sites. In 3.18b, circle size is based on the elevation above thalweg, where larger circles indicate higher elevation sites.   Figure 3.19a and b – Comparison between elevation and age as drivers for understory community composition change on upper terrace sites (elevation above thalweg ≥ 3 m). Landform age appears to better account for the variation in community composition. In 3.19a, circle size is based on the landform age, where larger circles indicate older sites. In 3.19b, circle size is based on the elevation above thalweg, where larger circles indicate higher elevation sites  1.5 2.0 2.5 3.0 -­2 .0 -­1 .5 -­1 .0 -­0 .5 0. 0 0. 5 A)  Understory  Comp.  =  f(Elev.  above  thalweg) in  floodplain  forests  (elev.  above  thalweg  =  <  3  m  ) circle  size  =  age Elevation  above  thalweg  (m) N M D S   a xi s   1   sc or e   (U nd er st or y   C om po si tio n) 0 100 200 300 400 500 -­2 .0 -­1 .5 -­1 .0 -­0 .5 0. 0 0. 5 B)  Understory  Comp.  =  f(Landform  Age) in  floodplain  forests  (elev.  above  thalweg  =  <  3  m  ) circle  size  =  Elev.  above  thalweg Landform  Age  (years) N M D S   a xi s   1   sc or e   (U nd er st or y   C om po si tio n) 0. 0 0. 2 0. 4 0. 6 0. 8 1. 0 A)  Understory  Comp.  =  f(Elev.  above  thalweg)   in  terrace  forests  (elev.  above  thalweg  =  >  3  m) circle  size  =  age Elevation  above  thalweg  (m) N M D S   a xi s   1   sc or e   (U nd er st or y   C om po si tio n) 0 200 400 600 0. 0 0. 2 0. 4 0. 6 0. 8 1. 0 Understory  Comp.  =  f(Landform  age)   in  terrace  forests  (elev.  above  thalweg  =  >  3  m) circle  size  =  elev  above  thalweg Landform  age  (years) N M D S   a xi s   1   sc or e   (U nd er st or y   C om po si tio n) 22.05.5 6.04.0 4.53.0 3.5 87  3.4	
  Discussion	
   3.4.1	
  Biogeomorphic	
  succession	
  within	
  the	
  overstory	
  of	
  alluvial	
  forests	
  –	
  refining	
  a	
   conceptual	
  model	
  for	
  the	
  Pacific	
  Coastal	
  Ecoregion	
   The large diversity of forest types is maintained by repeated rejuvenation and patch creation by the migrating and avulsing river.  Over a large time scale this ecological disturbance results in a mosaic of forest patches (Chapter 2).  We invoke the concept of a space for time substitution and assert that different forest stages within the forest mosaic represent a successional gradient.  These range from young, dense deciduous dominated pioneer bar forests to wide open 300 year old mature terraces dominated by towering Picea individuals and beyond to the dark 500+ year old stands dominated by Tsuga.  Although we have separated this gradient into categories it should be noted that forests categorized at the upper end of one type could often be described as the lower end of another type. Primary pathway of biogeomorphic succession in the Carmanah Valley. The youngest stands growing on fluvial deposits of Carmanah River are characterized by high densities of Alnus with a subcanopy of Picea, which contributes to this initial enduring cohort.  These pioneer bars, 0-30 years old, usually establish on the point bar of a meander bend or on island bars behind protective large wood within the channel.  During initial stage, Tsuga may be present in large numbers however our observations corroborate evidence that this species often dies off in the first several years after germination and does not thrive on gravel bars except on decaying wood (Pojar and McKinnon 2004, Van Pelt et al. 2006).  Conceptually, this forest type may be described as the stand initiation stage (Oliver and Larson 1990).  These dense riparian Alnus stands depend on episodes of hydrogeomorphic disturbance and are not common 88  in areas below reservoirs, which lack the flow variation needed to create fluvial deposits (Mallik and Richardson 2009). During years 30-70, intense density dependant mortality occurs within this cohort of Alnus and Picea, and this developing floodplain forest type is characteristic of stands within the stem exclusion stage (Oliver and Larson 1990).  Our results suggest that as these young forests age, rapid vertical growth rates of Alnus shade out many conifers, although a large population of suppressed Picea individuals persist in most stands.  Large decreases in density of all species is simultaneous with a large decrease in canopy openness from year 0-50, indicating that density dependant resource depletion is resulting in self thinning (Figures 3.5 and 3.6).  Tsuga are occasional in these forests except for on decaying large wood where they persist in larger numbers. This is presumably due to the combination of low light conditions and undeveloped soils of this forest stage.  Forest patches at this stage are relatively low lying on the floodplain mosaic and may experience regular flood inundation. This inundation results in fine sediment deposition and landform aggradation. We observed abundant organic matter that had been caught at the base of stems and acted as a shelter to sand deposits.  Over time, this process results in substantial landform lifting. After 60 years, Alnus begin to decrease in the stand due to age-related mortality although many individuals persist until 100 years or older.  At this point Picea dominates the canopy and stands with this key characteristic, ages 60-130 years, are referred to as established floodplains. Understory re-initiation occurs during this timeframe and Tsuga saplings are abundant at the upper end of this range. By ~120 years Alnus is completely absent from the stand and by 150-200 years we observed an abundant subcanopy layer of Tsuga in these transitional terrace forests. 89  Transitional terraces differ from established floodplains in their species composition, due to a lack of Alnus individuals and substantial Tsuga subcanopy, as well as their relationship with the channel.  Over the past century and a half of development, geomorphic lifting has been occurring in conjunction with biological succession (Figure 3.9).  Winter floods that result in overbank flow deposit sand and silt on floodplains and bars, and these peak flow events may also result in channel incision as bedload is transported downstream.  Together these processes lift floodplain surfaces to the point that they are rarely inundated except by the most extreme floods and may be described as terraces.  Vegetation play a role in this process, and over the first hundred years of development the extent of sediment accretion is integrally linked with the density and type of vegetation of the landform (Allmendinger et al. 2005, Gurnell and Petts 2006).  As floodwaters flow through forests, the friction of dense stem densities causes the water to slow and deposit sediment.  The combined biologic and geomorphic development of these landforms is referred to as biogeomorphic succession (Corenblit et al. 2007). Mature terraces, aged 250-500 years, are characterized by very large Picea individuals dominating the canopy but interspersed with smaller diameter and lower height Tsuga individuals.  During this time period, these giants begin to succumb to fungal rot and other forces of age related mortality (Van Pelt et al. 2006).  Because Picea only regenerate in the bright and freshly disturbed conditions of the floodplain, by 400-500 years Tsuga dominates the canopy. The complete absence of Picea pioneers from the initial cohort signifies the transition from mature to old growth terraces.  Mean Tsuga density is relatively stable from 100 years onward over all forests sampled, reflecting the ability of this species to grow as a subcanopy species and its ability to recruit in developed forests and low light conditions.  This life history strategy results in Tsuga dominating this oldest forest stage; however, these forests also contain a 90  substantial component of Abies. This composition of Tsuga and Abies interspersed in a vertically diverse canopy may be maintained for centuries with gap dynamics processes. Alternate pathways of biogeomorphic succession In general, the variability of forest composition and structure is well described as a function of age, however, there is considerable variability along this gradient that may be attributed to alternate pathways of biogeomorphic succession.  We believe that variation along the second NMDS axis may be mechanistically related to site wetness or water table depth, which may promote multiple pathways of succession.  Alternative trajectories of succession, similar in age to developing floodplain sites, are apparent at abandoned channel sites. Differences in composition and structure at this young stage may be responsible for the variability seen in older stages (Kalliola and Puhakka 1988, Van Pelt et al. 2006).  Three different trajectories of developing floodplains are described: dry succession, wet succession, and wet-disturbed succession. Dry succession occurs on point bars or sheltered gravel bars near the primary river channel and is the most common trajectory of floodplain forests in this system.  This was described above as the primary successional pathway and examples of this trajectory are characterized by very dense, young stands with an abundance of Alnus individuals, many Picea saplings and few Tsuga saplings.  These sites may be temporarily inundated by peak winter floods; however, because they occur on the point bar of a meander bend or in the sheltered wake behind a log jam, the hydraulic force of the inundating flood is generally insufficient to scour the site.  Furthermore, these sites are generally well drained and at higher elevation than wet succession sites.  In nearby catchments similar drier, elevated sites are commonly dominated by Pseudotsuga menziesii and Picea are restricted to wetter, low elevation zones (Stolnack and 91  Naiman 2010), although in Carmanah River valley the wetter climate allows Picea to persist here. Sites following other trajectories of succession were designated in the field as well as in aerial photography analysis as a subcategory of developing floodplains, based on their geomorphic history. These abandoned channel sites were located in the depression of abandoned secondary channels.  Depending on the hydrologic regime at the particular site, they may develop as either wet succession or wet-disturbed succession.  Both trajectories appear to be inundated quite frequently during high flow periods, although the hydraulic force of the inundating flow differs between types. Wet-succession abandoned channel forests develop on a silty substrate with a very high water table and are characterized by an abundance of Rubus spectabilis.  These sites have a much lower density of trees presumably due to competition with Rubus.  However, both Alnus and Picea persist in low numbers on these sites.  These sites appear to accrete fine sediments rapidly as floods bring slow-moving, silt-laden waters along the path of former or ephemeral channels gradually depositing and building a deep layer of muddy substrate (sites 2401 and 6401 are examples of this type). Wet-disturbed succession is very similar in that these sites also occur along abandoned or ephemeral channels that may fill during high flows; however, they contain only Alnus individuals and the density of Rubus is generally lower than wet succession sites.  The other difference is that the substrate of these sites is scoured gravel as opposed to accreted silts indicating that these sites experience a profoundly different hydrologic regime.  The regular flood inundation experienced here is of sufficient hydraulic force that sand and silt as well as Picea seedlings are scoured and removed from the site (site 5513 is an example of this type). 92  These situations described above are typical of many sites although in practice stands may experience variations of any of these successional trajectories resulting in varying stem densities, growth rates, and basal areas. The role of individual basin characteristics on forest composition and succession Our research builds on the work of Fonda (1974), Van Pelt et al. (2006), and Naiman et al. (2010) to further refine understanding of alluvial forest dynamics in the Pacific Coastal Ecoregion.  These studies have examined riparian forests of large rivers that are subject to hydrologic regimes with a large snowmelt component resulting in a substantial spring freshet.  In general, the conceptual model of Van Pelt et al. 2006 was well supported by our results. The alluvial forest dynamics in the Carmanah River valley can be described as a primary pathway of succession initiated by geomorphic disturbance within the main channel.  Like Van Pelt et al. (2006), we also found evidence of alternate pathways of succession driven by wetter initial geomorphic conditions that occur on lower elevation patches in abandoned secondary channels. Although these basic mechanisms seem to apply within the Carmanah River valley mosaic, differences in basin scale, geography, hydroclimate, faunal assemblages and sampling design illuminate facets within the general model that vary, depending on individual watershed characteristics. Similar forest structures and biogeomorphic processes occur on floodplains of larger rivers, however variations in forest composition and age were apparent between forests of the Carmanah River and Queets and Hoh Rivers in Washington State.  We observed several differences in detail that may be unique to smaller, flashier pluvial rivers such as the Carmanah. Both large and small rivers floodplain forests commonly originate as dense stands of primarily deciduous vegetation. However pioneer bars of the Queets and Hoh are generally composed of 93  Salix, Populus and Alnus while Carmanah supports only Alnus (Fonda 1974, Van Pelt et al. 2006, Naiman et al. 2010.)  This compositional variation is likely a reflection of the difference in hydroclimate.  The large rivers of the Olympic peninsula are subject to a prolonged spring freshet as snow and glacial meltwaters drain through the catchment.  This results in a long receding limb of the hydrograph during the spring and early summer, which facilitates the distribution of Salicaceous seeds and establishment of seedlings.  As the freshet waters recede gradually, propagules of Populus and Salix are able to germinate on gravel bars exposed by peak flows of the previous winter or spring.  The gradual rate of recession allows roots of seedlings to grow at a rate that maintains contact with the water table (Mahoney and Rood 1998).  Carmanah River has a much flashier flood response than these rivers for two reasons. First, it has a smaller basin area, which decreases basin response time. Second, because maximum elevations within the Carmanah watershed are lower and the entire watershed is close enough to the moderating Pacific Ocean, snow accumulation is minimal within the catchment and there is no spring freshet.  We observed no occurrences of Populus within the catchment and observed only two Salix individuals.  This is likely due to the germination requirements of these species, which are not met by flashy, pluvial streams such as the Carmanah River.  Thus pioneer bars and developing floodplains are dominated by Alnus, which is the only deciduous pioneer species able to establish in these conditions. Another difference on the Carmanah River is the nearly ubiquitous presence of Picea on young floodplain sites.  Research in larger rivers reports that Picea often colonize as a second wave many years after Alnus, after stem density decreases (Van Pelt et al. 2006).  We found Picea present on all young sites except for one, usually with a minimal lag in colonization behind Alnus (Figure 3.1).  On 74% of sites that contained Alnus the establishment lag of Picea was < 10 years, and in 53% of sites the establishment lag of Picea was < 5 years.  The elevated 94  early Picea abundance along the Carmanah River could be due to a difference in faunal composition and resulting levels of floodplain herbivory.  Roosevelt elk, who have been shown to have a substantial effect on riparian community composition (Beschta and Ripple 2006), do not exist in this region but are abundant on the floodplains of other larger rivers and (Van Pelt et al. 2006).  Also, field observations indicate that deer are scarce near our study site.  In nearby large river systems, preferential feeding habits of herbivores may exclude young conifer seedlings from easily grazed young floodplains, whereas the lack of an abundant ungulate population in the Carmanah River valley may explain the proliferation of young conifers on developing floodplain sites. Our chronosequence of 600-700 years was much longer than the 330 year chronosequence described for the Queets River (Van Pelt et al. 2006).  This may be a difference in sampling design but also a reflection of the characteristics of terrace forests in smaller rivers. Larger rivers have a higher rate of hydrogeomorphic disturbance of mature terrace forests, therefore preventing many patches from reaching the old growth terrace stage (Chapter 2).  In Carmanah, this old-growth stage is quite common along the river corridor covering 35% of the area we believe to be of fluvial origin (Chapter 2).  Within our 3 km long study reach, over 2 ha of this forest type was eroded by Carmanah River within the past 70 years, which results in old growth terraces contributing a substantial volume of large wood to the river system. Although this forest type has a half-life of ~1500 years, it nevertheless functions as an integral part of the fluvial-habitat / alluvial-forest system (Chapter 2). 95  	
   3.4.2	
  Understory	
  dynamics	
  and	
  the	
  declining	
  influence	
  of	
  the	
  contemporary	
  hydrologic	
   regime	
  over	
  time	
   Like the forest canopy, understory composition follows a successional gradient from young to old alluvial sites. However, results indicate that other environmental factors, which change as biogeomorphic succession progresses across centuries were also driving species composition.  Furthermore, we observed greater understory variation within forests of similar ages that appears to be driven by differences in site elevation, and light availability. Our analysis revealed that on lower elevation floodplain sites, understory composition is driven by the frequency of inundation, as indicated by a strong relationship between NMDS1 scores and elevation above thalweg. That is, species occurrences were stratified according to the frequency with which they are inundated by floodwaters.  Certain species were present only in low elevation sites presumably due to adaptations that allow these species to persist under floodwaters that cause low oxygen soil conditions during prolonged floods (Hupp and Ostercamp 1996, Chapin et al. 2002).  However, on these low elevation sites there is also a clear but nonlinear relationship with landform age, which indicates that understory composition may also be mechanistically related to successional processes.  That is, on these young sites several understory species, which are adapted to floods, may be outcompeting other species based on ruderal life history strategies.  Furthermore, we observed that light availability also plays a strong role in determining species cover and is highly correlated with landform age and elevation on floodplain sites (Hall and Harcombe 1998). On higher elevation, upper terrace sites, which rarely experience hydrologic inundation, we did not find a strong relationship with elevation above thalweg.  In these older and raised 96  forests, successional processes appear to drive the dynamism of community composition without influence from the contemporary hydrologic regime.  Here, understory vegetation dynamics appear to be linked to overstory dynamics, and species of trees, shrubs, herbaceous plants, ferns, and bryophytes increase and decrease in dominance over tens or hundreds of years as overstory succession progresses. In general, across lower and upper elevation sites, the dynamics of understory plants appear to be a function of changing abiotic conditions such as light availability, elevation above thalweg, and like other factors not measured, such as temperature, soil moisture and soil nutrient conditions, which are all controlled by biogeomorphic development (Fonda 1974).  Several studies have described a relationship between vegetation and flood inundation frequency within the context of a presumed static geomorphology (e.g., Bedinger 1978, Hall and Harcomb 1998, Chapin et al. 2002).  These studies have emphasized the influence of elevation above the river channel without regard to forest succession or geomorphic processes which, over time, alter environmental conditions including the elevation of the landform. Therefore, in unstable reaches, which may be the norm along geomorphically active rivers in the Pacific Coastal Ecoregion, the relationship between elevation and vegetation composition may be weak especially if both overstory and understory composition is considered (Teversham and Slaymaker 1976). Furthermore, unlike the large rivers that have dominated research in this field, the Carmanah River valley lacks a prolonged period of springtime inundation and therefor flood inundation may not be a strong environmental filter in this and similar flashy watersheds.  Thus, much of the heterogeneity of vegetation along the river corridor is due to successional processes rather than frequency of inundation. In conclusion, understory composition is tightly linked to successional processes although there are definitely flood frequency-based selection pressures operating on lower floodplain 97  landforms.  Species distributions are a reflection of several environmental filters that are changing over time due to the landform aggradation and overstory successional processes.  The strong influence of geomorphic and biological change demands an integrated view of how biogeomorphic succession drives the dynamics of factors that plants directly respond to such as light availability, soil depth, inundation frequency, soil moisture and microclimate.  3.5	
  Conclusions	
   The heterogeneity within the mosaic of forest patches in the Carmanah valley bottom is a result of the continuous fluvial reworking of theses alluvial forests and the integrated processes of biological succession and landform aggradation. This joint process, described as biogeomorphic succession, results in the hundreds of stands at different seral stages and geomorphic levels, which together comprise the mosaic of the Carmanah valley. We have described the dynamics and environmental drivers that shape overstory and understory composition within this patchwork. Our results indicate that overstory composition and structure is a function of landform age or time since hydrogeomorphic disturbance.  However, there is great variability in this relationship and we have observed evidence of alternate pathways of succession that may begin to explain this variability.  Forests which develop on landforms of different geomorphic histories and initial hydrologic inundation regimes may set up multiple pathways of succession which result in high diversity within the riparian landscape. This is also true in regards to understory composition, which was related to landform age across the entire range of sites. However, understory species distributions at low elevation floodplain sites were also driven by geomorphic history, or elevation above thalweg. 98  Furthermore, light availability was shown to be a significant factor in determining community composition.  Thus, species composition changes over time as overstory succession and geomorphic development alter environmental conditions, such as light availability or flood inundation frequency, at the understory level. Thus, within the context of a dynamic biogeomorphic successional framework, environmental filters such as flood inundation frequency and light availability determine community composition, however, within this framework, the relative importance of environmental filters change over time dependant on the biogeomorphic successional evolution of the landform.   	
   99  Chapter	
  4: Conclusions	
   4.1	
  General	
  conclusions	
   The Carmanah River valley alluvial forest ecosystem owes its great diversity of plant life, including the large Picea sitchensis individuals for which it has been preserved as a provincial park, to a hydrogeomorphic disturbance history that perpetuates a dynamic mosaic of forest patches that range in age and species composition.  Over the past century, the Carmanah River has reworked nearly 30% of the alluvial forest in our study area, and over the past 500 years the channel has eroded approximately 65% of this area. Within portions of this eroded area, the river has disturbed and rejuvenated the floodplain forest multiple times.  Each disturbance event sets up a path of vegetation succession and geomorphic development after initial substrate deposition (Figure 4.1).  Most commonly, young forests are characterized by a high density of Alnus with a subcanopy of Picea. As Alnus individuals die, Picea increasingly dominates the canopy.  Within these forests, Tsuga regeneration results in many suppressed individuals of this species thriving within the understory.  The original cohort of Picea dies after 300-500 years (Van Pelt et al. 2006), which allows Tsuga to dominate old-growth terrace forests.  Picea or Alnus do not tend to regenerate under these dense Tsuga canopies and without disturbance Tsuga may remain dominant indefinitely. During the floodplain forest’s first century of biological succession, geomorphic lifting is also occurring via the processes of landform accretion and river incision.  Like the forest canopy, understory composition was also related to landform age; however, species distributions at low elevation floodplain sites also appeared to be associated with flood inundation frequency.  Light availability was also a significant factor in driving community composition.  Overall, understory compositional change was tightly linked to overstory succession and geomorphic development as 100  these processes alter environmental conditions at the understory level.  Thus, species distributions were driven by several environmental filters, which changed over time as biogeomorphic succession occurred. 4.2	
  Climate	
  change	
  and	
  forest	
  management	
  within	
  the	
  riparian	
  mosaic	
   The mosaic of the Carmanah River valley is dynamic.  The forest is undergoing disturbance and rejuvenation in some locations while other patches are progressing along specific successional trajectories.  Developing forests are more often eroded than mature forests on a per total area basis, thus older patches tend to persist longer than young forests.  This self- organisation results from feedbacks between the biological composition of a landform and events that influence its geomorphological maintenance or development (Corenblit et al. 2007, Francis et al. 2009).  Decreased erosion of mature forests may be partially due to the cohesive bank strength that develops as the forest ages (Beechie et al. 2006, Eaton and Giles 2009).  Thus, the current proportional distribution of forest types can be attributed to the interaction between the recent regime of fluvial disturbance and the biogeomorphic resistance and resilience of forest patches (Tabacchi et al. 2009).  The landscape composition owes its configuration to a disturbance regime that allows for successional development between episodes of rejuvenation. If resistance- or resilience-related mechanisms are prevented from operating fully, instability of the riparian mosaic composition and decreased biological diversity may result (Roxburgh et al. 2004).  Two examples of such factors that may interfere with the development of resilience are forest management activities that systematically remove portions of the alluvial forest or increased disturbance rates that may occur due to climate change. The Carmanah River currently supports a higher ratio of mature to developing riparian forests compared to the larger, nearby Queets River (Latterell et al. 2006).  Presumably the 101  greater proportion of young forests in the Queets is due to higher magnitude flows, which result in a higher rate of forest erosion in this system (Konrad et al. 2011).  The balance of the forest mosaic composition depends on the rates of biogeomorphic succession as well as the rates of hydrogeomorphic disturbance for each forest stage.  The current balance of the riparian system is such that the half-life of each forest type is comparable in magnitude to the succession rate or the time needed to advance to the next stage of each forest type (Table 4.1).  An equal rate of succession and patch half-life allows half the forest area to be recruited to the next stage of riparian forest at each step of succession. Over time this has resulted in a steady supply of mature forest for which the Carmanah River valley is known. The rate of succession is controlled by biological parameters, however the erosion rate is linked to the intensity of storms that cause hydrogeomorphic disturbance.  It is predicted that with climate change, the frequency of large storms may increase, thus the erosion rate may also increase.  If the hydrogeomorphic disturbance rate is increased it would be expected that the landscape configuration would be transformed towards a decreased ratio of mature to developing forests in this and other alluvial ecosystems. Forest management also has the capacity to change the landscape composition by systematically removing resistant portions of riparian forest.  Forest harvesting within valley bottom forests is typically larger in extent than the area that may be disturbed due to natural processes (Pearson 2010).  If mature terrace area is removed during forest harvesting activities, the capacity for resistance to channel widening and channel migration during hydrogeomorphic disturbance events may be diminished.  Typically in British Columbia, a 30-50 m buffer is retained around fish-bearing streams.  In the case of the Carmanah River, this 50 m wide riparian strip would contain a higher proportion of developing compared to mature forest, relative to the entire alluvial forest area.  In areas, this band of forest may be quickly eroded with subsequent 102  hydrogeomorphic disturbance events leaving immature replanted forests at the channel edge. The resilience of the entire alluvial ecosystem depends on a mosaic of forest patches with different geomorphic and biological functions.  Forest harvesting that systematically removes mature portions of the alluvial mosaic may result in a riparian system that is eroded more quickly by subsequent hydrogeomorphic disturbance events.  If the development of resistance is hindered and the disturbance regime is increased, the capacity for the alluvial system to develop and maintain mature patches may be compromised. Thus, an “emulation of natural disturbance” approach to forest management could be adopted to maintain a forest mosaic of natural proportional forest composition (Moore and Richardson, in prep.)  Table 4.1 – Succession rate vs. erosion rate for each forest type. The rate of succession is the time needed to advance from one forest stage to the next. The half-life is calculated based on 70 years of data fitted to decay curves that describe the disturbance rate of forest patches that existed in 1937. Forest stage Age range (y)  Rate of Succession (y) Half life (Erosion rate) (y) Pioneer Bar 0-25 25 24 Developing Floodplain 25-60 35 47 Established Floodplain 60-110 50 78 Transitional Terrace 110-220 110 97 Mature Terrace 220-500 390 357 Old Growth Terrace 500+ - 1516 	
    103   Figure 4.1 – Conceptual model of the Carmanah River valley alluvial forest.  Primary and alternate pathways of biogeomorphic succession are illustrated in the above figure.  Hydrogeomorphic disturbance may erode part of a forest patch at any stage, converting it to channel area. Half-life indicates the estimated time in which half the area will be eroded by hydrogeomorphic disturbance processes.  Biological and geomorphic characteristics are noted below each forest type. Forest type names are based on Latterel et al. (2006) and Van Pelt et al. (2006). Secondary Channel Abandoned   Channel avulsion years ~25Half  Life  = Primary  Pathway  of  Biogeomorphic  Succession Alternate  Pathway  of  Biogeomorphic  Succession years ~50 years ~75 years ~100 years ~350 years ~1500 stand  initiation Alnus  +  Picea  saplings frequent  flood  inundation +rapid  landform  accretion ~5  years ~25  years ~60  years ~110  years ~220  years ~500  years stem  exclusion  in  Alnus   canopy  /  Picea  subcanopy frequent  flood  inundation +rapid  landform  accretion Alnus  begin  to  die  off, understory  re-­initiation infrequent  flood  inundation +landform  accretion Picea  dominated  canopy with  Tsuga  subcanopy rare  flood  inundation Mixed  Picea/Tsuga  canopy with  rare  Abies very  rare  flood  inundation Tsuga/Abies  canopy extremely  rare   flood  inundation years 0-­10 Hydrogeomorphic  Disturbance years 0-­10 Primary Channel Pioneer   Bar Developing   Floodplain Established   Floodplain Transitional   Terrace Mature   Terrace Old  Growth   Terrace 104  	
   4.3	
  Future	
  research	
   This project suggests several interesting lines of further research.  We used a space for time substitution to examine pathways of succession in these forests.  Although we have confirmed the existence of alternate pathways in abandoned channel vs. primary channel stands there is a need to further refine knowledge of these trajectories and the forces that drive alluvial forest stand dynamics.  Permanent plots could be established at sites that span a gradient of geomorphic histories and elevations. This would allow researchers to monitor pathways of succession in plots exposed to different initial conditions and hydrologic regimes. Another interesting finding of this study is the difference in composition of developing floodplain forests between the Carmanah and nearby larger rivers. There is a need to evaluate Populus and Salix presence in watersheds of the Pacific Coastal Ecoregion to assess if a threshold related to basin size and hydroclimate exists.  We know these species have germination requirements that rely on large springtime floods (Barsoum 2002, Polzin and Rood 1996), which explains their absence at Carmanah.  However, further research is required in order to examine the range of basin parameters will produce these ecologically meaningful floods. This study has also provided preliminary evidence that older terrace sites may be more resistant to erosion during hydrogeomorphic disturbance.  However, further research, which directly compares the response to a single or series of flood events in old vs. young forests is needed to clarify the extent of this trend. Finally, this research and other projects that have dealt with alluvial landscape dynamics could be used to assess areas of forest and quantities of wood that are recruited to the stream system on a decadal scale based on floodplain size and basin characteristics.  This information 105  would be very useful in defining sustainable areas of harvest near streams and rivers in British Columbia and elsewhere.  The current 30-50 m buffer for fish-bearing streams may be insufficient in highly dynamic systems, as large rivers may migrate or avulse this distance over the course of one or several flood seasons.  Further, there is a need to define boundaries and sustainable harvest levels in non fish-bearing streams.  Based on these studies and future research over a range of watershed sizes, ratios of mature to developing forest area could be used to define targets that would be useful, not only to ensure that harvesting activities remain within the natural range of variability, but also in defining restoration targets.   	
   106  References	
   Abbe, T. S., and Montgomery, D.R. (1996)  Large woody debris jams, channel hydraulics and habitat formation in large rivers. Regulated Rivers: Research & Management 12, 201-221. Abbe, T. B. (2000) Patterns, mechanics, and geomorphic effects of wood accumulations in a forest river system. Dissertation. University of Washington, Seattle, Washington, USA. Anderson, R.J., Bledsoe, B.P., and Hession, W.C. (2004) Width of streams and rivers in response to vegetation, bank material, and other factors. Journal of the American Water Resources Association 40, 1159-1172. Andrews, E.D. and Nankervis, J.M. (1995) Effective discharge and the design of channel maintenance flows for gravel-bed rivers. Geophysical monographs / American Geophysical Union 89. Allmendinger, N. E., Pizzuto, J. E., Potter, N., Johnson, T. E., and Hession, W. C. (2005) The influence of riparian vegetation on stream width, eastern Pennsylvania, USA. Geological Society of America Bulletin 117, 229-243. Balian, E.V. and Naiman, R.J. (2005) Abundance and production of riparian trees in the lowland floodplain of the Queets River, Washington. Ecosystems 8, 841-861. Barnes, W. J. (1978) The distribution of floodplain herbs as influenced by annual flood elevation. Wisconsin Academy of Sciences, Arts and Letters 66, 254-366. Barsoum, N. (2002) Relative contributions of sexual and asexual regeneration strategies in Populus nigra and Salix alba during the first years of establishment on a braided gravel bed river. Evolutionary Ecology 15, 255–279. Bedinger, M. S. (1979) Relation Between Forest Species and Flooding. In: Greeson, P.E., Clark, J.R., and Clark, J.E. Eds. Wetland functions and values: the state of our understanding. American Water Resources Association 426-435. Beechie, T.J., Liermann, M., Pollock, M.M., Baker, S., and Davies, J. (2006) Channel pattern and river-floodplain dynamics in forested mountain river systems. Geomorphology 78, 124-141. Beschta R.L. and Ripple, W.J. (2006) River channel dynamics following extirpation of wolves in northwestern Yellowstone National Park, USA. Earth Surface Processes and Landforms 31, 1525-1539. Bendix, J., and Hupp, C. (2000) Hydrological and geomorphological impacts on riparian plant communities. Hydrological Processes 14, 2977-2990. Bird, S. A. (1993) Stream Channel and Riparian Zone Response to the Development of a Lateral Sediment Wedge in the Queen Charlotte Islands, B.C. MSc Thesis, University of Western Ontario. Brayshaw, D. (2011) Bankfull and effective discharge in small mountain streams of British Columbia. PhD Thesis. University of British Columbia. 107  Bren, L. J., and Gibbs, N. L. (1986) Relationships between Flood Frequency, Vegetation and Topography in a River Red Gum Forest. Australian Forest Research 16, 357-370. Brummer, C. J., T. B. Abbe, J. R. Sampson, and D. R. Montgomery. (2006) Influence of vertical channel change associated with wood accumulations on delineating channel migration zones, Washington, USA. Geomorphology 80: 295–309. Chapin, D. M., Beschta, R. L., and Shen, H. W. (2002) Relationships between flood frequencies and riparian plant communities in the upper Klamath Basin, Oregon. Journal of the American Water Resources Association 38, 603-619. Corenblit, D., Tabacchi, E., Steiger, J., and Gurnell, A. M. (2007) Reciprocal interactions and adjustments between fluvial landforms and vegetation dynamics in river corridors: a review of complementary approaches. Earth Science Reviews 84, 56-86. Corenblit, D., Gurnell, A. M., Steiger, J., and Tabacchi, E. (2008) Reciprocal adjustments between landforms and living organisms: extended geomorphic evolutionary insights. Catena 73, 261-273. Corenblit, D., Steiger, J., Gurnell, A. M., and Naiman, R. J. (2009) Plants intertwine fluvial landform dynamics with ecological succession and natural selection: a niche construction perspective for riparian systems. Global Ecology and Biogeography 18, 507-520. Eaton, B. C., and Giles, T. R. (2009) Assessing the effect of vegetation-related bank strength on channel morphology and stability in gravel-bed streams using numerical models. Earth Surface Processes and Landforms 34, 712-724. Fannin, R.J., J. Jaakkola, J.M.T. Wilkinson, and E.D. Hetherington. (2000) Hydrologic response of soils to precipitation at Carnation Creek, British Columbia, Canada. Water Resources Research 36, 1481-1494. Fetherston, K.L., Naiman, R.J., and Bilby, R.E. (1995) Large woody debris, physical process, and riparian forest development in montane river networks of the Pacific Northwest. Geomorphology 13, 133-144. Fonda, R.W. (1974) Forest succession in relation to river terrace development in Olympic National Park, Washington. Ecology 55 (5), 927-942. Francis R.A. (2006) Allogenic and autogenic influences upon riparian vegetation dynamics. Area 38, 453-464. Francis, R.A., Corenblit, D., and Edwards, P.J. (2009) Perspectives on biogeomorphology, ecosystem engineering and self-organisation in island-braided fluvial ecosystems. Aquatic Sciences 71, 290-304. Friedman, J., Auble, G., Andrews, E., and Kittel, G. (2006) Transverse and longitudinal variation in woody riparian vegetation along a montane river. Western North American Naturalist 66, (1) 77-91. Geerling, G.W., Ragas, A.M.J., Leuven, R.S.E.W., Van den Berg, J.H., Breedveld, M., Liefhebber, D., and Smits, A.J.M. (2006) Succession and rejuvenation in floodplains along the river Allier (France). Hydrobiologia 565, 71–86. 108  Gonzalez, E., Gonzalez-Sanchis, M., Cabezas, A., Comin, F.A., and Muller, E. (2010) Recent changes in the riparian forest of a large regulated Mediterranean river: Implications for management. Environmental Management 45, 669–681. Gurnell, A. (1997) The hydrological and geomorphological significance of forested floodplains. Global Ecology and Biogeography Letters 6, 219-229. Gurnell, A.M., Petts, G.E., Hannah, D.M., Smith, B.P.G., Edwards, P.J., Kollmann, J., Ward, J.V., and Tockner, K. (2001) Riparian vegetation and island formation along the gravel-bed Fiume Tagliamento, Italy. Earth Surface Processes and Landforms 26, 31-62. Gurnell, A., and Petts, G. (2006) Trees as riparian engineers: the Tagliamento River, Italy. Earth Surface Processes and Landforms 31, 1558-1574. Hall, R. B. W., and Harcombe, P. A. (1998) Flooding alters apparent position of floodplain saplings on a light gradient. Ecology 79, 847-855. Hartman, G.F. and J.C. Scrivener (1990) Impacts of forestry practices on a coastal stream ecosystem, Carnation Creek, British Columbia. Canadian Bulletin of Fisheries and Aquatic Sciences 223. Hassan, M.A., Hogan, D.L., Bird, S.A., May, C.L., Gomi, T. and Campbell, D. (2005) Spatial and temporal dynamics of wood in headwater streams of the Pacific Northwest. Journal of the American Water Resources Association 41, 899-919. Hernandez, O., Merritt, R.W., and Wipfli, M.S. (2005) Benthic invertebrate community structure is influenced by forest succession after clearcut logging in southeastern Alaska. Hydrobiologia 533, 45-59. Hetherington, E.D. (1988) Hydrology and logging in the Carnation Creek watershed-what have we learned?. In: T.W. Chamberlin, Ed.  Proc. workshop: applying 15 years of Carnation Creek results. Pacific Biological Station, Nanaimo, B.C. 11-15. Hetherington, E.D. (1982) A first look at logging effects on the hydrologic regime of Carnation Creek experimental watershed. In: G.F. Hartman, Ed. Proc. Carnation Creek workshop, a 10-year review. Pacific Biological Station, Nanaimo, B.C. 45-63. Hickin, E. J. (1984) Vegetation and river channel dynamics. Canadian Geographer 28, 111-126. Hogan, D. L., Bird, S. A., and Hassan, M. A. (1998) Spatial and temporal evolution of small coastal gravel-bed streams: The influence of forest management on channel morphology and fish habitat. In: Klingeman, P.C, Beschta, R.L., Komar, P.D., and Bradley, J.B., Eds. Gravel-Bed Rivers in the Environment. Water Resources Publications, LLC. Highlands Ranch. 365-392. Honda, M. (2008) Relationships between flood frequency and riparian plant community structure in mountain streams of western Montana. MSc Thesis, The University of Montana. Huang, H. Q., and Nanson, G. C. (1997) Vegetation and channel variation; a case study of four small streams in southeastern Australia. Geomorphology 18, 237-249. 109  Hughes, F. M. R. (1990) The influence of flooding regimes on forest distribution and composition in the Tana River floodplain, Kenya. Journal of Applied Ecology 27, 475-491. Hupp, C. R. (1982) Stream-grade variation and riparian-forest ecology along Passage Creek, Virginia. Bulletin of the Torrey Botanical Club 488-499. Hupp, C. R. (1983) Vegetation pattern on channel features in the Passage Creek Gorge, Virginia. Castanea 48, 62-72. Hupp, C. R., and Osterkamp, W. R. (1985) Bottomland vegetation distribution along Passage Creek, Virginia, in relation to fluvial landforms. Ecology 66, 670-681. Hupp, C. R., and Osterkamp, W. R. (1996) Riparian vegetation and fluvial geomorphic processes. Geomorphology 14, 277-295. Kalliola, R., and Puhakka, M. (1988) River dynamics and vegetation mosaicism: a case study of the River Kamajohka, northernmost Finland. Journal of Biogeography 15, 703-719. Konrad, C., Berge, H., Fuerstenberg, R., Steff, K., Olsen, T., Guyenet, J. (2011) Channel dynamics in the Middle Green River, Washington, from 1936 to 2002. Northwest Science 85 1-14. Little, P.J., Bird, S.A. and Hassan, M.A. (2010) 2009 Longitudinal profile of the BC Ministry of Forests Carnation Creek Experimental Watershed. Unpublished report. Little, P.J., Gonzales, E. and Corenblit, D. (in prep.) Investigating the decline of Salix alba on the River Garonne, France by assessing functional relationships between geomorphic site and tree establishment. Latterell, J. J., Bechtold, J. S., O'Keefe, T. C., Van Pelt, R., and Naiman, R. (2006) Dynamic patch mosaics and channel movement in an unconfined river valley of the Olympic Mountains. Freshwater Biology 51, 523-544. Latterell, J. J., and Naiman, R. (2007) Sources and dynamics of large logs in a temperate floodplain river. Ecological Applications 17, 1127-1141. Leopold, L.B. (1992) Sediment Size that Determines Channel Morphology. In. Billi, P., Hey, R.D., Thorne, C.R. and Tacconi, P. Eds. Dynamics of Gravel-bed Rivers. John Wiley & Sons. 297-311. Moore, R.D. and Richardson, J.S. (in prep.) Natural disturbance and forest management in riparian zones: Catchment- and landscape-scale contexts. Naiman, R., and Décamps, H. (1997) The ecology of interfaces: riparian zones. Annual Review of Ecology and Systematics 28, 621-658. Naiman, R.J., Bechtold, J., Drake, D., and Latterell, J. (2005) Origins, patterns, and importance of heterogeneity in riparian systems. In: Lovett, G., Jones, C.G., Turner, M.G., Weathers, K.C., Eds. Ecosystem function in heterogeneous landscapes. Springer-Verlag, New York. 279-309. Naiman, R.J., Latterell, J.J., Pettit, N.E., Olden, J.D. (2008) Flow variability and the biophysical vitality of river systems. Comptes Rendus Géoscience 340, 629-643. 110  Naiman, R.J, Bechtold, J., Beechie, T., and Latterell, J. (2010) A Process-Based View of Floodplain Forest Patterns in Coastal River Valleys of the Pacific Northwest. Ecosystems 13, 1-31. Oksanen, J., Kindt, R., Legendre, P., O'Hara, B., Simpson, G.L., Solymos, P., Stevens, M.H.H., and Wagner, H. (2009) vegan: Community Ecology Package. R package version 1.15-4. http://CRAN.R-project.org/package=vegan Oliver, C.D. and Larson, B.C., (1990) Forest Stand Dynamics. McGraw-Hill, New York. Pearson, A.F. (2010) Natural and logging disturbances in the temperate rain forests of the Central Coast, British Columbia. Canadian Journal of Forest Research 40, 1970-1984. Perona, P., Molnar, P., Savina, M. and Burlando, P. (2009) An observation-based stochastic model for sediment and vegetation dynamics in the floodplain of an Alpine braided river. Water Resources Research 45. Peterson, E.B., Peterson, N.M., Weetman, G.F., Martin, P.J. (1997) Ecology and Management of Sitka Spruce Emphasizing Its Natural Range in British Columbia. UBC Press, Vancouver. Pojar, J. and MacKinnon, A. (2004) Plants of the Pacific Northwest Coast: Washington, Oregon, British Columbia & Alaska. Lone Pine, Vancouver. Polzin, M.L. and Rood, S.B. (2006) Effective disturbance: Seedling safe sites and patch recruitment of riparian cottonwoods after a major flood of a mountain river. Wetlands 26, 965–980. Primack, A. G. B. (2000) Simulation of climate-change effects on riparian vegetation in the Pere Marquette River, Michigan. Wetlands 20, 538-547. Richardson, J.S., Bilby, R.E., and Bondar, C.A. (2005) Organic matter dynamics in small streams of the Pacific Northwest. Journal of the American Water Resources Association 41, 921-934. Robertson, K.M. and Augspurger, C.K. (1999) Geomorphic processes and spatial patterns of primary forest succession on the Bogue Chitto River, USA. Journal of Ecology 87, 1052- 1063. Roxburgh, S., Shea, K. and Wilson, J. (2004) The intermediate disturbance hypothesis: patch dynamics and mechanisms of species coexistence. Ecology 85, 359-371. Sakai, T., Tanaka, H., Shibata, M., Suzuki, W., Nomiya, H., Kanazashi, T., Iida, S., and Nakashizuka, T. (1999) Riparian disturbance and community structure of a Quercus-Ulmus forest in central Japan. Plant Ecology 140, 99-109. Shin, N., and Nakamura, F. (2005) Effects of fluvial geomorphology on riparian tree species in Rekifune River, northern Japan. Plant Ecology 178, 15-28. Stolnack S.A. and Naiman R.J. (2010) Patterns of conifer establishment and vigor on montane river floodplains in Olympic National Park, Washington, USA. Canadian Journal of Forest Research 40, 410-422. 111  Swanson F.J., Johnson S.L., Gregory S.V., and Acker S.A. (1998) Flood disturbance in a Forested Mountain Landscape. BioScience 48, (9), 681-689. Tabacchi, E., Steiger, J., Corenblit, D., Monaghan, M.T., and Planty-Tabacchi, A.M., (2009) Implications of biological and physical diversity for resilience and resistance patterns within Highly Dynamic River Systems. Aquatic Sciences 71, 279-289. Teversham, J., and Slaymaker, O. (1976) Vegetation composition in relation to flood frequency in Lillooet River valley, British Columbia. Catena 3, 191-201. Van Looy, K., Meire, P., and Wasson, J.G. (2008) Including riparian vegetation in the definition of morphologic reference conditions for large rivers: A case study for Europe’s western plains. Environmental Management 41, 625–639. Van Pelt, R., O'Keefe, T. C., Latterell, J. J., and Naiman, R. (2006) Riparian forest stand development along the Queets river in Olympic National Park, Washington. Ecological Monographs 76, 277-298. Whited, D., Lorang, M., Harner, M., and Hauer, F. (2007) Climate, hydrologic disturbance, and succession: Drivers of floodplain pattern. Ecology 88, 940–953. Wipfli, M.S. and Musslewhite, J. (2004) Density of red alder (Alnus rubra) in headwaters influences invertebrate and detritus subsidies to downstream fish habitats in Alaska. Hydrobiologia 520, 153-163. Wolman, M.G. and Miller, J.P. (1960) Magnitude and frequency of forces in geomorphic processes. The Journal of Geology 68, (1), 54-74 112  Appendix	
  A	
  –	
  Plot	
  overstory	
  composition	
  and	
  structure	
  statistics.	
  	
   Density, mean DBH, maximum DBH, Standard Deviation of DBH and Basal Area are calculated for each tree species present in each plot.  Plots are sorted by estimated landform age.  !"#$ !%&'(&) *+,-+.& /&0)(12-3456+7 *&+0-/89-3':7 *+,;-/89-3':7 !;/;-/89-3':7 8+)+<-=>&+-3:?56+7 @@AB @ CDEEAA A;A B;E A;? B;? !"#$%&'$(') ! ""### #$% %$& #$' #$( *+,-)&%+.,/-#%+% ) !)"&## #$# #$# #$# #$! 0%$1)&/-.-'23/4"") ) (%### #$# #$# #$# #$% F@AE BC BEAAA ?;B G;G ?;A BA;E !"#$%&'$(') %' (### )$* +$+ )$# "$' *+,-)&%+.,/-#%+% %' '*## %$" "$% )$) )$* 0%$1)&/-.-'23/4"") %'## #$" !$% %$) #$) ?E?H BD BCGAA B;A H;D B;@ @;? !"#$%&'$(') %( +#*" %$* "$( %$' !$# *+,-)&%+.,/-#%+% %! &+*" #$! '$' #$" #$! 0%$1)&/-.-'23/4"") %# +*" #$# #$# #$# #$# E@AD BD @?CAA ?;@ BB ?;B ?E;@ !"#$%&'$(') %( )+)## )$* %% )$% )'$+ *+,-)&%+.,/-#%+% %( ')## #$* %$& #$+ #$! B@A@ ?@ DCFH ?;D BG C;A BG;E !"#$%&'$(') )! "%!! !$' %+ '$* %+$% *+,-)&%+.,/-#%+% %! )!!! %$& %$& #$# #$' FEAH @C C@@@ E;G ?@ C;E ?C;G !"#$%&'$(') !' %#*" %) )! &$# %!$( *+,-)&%+.,/-#%+% )( !#*" &$! %! )$+ +$& 0%$1)&/-.-'23/4"") )& )## )$' '$% %$& #$% ?EBC @G E@@@ @;C @A C;D BE;B !"#$%&'$(') !+ "() %) !# "$( %)$% *+,-)&%+.,/-#%+% )' %+!! %$( +$( )$% %$% 0%$1)&/-.-'23/4"") )) )**" )$) %% )$# %$+ !(+-%&)5)(+"+% ( ') #$# #$# #$# ?CAB @G BCFH F;D ?D G;H B@;D !"#$%&'$(') !+ (!! "$* )( %#$* %%$( *+,-)&%+.,/-#%+% !& &!! &$* %) '$) )$# FEA@ @G BF@? B@ ?? E;G ?C;F !"#$%&'$(') !+ %!%* %& )) '$! )!$( *+,-)&%+.,/-#%+% )( !%* '$' %% !$& #$" EEB@ C@ HFE B? ?A F;A BA;G !"#$%&'$(') '! *'" %' )# '$" %#$" *+,-)&%+.,/-#%+% &( %$& %$& #$# 0%$1)&/-.-'23/4"") &( '$# '$# #$% CEAE EB ?AE@ BE CA BA;? E?;C !"#$%&'$(') &% ('" )! '# *$& '%$% *+,-)&%+.,/-#%+% &% %### ($% )+ +$% %%$! 0%$1)&/-.-'23/4"") %#& %$& %$& #$# 113  Appendix A - Overstory statistics continued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ppendix A - Overstory statistics continued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ppendix A - Overstory statistics continued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ppendix	
  B	
  –	
  Forest	
  type	
  plates	
   Channel mean percentage of study area: 11% 117  Pioneer Bar Half Life:   24 years mean DBH*:   2 cm mean percentage of study area:  3% Canopy:  Alnus rubra Landform Age:  1-25 years 118   Developing Floodplain Canopy:   Alnus rubra      Picea sitchensis Landform age:   25-60 years Half Life:   47 years mean DBH:   10 cm mean percentage of study area:   7% 119     Established Floodplain Canopy:   Picea sitchensis      Alnus rubra Landform Age:   60-110 years Half Life:   78 years mean DBH:   25 cm mean percentage of study area:   5% 120    Transitional Terrace Half Life:   97 years mean DBH:   37cm mean percentage of study area:   9% Canopy:   Picea sitchensis     Tsuga heterophylla Age range:   110-250 years 121   Mature Terrace Half Life:   357 years mean DBH:   45 cm mean percentage of study area:   29% Canopy:   Picea sitchensis     Tsuga heterophylla Age range:   250-500 years 122        Old Growth Terrace Age range:   400-700+ years Half Life:   1516 years mean percentage of study area:  36% Canopy:   Tsuga heterophylla      Abies amabalis mean DBH:   18 cm

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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072413/manifest

Comment

Related Items