UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Environmental and biotic influences on the abundance and distribution of an introduced grass species… Deane, Thomas James 2010

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2010_fall_deane_thomas.pdf [ 3.99MB ]
Metadata
JSON: 24-1.0071034.json
JSON-LD: 24-1.0071034-ld.json
RDF/XML (Pretty): 24-1.0071034-rdf.xml
RDF/JSON: 24-1.0071034-rdf.json
Turtle: 24-1.0071034-turtle.txt
N-Triples: 24-1.0071034-rdf-ntriples.txt
Original Record: 24-1.0071034-source.json
Full Text
24-1.0071034-fulltext.txt
Citation
24-1.0071034.ris

Full Text

  ENVIRONMENTAL AND BIOTIC INFLUENCES ON THE ABUNDANCE AND DISTRIBUTION OF AN INTRODUCED GRASS SPECIES: IMPLICATIONS FOR MANAGEMENT IN THE OKANAGAN VALLEY, BRITISH COLUMBIA.   by   Thomas James Deane   B.Sc. (Hons), The University of St. Andrews, 2006   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   MASTER OF SCIENCE   in   THE FACULTY OF GRADUATE STUDIES   (Zoology)   THE UNIVERSITY OF BRITISH COLUMBIA   (Vancouver)   June, 2010   ! Thomas James Deane, 2010  ii Abstract   Introduced plant species exert major influences on the structure and function of ecosystems, and are often implicated in biodiversity declines. The Eurasian annual cheatgrass, Bromus tectorum Linnaeus, has spread extensively in western North America since its introduction over 150 years ago; it extirpates native species, appears to have increased fire cycle periodicities, and provides cattle with inadequate nutrition.  Because cheatgrass abundance recently increased in pastures of grassland in the Okanagan Valley of British Columbia, Canada, assessing environmental and biotic factors that influence its abundance is important from a management perspective. In an observational study at five heterogeneous sites, I isolated a number of highly significant correlations; cheatgrass abundance was positively correlated with proximity to focal ponderosa pine (Pinus ponderosa C. Lawson) trees, but negatively correlated with other plant diversity. Also, soil pH and soil moisture were significantly lower in proximity to trees than at distances further away, suggesting soil chemistry could have affected cheatgrass abundance. Because other analyses indicated that cheatgrass abundance differed in relation to the identity of the other species present, I conducted community- wide and species-specific co-occurrence analyses; I asked whether invaded communities featured different assembly patterns and isolated the species that had the strongest co- occurrence patterns with cheatgrass. I found that communities lacking cheatgrass were more diverse in terms of grass species and appeared to be structured non-randomly. Invaded communities, however, displayed patterns indicative of ‘disassembly’ as co- occurrence relationships did not differ from null predictions. Five grass species grew relatively more frequently if cheatgrass was present; these were bluebunch wheatgrass (Pseudoroegnaria spicata, Pursh. A. Love), Kentucky bluegrass (Poa pratensis L.), western needlegrass (Stipa occidentalis Thurb.), sand dropseed (Sporobolus cryptandrus Torr. A. Gray) and needle-and-thread-grass (Stipa comata Trin & Rupr. Barkworth).  These results suggest that selective herbicide use in proximity to pine trees could be effective in controlling cheatgrass in these grasslands. I recommend manipulative experiments to assess the potential of this technique, as well as seeding experiments designed to characterize the most effective natural competitors against cheatgrass.  iii Preface   I was solely responsible for designing and conducting all research that appears in Chapter 2 of this thesis, and wrote the entire chapter after conducting the associated research. In completing Chapter 3 of this thesis, I analyzed data previously collected and described by Stephens et al. (2009). I was solely responsible for designing and conducting the novel data analyses, and wrote the entire chapter after conducting the associated research.                       iv Table of contents  Abstract .......................................................................................................................... ii Preface ........................................................................................................................... iii Table of contents............................................................................................................ iv List of tables ................................................................................................................... v List of figures................................................................................................................ vii Acknowledgements ...................................................................................................... viii   Chapter 1: General introduction –  cheatgrass, Bromus tectorum Linnaeus ............. 1  Chapter 2: Variations in the abundance and distribution of cheatgrass relative to biotic factors and a potential link to soil chemistry ..................................................... 4 Introduction .................................................................................................................... 4 Methods ......................................................................................................................... 7 Results .......................................................................................................................... 11 Discussion .................................................................................................................... 19 Conclusions................................................................................................................... 25  Chapter 3: Community-wide differences in cheatgrass-invaded communities and species-specific co-occurrence relationships............................................................... 26 Introduction .................................................................................................................. 26 Methods ....................................................................................................................... 29 Results .......................................................................................................................... 33 Discussion .................................................................................................................... 38 Conclusions................................................................................................................... 44  Chapter 4: General conclusions ................................................................................ 45 Summary....................................................................................................................... 45 Suggestions for future research ...................................................................................... 48 Management implications.............................................................................................. 50  References ................................................................................................................... 51          v List of tables  Table   2.1:  Elevation and habitat characteristics of the five study sites........................... 8  Table 2.2: Site-specific linear equations describing significant rates of decline in cheatgrass abundance (number of stems per 0.25m 2  quadrat) with distance from focal pine trees. For all sites, degrees of freedom = 95. Linear equations for each site should be read as the number of cheatgrass stems adjacent (0m) to a pine tree – number of cheatgrass stems/m from a pine tree .................... 12  Table 2.3: Site-specific linear equations describing significant rates of decline in cheatgrass abundance (number of stems per 0.25m 2 quadrat) with increasing number of other species. For all sites, degrees of freedom = 95 (65 for control data). Linear equations for each site should be read as the number of cheatgrass stems – number of cheatgrass stems/other species found............. 13  Table 2.4: Site-specific linear equations describing significant rates of increase in soil pH with distance from focal pine trees. For all sites, degrees of freedom = 47. Linear equations for each site should be read as the soil pH adjacent (0m) to a pine tree + pH units/m from a pine tree ........................................................ 16  Table 2.5: Site-specific linear equations describing rates of increase in soil moisture (%) with distance from focal pine trees. For all sites, degrees of freedom = 47. Linear equations for each site should be read as the soil moisture (%) adjacent (0m) to a pine tree + moisture units (%)/m from a pine tree. Soil moisture (%) at Anarchist Mountain was not significantly correlated with distance from focal pine trees (p=0.53) .............................................................................  16  Table 2.6: Site-specific descriptions of correlations between cheatgrass abundance and percentage component of other species comprising native bunchgrasses (‘NB’) and introduced species (‘IS’). For all sites, degrees of freedom = 95. Proportion increase/decrease for each site should be read as the proportion (%) increase/decrease of other species comprising species belonging to each functional group per stem of cheatgrass ....................................................... 18  Table 3.1: List of the 28 grass species (Family Poaceae) in co-occurrence and diversity analyses, featuring a total of 19 native and 9 introduced, and 10 annual and 18 perennial species. The four species appearing in bold font were those for which spatial distribution patterns were analyzed in considering potential environmental gradients............................................................................... 30  Table 3.2: Statistical comparisons of C-score analyses using the SS and RKT algorithms. All statistically significant, directional results were repeated using both algorithms, although absolute values differ. ‘Intact’ plots are in bold font within years; ‘invaded’ plots are in regular font ........................................... 35   vi Table 3.3: Statistical comparisons of species combinations analysis using SS and RKT algorithms. All statistically significant, directional results were found under both algorithms, although absolute values differ. ‘Intact’ plots are in bold font within years; ‘invaded’ plots are in regular font ........................................... 35  Table 3.4: The 6 species that displayed patterns of interest in co-occurrence relationships with cheatgrass. These species were either found substantially more or less frequently in association with cheatgrass (‘invaded’ plots) than in ‘intact’ plots in either 2001 or 2005, or displayed negative patterns in 2001 and positive patterns in 2005. Bold font is used for species found less frequently in ‘invaded’ plots; regular font is for species found more frequently. Values represent the number of times more or less frequently the species was found in ‘invaded’ plots. Letters in brackets after species names show whether the species is native or introduced (N or I), annual or perennial (A or P)............ 36                                  vii List of figures  Figure 2.1: Cheatgrass declines in abundance (number of stems per 0.25m 2  quadrat) with distance from focal pine trees. Different symbols and colors represent the five study sites, n=96 per site.............................................................................. 11  Figure 2.2: Cheatgrass declines in abundance (number of stems per 0.25m 2  quadrat) as the number of other species present within a quadrat increases. Different symbols and colors represent the five study sites, n=96 per site.................... 13  Figure 2.3: Cheatgrass declines in abundance (number of stems per 0.25m 2  quadrat) as the number of other species present within a quadrat increases in control (no pine tree) transects. Different symbols and colors represent the five study sites, n=96 per site ....................................................................................... 14  Figure 2.4: Changes in soil pH with distance from focal pine trees at all five sites. Different symbols and colors represent different sites, n=48 per site ............ 15  Figure 3.1: Within-year comparisons of mean grass species richness (± SE) between ‘intact’ (black bars) and ‘invaded’ (white bars) plots. Significant differences at p=<0.017 existed for comparisons in 2001, 2002 and 2003. N=44 plots for 2001 ‘intact’, 46 (‘01 ‘invaded’), 51 (’02 ‘intact’), 39 (’02 ‘invaded’), 44 (’03 ‘intact’), 46 (’03 ‘invaded’), 39 (’05 ‘intact’), 51 (’05 ‘invaded’). No data were collected in 2004 ................................................................................. 33  Figure 3.2: Standardized effect sizes generated from C-score analysis, comparing 30,000 simulated null matrices to the observed matrix for each year and status. ‘Intact’ plots (black bars) for all year classes had significantly higher C-scores than expected by chance at p=!0.004. ‘Invaded’ plots (white bars) had significantly higher C-scores in 2001, but not so in 2002, 2003 and 2005. Standardized effect sizes between 2.0 and -2.0 indicate non-significant deviations from predictions at p=!0.05 ........................................................ 34  Figure 3.3: Annual and early-year (January – April) precipitation in Penticton (black bars are annual data, patterned bars are early-year data) .....................................  37             viii Acknowledgements   First and foremost, I would like to extend sincere thanks to my supervisors, Judy Myers and Mark Vellend, for their support, encouragement and patience in aiding the development of this research project from a few initial interests and ideas to a complete M.Sc. thesis. I also extend my thanks to Roy Turkington, who proved to be a fantastic committee member whose input greatly improved the quality of the final product.  I thank Michelle Tseng for her help in developing my statistical knowledge, and fellow lab-mates Michelle Franklin, Caroline Jackson, Amanda Brown, Rana Sarfraz and Andrea Stephens for their valuable input on all things biological throughout my time at UBC. I thank Andrea, in particular, for her useful advice and shared driving from Vancouver to my field sites in the Okanagan Valley, where she also worked. Adrian MacKay was an excellent summer assistant, working with enthusiasm on the many varied tasks he was given, while Ishwarya Chaitanya also contributed to data collation. I also thank Linda Edwards, who allowed me to use her house in Keremeos as a base from which to collect field data over the course of my studies.  I am forever grateful to my parents, Jill and Bob, for their ongoing support in all that I do, and for their light-hearted approach to scientific matters. On a similar note, I extend great appreciation to my girlfriend Ashley, who provided me with the inspiration to complete this project, for looking after me brilliantly, and for her ability to make research-related frustrations seem unimportant whenever setbacks occurred.  This research was financially supported by a Canada Memorial Foundation Scholarship, a University Graduate Fellowship Award from the University of British Columbia, and by National Science and Research Council Awards awarded to Judy Myers.   1 Chapter 1: General introduction – cheatgrass, Bromus tectorum Linnaeus    Non-native, introduced plant species are one of the major threats to global biodiversity (Pejchar & Mooney 2009, Downey et al. 2010). Although not always solely problematic (De Wit et al. 2001), their spread and proliferation often exerts considerable negative effects on ecosystem processes such as nutrient cycling (Gordon 1998) and pollination (Spira 2001). These in turn tend to have hugely detrimental financial consequences (Pimentel et al. 2005).  The Eurasian annual cheatgrass, Bromus tectorum Linnaeus, is one such introduced species, whose rapid spread throughout western North America has had huge consequences for the arid grassland and shrubsteppe habitats in which it has flourished (Mack 1981, Upadhyaya et al. 1986). Accidentally introduced approximately 150 years ago, probably via contaminated grain, cheatgrass has transformed communities from biologically diverse, perennial-rich assemblages to species-poor aggregations; sometimes its dominance is so great that it exists in monoculture (Bradley & Mustard 2006, Monaco et al. 2003). Cheatgrass possesses a number of characteristics that have allowed it to establish and spread so extensively; it is adapted to arid conditions but competes well for available water and other nutrients, using all available resources to produce high quantities of seed (Young & Allen 1997). It can germinate in the autumn if conditions are favourable (Bradley & Mustard 2006), allowing it to monopolize resources the following spring when perennials are only just emerging. Seeds can also lie dormant for many years (Smith et al. 2008) and germinate over a wide range of temperatures (Meyer et al. 1997). Cheatgrass is also well adapted to fire, because seeds can exist in the seed bank for many years, enabling swift re-colonization following burning; within two years of fire, it can recover to occupy >99% of a seed bank (Humphrey & Schupp 2001). The problem of cheatgrass is thus ecological and economic; wherever it is common, habitats are typically depauperate in terms of biodiversity and important ecosystem processes are negatively impacted (DiTomaso 2000). In turn, this requires significant expenditure in attempting to restore habitats, prevent further invasion, and control wildfires, which occur in affected areas at frequencies far exceeding historical  2 benchmarks (Haubensak et al. 2009). Although cheatgrass provides some nutrition to cattle that graze rangelands in early spring, it senesces before mid-summer, at which point it has relatively low nutritive value, is unpalatable to livestock (Currie et al. 1987) and so sharp that it causes injury (Morrow & Stahlman 1984). Restoring invaded communities is a real challenge, but a number of techniques are used in attempts to do so. These include prescribed burns (Brooks et al. 2004), the use of herbicides (Davison 2007), heavy grazing and/or mowing (Davies et al. 2009), aggressive seeding of other introduced species (Goodrich & Rooks 1999), and seeding of native, perennial species mixes (Cox & Anderson 2003). However, cheatgrass is so well adapted to invading arid habitats that restoration attempts often fail to reduce spread or encourage the re-colonization of favoured species (Clements & Young 2007). Despite this grim reality, there is sufficient hope that under some circumstances, cheatgrass can be suppressed (Pellant 1996,), usually when it experiences competition from other species (Ott et al. 2003). The major problem in many of these restoration attempts is that the mix of environmental and biotic components is unique to each habitat, which results in successes being context dependent; species that compete successfully and resist invasion in one habitat do not always do so in a different one. However, although restoration attempts almost always fail to eliminate cheatgrass completely, promotion of native species that reduce its dominance can be highly cost-effective as they invest in biodiversity and fire prevention (Epanchin-Niell et al. 2009).   For these reasons, it is important that the environmental variables and species- specific relationships that appear to explain much of the variation in the abundance and distribution of cheatgrass are investigated on local scales. This information should be of use from a restoration perspective because different management plans can theoretically be applied on such scales. In this thesis, I present cheatgrass abundance and distribution analyses, and detailed co-occurrence analyses that aimed to provide such information for grasslands in the Okanagan Valley of British Columbia, Canada. Cheatgrass has been present in the warmer, drier parts of Canada for well over 100 years (Valliant et al. 2007), but it has become more abundant and widely distributed in pastures of the Okanagan Valley recently (Stephens et al. 2009). Because these grasslands also feature an incredibly high diversity of plant and animal species (Bezener  3 et al. 2004), many of which are endangered, it is of critical importance that restoration attempts are actively pursued in this region.                               4 Chapter 2: Variations in the abundance and distribution of cheatgrass relative to biotic factors and a potential link to soil chemistry  Introduction  Cheatgrass, Bromus tectorum Linnaeus, is a Eurasian annual grass that has spread rapidly throughout western North America since its accidental introduction approximately 150 years ago (Mack 1981, Upadhyaya et al. 1986, D’Antonio & Vitousek 1992). The plant possesses many attributes that make it a successful invader; it produces a high number of seeds, which can lie dormant for many years before germinating under favorable conditions (Smith et al. 2008). Cheatgrass also grows quickly and can germinate over a wide range of temperatures (Meyer et al. 1997). It can thus monopolize soil moisture and nutrients by germinating and/or growing in periods when many other plants do not. For example, it can germinate in the autumn, over-winter in a semi- dormant state, and grow rapidly when precipitation arrives in the spring, The proliferation of cheatgrass is a concern for a multitude of reasons. Firstly, the plant is a strong competitor and can quickly establish mono-specific stands in formerly productive and diverse desert and grassland habitats, which consequently suffer in terms of ecosystem processes (DiTomaso 2000, Humphrey & Schupp 2001). Secondly, invaded areas tend to experience wildfires much more frequently than historically normal. Following fire, cheatgrass re-colonizes and spreads more efficiently than other species, essentially creating a positive feedback (Melgoza et al. 1990, Rowe & Brown 2008). Fires devastate diverse habitats in a matter of hours, cause major safety concerns, and are expensive to regulate. Cheatgrass is also common on many pastures used for cattle forage, but the plant provides only average nutrition in the spring and seeds become so sharp as to be harmful to cows when drier later in the season (Morrow & Stahlman 1984). Cheatgrass has been characterized for some time as one of the most problematic invaders in western USA, but it also occurs in warmer areas of Canada, such as the Okanagan Valley in British Columbia. This region is located approximately 300km east of the Pacific Ocean and features a valley that runs north from the American border for approximately 160km (Rayne et al. 2009). Summers are long and dry, with temperatures reaching over 40°C at their peak. Known for its warm temperatures that promote arid grasslands, it provides almost ideal habitat for the establishment and spread of cheatgrass,  5 especially because precipitation is typically greatest in autumn and spring, when seeds can germinate and shoots can grow, respectively. Grasslands in the region feature a mixture of native and introduced species. Plant diversity can be high in intact habitats, which is important because over 30% of British Columbia’s threatened species depend upon these grasslands (Crowe 2008). Among the better-known introduced species are diffuse knapweed (Centaurea diffusa Lamarck) and crested wheatgrass (Agropyron cristatum L.). Meanwhile, native perennial species such as bluebunch wheatgrass (Pseudoroegnaria spicata Pursh) and Sandberg bluegrass (Poa secunda J. Presl) are favored in restoration attempts and by ranchers. Numerous shrubs and trees are also present, including Ponderosa pine (Pinus ponderosa C. Lawson), a species that is numerous in these grasslands.  This makes it of particular interest because large trees can be critical in shaping community structure and diversity in a range of habitats via direct and indirect mechanisms. For example, trees can provide shade and shelter from environmental stressors (Garrett et al. 2004), and influence soil nutrient and moisture composition (Eaton & Farrell 2004, Maestre et al. 2009). In this observational study, five sites of varying characteristics throughout the Okanagan Valley were chosen to assess spatial variation in cheatgrass abundance. Proximity to focal pine trees, number of other species present, aspect, soil pH, soil moisture, and designation of co-occurring species guilds were analyzed as potential explanatory variables. Although there is no consensus on whether native diversity per se affects the likelihood of invasion into an area (see Davis et al. 2000 and Sax 2002 for differing opinions), cheatgrass appears to experience species-specific competition and facilitation in these habitats (Booth et al. 2003, Chambers et al. 2007). As such, these analyses were conducted to provide information of use to those charged with restoring these landscapes to something approaching their native-dominated states. For example, artificial seeding of other species is a common technique used to stop the invasion and spread of cheatgrass, and selective herbicide application is another option. However, there is no consensus on what constitutes an ideal seed mix (Cox & Anderson 2004) because documentation of the most effective competitors is limited, contradictory, or site-specific. Cheatgrass has been shown to be vulnerable to herbicide  6 (Pellant et al. 1999), but applications are expensive over large spatial scales and have the potential to negatively affect non-target species (Crone et al. 2009). Specifically, strong associations with physical and biotic presences could indicate the factors most important in dictating its abundance and distribution, paving the way for the use of more effective native seed mixes or spatially directed herbicide applications.                                         7 Methods  Cheatgrass abundance and correlations with biotic factors  Cheatgrass abundance in relation to Ponderosa pine trees was measured by counting the number of individual stems per 0.25m2 quadrat (0.5m x 0.5m) along 10m transects. The number of stems was deemed a more reliable, quantitative measure than percentage cover, because the latter often relies on visual judgment, which is typically more prone to error. Transects were established by walking 10m in four directions (north, south, east, west) from a focal pine tree. To prevent groups of trees confounding the distribution of cheatgrass, transects were only established when the focal tree was the only one in the given area (10m x !r2). Quadrat sampling was conducted so that samples of cheatgrass abundance were recorded at 2m intervals along each transect (adjacent to a pine tree (0m), 2m, 4m, 6m, 8m, 10m). In total, 16 trees were used for each of five sites from which data were collected. In addition to cheatgrass abundance, the total number of ‘other’ plant species was recorded per quadrat; these other species were simply recorded as ‘present’. Basic plant functional group/growth form was noted, enabling consideration of positive/negative relationships between cheatgrass abundance and (1) other introduced species and (2) native bunchgrasses. To provide control comparisons for each analysis, data were collected in the same manner along transects that did not have pine trees. Control transects were established with 11 ‘start points’ for each of the five sites. Each ‘start point’ (0m) was chosen randomly.  Study sites and sampling dates   Data were collected on June 15 – June 21, 2009, June 28 – July 5, 2009, and July 13 – July 17, 2009. Table 1.1 provides basic habitat information for each site, including the most common species and any other notable characteristics.    8 Table 2.1: Elevation and habitat characteristics of the five study sites.  Statistical analyses   All data were analyzed using the statistical software package ‘R’. Linear models were used to determine relationships between cheatgrass abundance and the other factors considered (distance from focal pine trees, number of ‘other’ species present, aspect). To avoid pseudo-replication, pine tree transects were blocked for each tree in site analyses to give n=16 for each distance at each site (total n=96 per site); control transects were also blocked for ‘start points’ in the same manner to give n=11 for each distance at each site (total n=66 per site). Aspect was a non-significant factor in accounting for cheatgrass abundance (see Results), so means were taken from the four aspect-based measurements of each distance per transect.  Soil pH and soil moisture correlations with focal pine trees   To consider whether potential differences in soil moisture or soil pH were associated with the observed differences in cheatgrass abundance as distance from focal Site name GPS location; elevation Common species; notable habitat characteristics Anarchist Mountain 49°00’59.46”N, 119°16’48.01’W; 3720 ft Ponderosa pine, Douglas fir, aspen; diffuse knapweed common Okanagan Mountain 49°46’50.21”N, 119°35’20.57’W; 1380 ft Spruce species, Ponderosa pine, Douglas fir; prone to fire Oliver 49°09’09.61”N, 119°32’14.33”W; 1350 ft Mainly Ponderosa pine; close to small roads and human disturbance Vaseux Lake 49°17’48.75”N, 119°31’34.54’W; 1270 ft Ponderosa pine, bunchgrasses; recovering from major fire in 2003 White Lake 49°19’22.45”N, 119°37’53.91’W; 1850 ft Diffuse knapweed, bunchgrasses, Ponderosa pine.  9 pine trees increased (see Results), soil samples were taken from the same transects as those previously used. To obtain each sample, a garden trowel was used to dig to a depth of ~20cm (the depth at which cheatgrass roots were found to have reached), at which point a coring device was used to collect 40-50g of soil. Samples were placed in plastic beakers, sealed with wax tape to prevent subsequent gaseous exchange, and stored in a dark, cool environment until being analyzed. Soil samples were only taken from southerly aspects. Eight of the 16 trees used to establish each original transect were randomly chosen at each site, and samples were taken for all distances (0m, 2m, 4m, 6m, 8m, 10m). As such, 8 samples were provided for each distance at each site (n= 48 samples per site, total n=240). To test for differences in soil pH, 20g of fine soil was added to 20ml dH2O and shaken vigorously for 2-3 minutes until dissolved. At this point, a pH meter was used to record the pH of each sample. To test for differences in soil moisture content, a digital soil moisture probe that produced a % moisture content reading (integers only) for each tested sample was used. To provide control comparisons, data were collected in the same manner but along transects that did not have pine trees. Five of the 11 control transects previously established at each site were sampled and samples were taken for all distances. As such, excepting Okanagan Mountain where control samples were not taken, 5 samples were provided for each distance at each site (n=30 samples per site, total n=120). All soil samples were collected on July 15 – July 17, 2009.  Statistical analyses   All data were analyzed using the statistical software package ‘R’. Linear models were used to determine relationships between distance from focal pine trees and (1) soil pH, and (2) soil moisture.  Cheatgrass abundance and correlations with functional groups comprising ‘other species’   There was a highly significant negative correlation between cheatgrass abundance and the number of other species present in quadrats (see Results), so a more detailed  10 investigation into the functional group/growth form of species in these interactions was conducted. Of the individuals categorized as ‘other’ species within quadrats, two further distinctions were made: (1) other introduced species, and (2) native bunchgrasses. The species characterized as (1) other introduced species were: diffuse knapweed (Centaurea diffusa), crested wheatgrass (Agropyron cristatum), Dalmatian toadflax (Linaria dalmatica L.), Tansy ragwort (Senecio jacobaea L.), up to 3 non-cheatgrass Bromus spp., 2 non-native fescues (Vulpia spp.) and bulbous bluegrass (Poa bulbosa L.). The species characterized as (2) Native bunchgrasses were: bluebunch wheatgrass (Pseudoroegnaria spicata), sandberg bluegrass (Poa sandbergii J. Presl), basin wildrye (Leymus cinereus Scribn. & Mehr), Idaho fescue (Festuca idahoensis Elmer), western needlegrass (Stipa occidentalis Thurb.) and squirreltail (Elymus elymoides Raf.).   Statistical analyses  All analyses were conducted in the statistical software program ‘R’. The percentage component of ‘other’ species made up of (1) other introduced species, and (2) native bunchgrasses, was calculated. To determine relationships between cheatgrass abundance and percentage components of co-occurring species, linear models were used. Pine tree data were obtained from all quadrats (n=96 per site), whereas control data were obtained for all the quadrats used from 5 of 11 null transects per site (n=30 per site).                  11 Results  Correlations between cheatgrass abundance and distance from focal pine trees      Cheatgrass abundance declined significantly (F=307.12, p=<0.0001, n=480) as distance from focal pine trees increased (Figure 2.1). This pattern was consistent at all five sites. However, regression coefficients differed for each site, such that plots at Vaseux Lake showed the greatest rate of decline in cheatgrass abundance with distance and Oliver the least (Table 2.2). Regardless of distance, cheatgrass abundance was greatest at Vaseux Lake and lowest at Anarchist Mountain (Figure 2.1). Among-site Tukey’s post hoc comparisons indicated that all five sites differed significantly at p=<0.0001, except for the Okanagan Mountain and White Lake comparison (p=0.29). For the control data, cheatgrass abundance was not related to distance along transects at any of the sites (Anarchist Mountain p=0.18 to Vaseux Lake p=0.65).                Figure 2.1: Cheatgrass declines in abundance (number of stems per 0.25m 2  quadrat) with distance from focal pine trees. Different symbols and colors represent the five study sites, n=96 per site.      12 Table 2.2: Site-specific linear equations describing significant rates of decline in cheatgrass abundance (number of stems per 0.25m 2  quadrat) with distance from focal pine trees. For all sites, degrees of freedom = 95. Linear equations for each site should be read as the number of cheatgrass stems adjacent (0m) to a pine tree – number of cheatgrass stems/m from a pine tree.   Correlations between cheatgrass abundance and number of ‘other’ species present   Cheatgrass abundance declined significantly (F=407.31, p=<0.0001, n=480) as the number of ‘other’ species present within quadrats increased (Figure 2.2). This pattern was consistent at all five sites. Regression coefficients differed between sites, such that plots at Vaseux Lake and White Lake showed the greatest rates of decline in cheatgrass abundance as the total number of ‘other’ species present increased (Table 2.3). Among- site Tukey’s post hoc comparisons indicated that all five sites differed significantly at p=<0.0001, except for the Oliver and Okanagan Mountain comparison (p=0.034). For the control data, cheatgrass abundance also declined significantly (F=386.21, p=<0.0001, n=330) as the number of ‘other’ species present in quadrats increased (Figure 2.3). This pattern was consistent at all sites. Regression coefficients differed between sites, such that plots at Vaseux Lake and Oliver showed the greatest rate of decline in cheatgrass abundance as the total number of ‘other’ species present increased (Table 2.3). Among-site Tukey’s post hoc comparisons indicated that all sites differed significantly from each other at p=<0.0001.       Site Linear Equation p-value R 2 value F-statistic Anarchist Mountain 24.83 – 2.64/m 0.005 0.48 89.47 Okanagan Mountain 35.87 – 3.22/m 0.009 0.37 57.22 Oliver 27.18 – 2.39/m 0.004 0.43 71.64 Vaseux Lake 52.16 – 4.52/m 0.002 0.49 94.01 White Lake 35.77 – 3.60/m 0.002 0.50 97.81  13                 Figure 2.2: Cheatgrass declines in abundance (number of stems per 0.25m 2  quadrat) as the number of other species present within a quadrat increases. Different symbols and colors represent the five study sites, n=96 per site.  Table 2.3: Site-specific linear equations describing significant rates of decline in cheatgrass abundance (number of stems per 0.25m 2 quadrat) with increasing number of other species. For all sites, degrees of freedom = 95 (65 for control data). Linear equations for each site should be read as the number of cheatgrass stems – number of cheatgrass stems/other species found.  Site, data type Linear Equation p- value R 2 value F- statistic Anarchist Mountain 27.33 – 2.03/os 0.008 0.34 48.49 Anarchist Mountain, control 30.11 – 2.74/os 0.011 0.24 46.30 Okanagan Mountain  42.92 – 4.34/os 0.014 0.22 28.51 Okanagan Mountain, control 34.12 – 2.91/os 0.007 0.35 49.14 Oliver  42.44 – 4.66/os 0.001 0.55 118.1 Oliver, control 46.19 – 5.78/os <0.001 0.58 123.2 Vaseux Lake 63.32 – 6.01/os 0.004 0.47 85.71 Vaseux Lake, control 51.13 – 4.41/os 0.004 0.43 84.62 White Lake 51.44 – 5.91/os <0.001 0.66 186.01 White Lake, control 39.52 – 3.59/os 0.006 0.39 70.40   14                      Figure 2.3: Cheatgrass declines in abundance (number of stems per 0.25m 2  quadrat) as the number of other species present within a quadrat increases in control (no pine tree) transects. Different symbols and colors represent the five study sites, n=96 per site.   Cheatgrass abundance by aspect   Cheatgrass abundance did not differ significantly with aspect at any of the study sites (p=0.11 at Vaseux Lake to p=0.65 at Oliver). For control data, cheatgrass abundance did not differ significantly with aspect at any of the study sites (p=0.21 at White Lake to p=0.63 at Anarchist Mountain).  Correlations between soil pH and distance from focal pine trees   Soil pH increased significantly with distance from focal pine trees (F=478.3, p=<0.0001, n=240). This pattern was consistent at all five sites (Figure 2.4). Regression   15 coefficients differed among sites, such that plots at Vaseux Lake and Okanagan Mountain showed the greatest rate of increase in pH with distance (Table 2.4). Among-site Tukey’s post hoc comparisons indicated that all sites differed significantly at p=0.01 or greater, except for the Vaseux Lake and Okanagan Mountain comparison (p=0.34). Qualitatively, soil pH was weakly acidic adjacent (0m) to focal trees at all sites and remained so (<7.0) at 10m at Anarchist Mountain, Oliver and White Lake. At Vaseux Lake and Okanagan Mountain, soil pH became neutral at ~7.5m from focal pine trees and was weakly basic at 10m (Figure 2.4). For control data, soil pH was not significantly correlated to distance from ‘start points’ at any of the four sampled sites (p=0.26 at Anarchist Mountain to p=0.96 at Vaseux Lake).                   Figure 2.4: Changes in soil pH with distance from focal pine trees at all five sites. Different symbols and colors represent different sites, n=48 per site.     16 Table 2.4: Site-specific linear equations describing significant rates of increase in soil pH with distance from focal pine trees. For all sites, degrees of freedom = 47. Linear equations for each site should be read as the soil pH adjacent (0m) to a pine tree + pH units/m from a pine tree.  Site Linear Equation p-value R 2 value F-statistic Anarchist Mountain 6.22 + 0.04/m 0.004 0.54 71.10 Okanagan Mountain 6.47 + 0.07/m 0.004 0.53 69.95 Oliver 6.64 + 0.05/m 0.008 0.46 52.91 Vaseux Lake 6.48 + 0.07/m 0.003 0.60 89.98 White Lake 6.68 + 0.03/m 0.012 0.39 38.87   Correlations between soil moisture and distance from focal pine trees   Soil moisture content increased significantly with distance from focal pine trees (F=136.7, p=<0.004, n=240). This pattern was consistent at four of the five sites (Table 2.5); Anarchist Mountain was the non-significant exception (p=0.53). Regression coefficients differed among sites, such that plots at Vaseux Lake and White Lake saw the greatest rate of increase with distance (Table 2.5). Among-site Tukey’s post hoc comparisons were significant at p=<0.001 or greater, except for the Oliver and Okanagan Mountain comparison (p=0.016). Anarchist Mountain was not included in comparisons. For control data, soil moisture was not significantly correlated with distance from transect ‘start points’ at any of the four sampled sites (p=0.39 at White Lake to p=0.81 at Anarchist Mountain).  Table 2.5: Site-specific linear equations describing rates of increase in soil moisture (%) with distance from focal pine trees. For all sites, degrees of freedom = 47. Linear equations for each site should be read as the soil moisture (%) adjacent (0m) to a pine tree + moisture units (%)/m from a pine tree. Soil moisture (%) at Anarchist Mountain was not significantly correlated with distance from focal pine trees (p=0.53).  Site Linear Equation p-value R 2 value F-statistic Anarchist Mountain 1.48 + 0.09/m 0.530 0.01 0.84 Okanagan Mountain 1.87 + 0.07/m 0.022 0.08 7.14 Oliver 1.92 + 0.05/m 0.035 0.07 4.53 Vaseux Lake 1.34 + 0.13/m 0.008 0.31 27.73 White Lake 1.70 + 0.13/m 0.017 0.24 19.88    17 Correlations between cheatgrass abundance and proportions of functional groups comprising ‘other species’   The percentage component of other species comprising native bunchgrasses increased significantly as cheatgrass abundance increased (F= 114.5, p<=0.001, n=480). This pattern was consistent at four of the five sites (Table 2.6); Vaseux Lake was the non- significant exception (p=0.18). Regression coefficients differed among sites, but R2 values were relatively low for all sites and among-site comparisons were not made. For control data, the percentage component of ‘other’ species comprising native bunchgrasses was positively correlated with cheatgrass abundance at Anarchist Mountain (F=11.23, p=0.002, n=96, R2=0.09. There were no other significant relationships (Oliver p=0.14 to White Lake p=0.24). The percentage component of other species comprising native bunchgrasses declined significantly as cheatgrass abundance increased (F=67.3, p=0.034, n=480). However, this pattern was only consistent at three sites (Table 2.6); Okanagan Mountain (p=0.38) and White Lake (p=0.57) were the non-significant exceptions. Regression coefficients differed among sites, but R2 values were relatively low for all sites and among-site comparisons were not made. For control data, the percentage component of other species comprising introduced species was not significantly correlated with cheatgrass abundance (p=0.064 at White Lake to p=0.47 at Anarchist Mountain).                    18 Table 2.6: Site-specific descriptions of correlations between cheatgrass abundance and percentage component of other species comprising native bunchgrasses (‘NB’) and introduced species (‘IS’). For all sites, degrees of freedom = 95. Proportion increase/decrease for each site should be read as the proportion (%) increase/decrease of other species comprising species belonging to each functional group per stem of cheatgrass.                           Site, data type Proportion (%) increase/decrease p- value R 2 value F- statistic Anarchist Mountain, ‘NB’ + 0.17/stem 0.038 0.03  2.54 Anarchist Mountain, ‘IS’ – 0.24/stem <0.001 0.14 14.62 Okanagan Mountain, ‘NB’  + 0.09/stem 0.027 0.03 2.81 Okanagan Mountain, ‘IS’ – 0.09/stem 0.380 <0.01 0.28 Oliver, ‘NB’  + 0.14/stem 0.018 0.06 5.72 Oliver, ‘IS’ – 0.18/stem <0.001 0.04 1.05 Vaseux Lake, ‘NB’ + 0.03/stem 0.180 0.01 0.39 Vaseux Lake, ‘IS’ – 0.20/stem 0.013 0.04 6.97 White Lake, ‘NB’ + 0.12/stem 0.007 0.08 9.63 White Lake, ‘IS’ – 0.05/stem 0.570 <0.01 0.24  19 Discussion    A highly significant positive relationship between pine trees and cheatgrass was detected at all five sites despite different habitat characteristics being evident. This pattern did not occur in control transects without a focal pine tree. Thus, the presence of pine trees is likely to influence the abundance of cheatgrass and be critically important in mediating the characteristics of herbaceous communities in Okanagan grasslands. Despite the commonality of the trend, rates of decline in cheatgrass abundance with distance were, unsurprisingly, site-specific. Cheatgrass abundance declined most with distance at Vaseux Lake, where it was most abundant adjacent to focal trees; abundance declined least with distance at Oliver and Anarchist Mountain, where it was least abundant at pine tree bases. The high relative abundance at Vaseux Lake is likely a result of a recent wildfire (2003) having affected this site. The fire cleared vegetation and opened up colonization space – something cheatgrass can quickly take advantage of due to its large seed production and competitive ability (Smith et al. 2008). Cheatgrass is a rapid colonizer and a suite of studies link increases in abundance with increased fire frequencies in grassland systems (Melgoza et al. 1990, Rowe & Brown 2008). Trees and shrubs can have facilitative effects – direct and indirect - on many plant species in diverse habitats; resource modification, propagule concentration and mycorrhizal associations are just some of the mechanisms these species mediate to promote such positive interactions (Callaway 1995). These associations are particularly common in arid environments, which are otherwise stressful to young plants, with other annuals and grasses among those found to associate positively with trees (see Yavitt & Smith 1983, Frost & McDougald 1989, Callaway et al. 1991 for examples). A recent study considering the impact of shrubs on the occurrence of cheatgrass in California reported a similar positive relationship (Griffith 2010). In that experiment, cheatgrass composed 68% of shrub canopy space but only 11% of inter-shrub space, while the abundance of other herbaceous species did not differ between microhabitats. The study also showed that soil chemistry differed in the two microhabitats, which in turn suggests a mechanism for the observed result. However, Griffith’s (2010) most important finding was that increased survival and growth rates of cheatgrass under shrub canopy  20 translated into a positive effect at the population level. As such, the positive association between cheatgrass and focal pine trees in Okanagan grasslands has obvious implications for those charged with their restoration. Herbicide application has been effective in reducing cheatgrass abundance (Pellant et al. 1999, Vollmer & Vollmer 2008), but because it is expensive and can have negative effects on non-target species, it is not always considered as a management option. However, by applying herbicide selectively at focal pine tree bases, this could save a lot of money and minimize the risk of negatively impacting native species. Because cheatgrass density was reduced by 91% in plots treated with OUST Herbicide in the Great Basin, USA, (Pellant et al. 1999) there is clearly hope that focused applications could have a population level effect in reducing cheatgrass abundance – especially if patches in inter-tree space are sink populations. Dunning et al. (1992) state that species can be maintained at a landscape scale despite migrants coming from only small source populations. Whether patches away from trees represent sink populations in Okanagan grasslands is unknown, but examples of grasses persisting in habitats via source-sink mechanisms do exist, suggesting this is a possibility. For example, Kadmon & Shmida (1990) reported that the desert grass Stipa capensis exists in unfavorable slope (sink) habitats due to emigration from productive wadi (source) habitats nearby.  A highly significant negative relationship between cheatgrass abundance and the number of other species present in quadrats was found. This pattern was consistent at all five sites in pine tree and control transects. It was assumed most inter-specific interactions in these grasslands would result in competition; therefore, a more diverse community was expected to see relatively low cheatgrass abundance. It is also possible that environmental variations resulted in the observed patterns, with cheatgrass abundance being highest in regions other species were not suited to, and vice versa. This seems less likely, because cheatgrass is suited to similar conditions as most other species seen in these grasslands, had invaded most patches in the landscape, and appears capable of invading any community. One further possibility is that cheatgrass was suppressing other species and that plots where it was abundant thus tended to feature fewer other species. The difficulty in establishing which of these three potential explanations is the correct one highlights the care required in analyzing observational data.  21 Importantly, however, some sites saw greater rates of decline with other species in pine tree transects (Okanagan Mountain, Vaseux Lake, White Lake), while others saw greater rates of decline in control transects (Anarchist Mountain, Oliver). This suggests other species diversity did not influence cheatgrass uniformly, and that the identity of the other species present in quadrats were likely important in mediating cheatgrass abundance. This was expected, because some species are more competitive than others whereas some might even facilitate cheatgrass. For example, it is known that species such as squirreltail (Elymus elymoides) tend to compete more effectively with cheatgrass than others in Utah (Ott et al. 2003), while bunchgrasses such as the introduced crested wheatgrass (Agropyron cristatum) are generally believed to be more effective than others at resisting initial invasion (D’Antonio & Thomsen 2004). Quantifying the nature and strength of interaction other species exert on cheatgrass in the Okanagan should show which species should be preferentially seeded as competitors in these habitats. Aspect had no discernible effect on cheatgrass abundance at any site. Conditions can be warmer on south-facing slopes than on north-facing slopes due to solar radiation irregularities, and these variations can result in differences in nutrient cycling and plant growth (Aragon et al. 2007, Kota et al. 2007). Thus, it would not be surprising if aspect influences the abundance and distribution of cheatgrass in Okanagan communities at more northerly latitudes where solar radiation is a more limiting resource.  Soil pH and soil moisture inferences  Soil pH increased significantly with distance from focal pine trees at all five sites. This pattern did not occur in control transects without a focal pine tree, suggesting the pine trees were directly or indirectly responsible for increasing acidity in the soil environment. Rates of soil pH increase with distance were site-specific; Vaseux Lake and Okanagan Mountain had the highest rates of increase, and White Lake and Anarchist Mountain had the lowest. Because the general pattern was highly significant at all five sites, it suggests that pine trees affect soil pH at least as far as 10m away. This again underlines the likelihood that they exert strong effects on ecological processes in the region.  22 Having characterized the relationship between cheatgrass abundance and proximity to pine trees, it is recognized that cheatgrass (and not pine trees) could also be responsible for the observed differences in soil pH. This seems less likely because cheatgrass roots are relatively small and have less scope for altering environments to the same degree as large, mature pine trees that drop needles at their bases. In a Swedish study focusing on the impacts of Norway spruce and Scots pine, concentrations of nutrients in needles varied in a complex manner with season, climate and soil characteristics (Thelin et al. 1998). These nutrients (e.g. N, K, Cu, Zn, Mg, Ca), affect soil pH and the availability of nutrients and other compounds to plants. Importantly, pine needles have long been associated with reducing soil pH and even possess acid-forming compounds and resins that resist decay (McConnell & Smith 1971). Over forty years ago, Moir (1966) linked the rarity of native species under pine canopies to low levels of soil nitrogen, which can be driven by acidic input and poor microbial performance in breaking down litter. Although these results are only correlations, there is a suggestion that soil pH could influence cheatgrass abundance. This is because soil pH was significantly lower at given distances from a focal pine tree at Anarchist Mountain than at the other sites, and cheatgrass abundance was lowest at this site. Even a generalist invader such as cheatgrass will likely be more successful in environments displaying certain soil pH characteristics. Soil moisture (%) increased significantly with distance from focal pine trees but this only occurred at four sites, with the other (Anarchist Mountain) showing no significant pattern. Control transects showed no significant pattern at any site, again suggesting the trees affected soil chemistry. Rates of moisture increase (%) with distance were site-specific; Vaseux Lake and White Lake saw the greatest rates of increase, and these sites were the driest adjacent to focal pine trees. It is possible that cheatgrass is responsible for altering soil moisture as it can virtually monopolize moisture early in the growing season when competing with native plants. However, it seems the much larger pine trees would be more likely responsible for the pattern. Although it might be expected that soil near trees would contain more moisture than that in open, arid environments, the literature suggests relationships are context dependent. An experiment in a New Zealand habitat featuring Pinus radiata and a  23 ryegrass-dominated under-story found that the two species exert a complementary effect with regard to soil moisture over three or four seasons, but then the balance shifts in favor of the pine trees (Pollock et al. 2009). At this point their roots can dominate to the extent that soil moisture is almost entirely monopolized. Pollock et al. (2009) used young trees whereas the pine trees used in this study were of considerable age, suggesting they would exert an even greater influence. In contrast with the pattern in Okanagan grasslands, Fernandez et al. (2006) showed a positive balance exists between competition and facilitation for Ponderosa pine and a grass species, Festuca pallescens, in Patagonia. Cheatgrass abundance typically increases following seasons of greater than average precipitation (Meyer et al. 2001, Whittaker et al. 2008, Stephens et al. 2009), but it is generally more successful long-term in arid environments where water is a limiting resource. This is because once it is established it is hardy, an efficient user of soil moisture, and extremely hard to displace. As such, the potential importance of the relatively dry soils at Vaseux Lake should not be dismissed; this result might indicate a site that has seen proliferation of cheatgrass, which has effectively monopolized soil moisture and established a community that will be difficult for other species to invade. This study only provided readings for soil pH and soil moisture at one part of the summer, and it is known that soil chemistry changes over time, sometimes suddenly due to precipitation and/or mineralization. As a result, the patterns observed might vary over the course of a season. Also, differences in soil moisture and pH are linked because variations in moisture affect the relative availability of charged nutrients in soils. However, the variations suggest pine trees influence microhabitat characteristics and these are effectively exploited by cheatgrass.  Functional group correlations with cheatgrass   Cheatgrass abundance was positively correlated with the proportion of native bunchgrasses comprising other species but negatively correlated with the proportion of introduced species comprising other species. However, significant relationships did not exist for all sites, and R2 values only described a small degree of variation. Nevertheless, the existence of these trends requires more detailed analysis on a species-specific level;  24 clearly, information regarding the identity of species likely to compete well with cheatgrass in these habitats would be very valuable.  The positive correlation between cheatgrass abundance and native bunchgrasses could suggest that environmental conditions are suitable for both. Because exotic species richness is often negatively correlated with that of native plants (Vidra et al. 2006), this is potentially encouraging. Native plants are valued as they provide good quality forage for cattle (Morrow & Stahlman 1984) whereas grassland ecosystems featuring introduced species can be less productive and are typically less diverse (Christian & Wilson 1999). Furthermore, native perennial species can be effective at resisting the invasion and spread of introduced species, especially if competition occurs between mature plants (Corbin & D’Antonio 2004). As such, there is cause for optimism that the sites surveyed in this study possess communities that are likely to be more resistant to further invasion than many of the more susceptible areas in Western North America. But, as Getz & Baker (2008) point out, positive species associations can indicate facilitation, not competition; the results of this study cannot distinguish the exact nature of the relationship. The negative correlation between cheatgrass abundance and other introduced species is also interesting. Many plant community-based studies have found evidence to suggest introduced/invasive species tend to have positive effects on one another (Simberloff & Von Holle 1999). The authors, in describing these relationships, coined the phrase ‘invasional meltdown’ to describe the facilitative effects introduced species can have on one another in new habitats. As an example, the introduced perennial grass, Schizachyrium condensatum, invaded Hawaiian woods and grasslands and altered the fire regime to a more frequent than normal state, which in turn led to declines in native plant species and aided the invasion of a fellow introduced grass species, Melinis minutiflora (Hughes et al. 1991). However, other studies show it is not uncommon for introduced species to have negative impacts on other introduced species (e.g. cheatgrass competes with crested wheatgrass, Ott et al. 2003).      25 Conclusions  This study has provided a number of results that have formed hypotheses that are testable in situ, with the ultimate aim being the provision of information that can be put into practice to aid restoration of cheatgrass-invaded sites in the Okanagan. Perhaps the most important result is the positive association between focal pine trees and cheatgrass. From an applied perspective, it suggests the selective use of herbicide around tree bases could reduce cheatgrass abundance. Among the results that sprung from soil pH and soil moisture analyses were that pine trees were associated with more acidic and drier soils. This might indicate that cheatgrass is more competitive in these conditions, which would have further implications for management programs. Cheatgrass was generally positively associated with native bunchgrasses and negatively so with introduced species. In conjunction with the result that showed cheatgrass was less abundant as the number of other species increased, this suggests some species are more effective competitors than others. Characterization of these species would indicate those that should be preferentially seeded into these grasslands in an attempt to resist further cheatgrass invasion and promote the proliferation of native species.                     26 Chapter 3: Community-wide differences in cheatgrass-invaded communities and species-specific co-occurrence relationships  Introduction  Species that have been introduced to novel habitats provide some of the best examples of those that exert the greatest negative influences on communities in terms of their structure and function (Vitousek et al. 1996, Woods & Moriarty 2001). Non-native plant species, which can be introduced purposefully and non-purposefully, are now considered to be a major component of environmental change and biodiversity loss (Theoharides & Dukes 2007). The negative impacts attributed to introduced plant species can be so significant that major economic expenditure is required to prevent further spread once they are established (Pimentel et al. 2005); invasive weed control costs and associated crop losses in the United States alone are estimated to be approximately $35 billion per year. One introduced species whose spread throughout western North America has had major negative ecosystem-level consequences is cheatgrass, Bromus tectorum (Mack 1981). Unintentionally introduced from Eurasia over 150 years ago, it has quickly dominated formerly diverse grasslands and arid sagebrush communities (Reid et al. 2008). It can form monotypic stands and even in less severe circumstances, it generally reduces diversity and has negative impacts on ecosystem processes. It is also implicated in increasing the periodicity of fire regimes far above historical frequencies, essentially creating a positive feedback that translates into even greater domination of landscapes (Haubensak et al. 2009). Restoration of cheatgrass-infested grasslands is very challenging because once it is established the plant grows rapidly, produces large quantities of seed, and competes effectively for available soil moisture and nutrients (Monaco et al. 2003).  The common management options include herbicide application (Pellant et al. 1999, Baker et al. 2009) and extensive seeding of desired species into affected areas (Epanchin-Niel et al. 2009). The latter option is the most appealing but there is often no consensus on which species should be seeded as species-specific associations and relationships are often spatially and temporally dependent. As a result, restoration attempts featuring native grasses and forbs  27 are often unsuccessful (Clements & Young 2007). Nevertheless, the isolation of context- dependent relationships between cheatgrass and other species should highlight the species that are likely to give restoration attempts the best chance of success in discrete environments. This hypothesis fits broadly into a traditional view of community assembly and the forces dictating it; the focus being on species possessing different traits that are believed to predetermine in some manner their ability to occupy specific niches within a heterogeneous environment (Hutchinson 1957, Tilman 1988). However, despite some species appearing to exert far greater effects and assemble more readily into certain communities than others, such as cheatgrass in North American grasslands, this view has recently been challenged. The development of neutral-based models based on the assumption that species of a similar trophic status are ‘ecologically equivalent’ have called into question whether community assembly is dictated by generalized rules (Hubbell 2001, 2005).  Both neutral and niche-based theories have their detractors; many studies support the notion that species sharing similar traits and environmental requirements compete more strongly with one another (Fargione et al. 2003, Turnbull et al. 2005). Niche-based models, however, typically require unworkable levels of complexity, and, in many cases, multiple plausible hypotheses exist to explain observed phenomena (Chave et al. 2002). In testing whether community composition can be explained by neutral or non-random assembly rules, ecologists often consider indices such as community C-scores (Stone and Roberts 1990), number of checkerboard pairs (Diamond 1975) and number of species combinations (Pielou & Pielou 1968). Community C-scores quantify the average amount of co-occurrence in a given community; they highlight to what extent the total number of species present at a larger spatial habitat scale co-occur with one another in smaller discrete communities of the habitat. Checkerboard pairs analysis quantifies the specific number of species pairs that never co-occur, and species combinations analysis quantifies the specific number of unique species pairs that do co-occur. To test for significance, these indices all require community presence-absence matrices to be compared to those of simulated, null communities (Gotelli & McCabe 2002).  28  Grasslands of the Okanagan Valley in British Columbia, Canada, provide a good opportunity to test whether discrete plant communities are assembled differently based on their invasion status. Because cheatgrass abundance and distribution recently increased in monitored pastures in this region (Stephens et al. 2009), I asked whether communities invaded by cheatgrass featured different species associations, and whether these were indicative of altered assembly patterns. I hypothesized that ‘invaded’ areas would have lower grass species diversity and that co-occurrence analyses would show relatively high levels of community disassembly in ‘invaded’ areas. If communities are structured non-randomly (via competitive processes e.g.), community C-score, checkerboard and species combinations indices should differ significantly from those produced by null matrices whereas indices suggestive of community ‘disassembly’ should not differ from null matrix predictions (Sanders et al. 2003, Reshi et al. 2008). My prediction was based on the belief that a dominant invader such as cheatgrass would suppress and extirpate other species on a random basis wherever they were found; reduced biodiversity mediated by invasion can result in the breakdown of assembly rules that structure intact communities (Sanders et al. 2003). In addition to grass species diversity and community co-occurrence analyses, I conducted species-specific co-occurrence analyses between cheatgrass and other grass species to isolate the species found in association more or less frequently than was average for the whole community. This should indicate potential competitors with cheatgrass and thus has implications for the restoration of grasslands that have traditionally been seeded with other rapid growing introduced grasses (Mazzola et al. 2008).          29 Methods   Cover data were collected in 0.1m2 (0.2m x 0.5m) plots that fell along 10m transects that were established in five pastures in the White Lake area of the Okanagan Valley in British Columbia, Canada (Stephens et al. 2009). In this study, species-specific presence-absence data were converted from cover data from the Park Rill pasture (49°18'14.22725"N, 119°38'42.11677"W to 49°18'31.82939"N, 119°38'55.78915"W). These data were then used to assess grass species diversity and to conduct extensive co- occurrence analyses. This pasture was chosen because data from it were numerous; it was the only pasture chosen for this analysis because combining others would increase the likelihood that environmental heterogeneity would influence co-occurrence and diversity patterns more strongly. Analyses were conducted independently for plots that were ‘invaded’ by cheatgrass and those that were ‘intact’ (lacked cheatgrass). Sampling of the species present in plots took place in 2001, 2002, 2003 and 2005. Data used in analyses were taken from all 90 of these plots; the same plots were used in analyses for each independent year.  Grass diversity analyses   Grass diversity was assessed by calculating the mean number of grass species present in plots. This analysis was conducted for each year and plot status (‘intact’ and ‘invaded’) to determine whether cheatgrass had a significant effect on grass diversity and whether this changed over the time period of the study. Twenty-eight grass species were present at least once in plots from which data were taken for use in diversity and co- occurrence analyses, of which 19 were native and 9 introduced (Table 3.1).        30 Table 3.1: List of the 28 grass species (Family Poaceae) in co-occurrence and diversity analyses, featuring a total of 19 native and 9 introduced, and 10 annual and 18 perennial species. The four species appearing in bold font were those for which spatial distribution patterns were analyzed in considering potential environmental gradients.  Grass species  Status (Native/Introduced) Growth form Agropyron cristatum Introduced Perennial Apera interrupta Introduced Annual Bromus hordeaceus Introduced Annual Bromus tectorum Introduced Annual Bromus commutatus Introduced Annual Bromus racemosus Introduced Annual Distichlis spicata Native Perennial Distichlis stricta Native Perennial Distichlis spp. Native Perennial Elymus repens Introduced Annual Elymus trachycaulus Native Perennial Elymus spp. Native Perennial Festuca idahoensis Native Perennial Hordeum jubatum Native Perennial Koeleria macrantha Native Perennial Leymus cinereus Native Perennial Muhlenbergia asperifolia Native Perennial Muhlenbergia richardsonis Native Perennial Poa bulbosa Introduced Perennial Poa pratensis Native Perennial Poa secunda Native Perennial Poa cuscickii Native Perennial Poa spp. Introduced Perennial Pseudoroegneria spicata Native Perennial Sporobolus cryptandrus Native Perennial Stipa comata Native Perennial Stipa occidentalis Native Annual Vulpia octoflora Native Annual   Co-occurrence analyses    All co-occurrence analyses were conducted using the statistical software package Ecosim, version 7.0 (Gotelli & Entsminger 2006). For each year in the analysis, a presence-absence matrix of observed data was constructed for all plot communities that had been sorted into ‘intact’ and ‘invaded’ categories. Observed data matrices were compared statistically to null matrices created via a sequential swap (SS) algorithm and a Random Knight’s Tour (RKT) algorithm. I used both algorithms because there is disagreement over which should be used under different circumstances (Gotelli & Entsminger 2001, Manly & Sanderson 2002, and references therein).  31 To assess community co-occurrence patterns for each year of the study, I calculated two indices; Stone & Roberts’ (1990) checkerboard score of the matrix (C- score), and the number of unique species combinations. The community C-score first quantifies the total number of checkerboard units that can be found for each species pair in the matrix. A checkerboard unit is defined as one in which ‘species x’ is present and ‘species y’ is not, or ‘species y’ is present and ‘species x’ is not. Thus, for each species pair, the number of checkerboard units is (Ax - C)(Ay - C), where Ax is the number of occurrences for species x, Ay is the number of occurrences for species y, and C is the number of community plots in which both species occur. The final community C-score index is calculated as the average number of checkerboard units for each unique species pair. Communities that are structured non-randomly (by abiotic and/or biotic processes) are expected to have larger C-scores and fewer pair-wise species combinations than predicted by null matrices due to species segregation; randomly structured communities should have larger or similar C-scores and more pair-wise species combinations than predicted by null matrices due to uninhibited species aggregation. The number of species combinations index is simply calculated as the number of unique species combinations found in the matrix. Using both the SS and RKT algorithms, I created 30,000 simulated null matrices for comparison against observed matrices. In the simulated null matrices, row (species) sums were fixed such that each species appeared with the same frequency as in observed matrices, while all columns (plots) were fixed in the same manner. To make results directly comparable, standardized effect sizes were calculated directly from Ecosim 7.0, as: (observed index - mean of simulated null indices) / standard deviation of simulated null indices. Given a normal distribution, significant effect sizes (either positive or negative) will fall at least 2.0 units above or below 0 on the effect scale when testing for significance at p=0.05.  Species-specific cheatgrass associations  Co-occurrence comparisons were made between cheatgrass and the other grass species to isolate the species whose abundances seemed most affected by plot status. For the years of 2001 and 2005, the percentage of plots occupied by each species in both  32 ‘intact’ and ‘invaded’ plots was calculated. These percentages were then compared to assess whether the species in question occurred more frequently in the absence of cheatgrass (‘intact’), than in association (‘invaded’). For example, if species ‘x’ occurred in 10% of ‘intact’ plots in year ‘y’, but only 5% of ‘invaded’ plots in the same year, it was observed to occur twice as frequently in ‘intact’ plots (10/5=+2).  Consideration of environmental gradients   Although no environmental data were measured for plots, annual and early-year (January – April) rainfall values were available from a long-term weather station in Penticton, approximately 40km north of the Park Rill pasture. I also conducted a spatial analysis of the distribution of four of the most common species in these plots (bold species in Table 1). Linear models were used to assess whether distributions were correlated to distance from the start point of transects.                   33 Results Grass diversity  Within-year comparisons of ‘intact’ and ‘invaded’ single plots showed a significant effect of invasion status on grass species diversity for all years except 2005 (Figure 3.1). In each of these cases, ‘invaded’ plots contained significantly fewer grass species than ‘intact’ plots. In 2001, p=0.017 (Wilcoxon rank sum test, W=1378); in 2002 p=<0.0001, (W=1472); in 2003 p=<0.0001, (W=1524). In 2005, there was no significant effect (p=0.105, W=1262.5). Although there was no significant difference in the number of grass species found in ‘intact’ and ‘invaded’ plots in 2005, the trend remained for diversity to be lower in ‘invaded’ plots.  Figure 3.1: Within-year comparisons of mean grass species richness (±  SE) between ‘intact’ (black bars) and ‘invaded’ (white bars) plots. Significant differences at p=<0.017 existed for comparisons in 2001, 2002 and 2003. N=44 plots for 2001 ‘intact’, 46 (‘01 ‘invaded’), 51 (’02 ‘intact’), 39 (’02 ‘invaded’), 44 (’03 ‘intact’), 46 (’03 ‘invaded’), 39 (’05 ‘intact’), 51 (’05 ‘invaded’). No data were collected in 2004.     34 Co-occurrence index comparisons  C-score  C-score index analysis of plots between 2001 and 2005 showed different patterns for ‘intact’ and ‘invaded’ plots (Figure 3.2). ‘Intact’ plots had significantly higher C- scores than predicted by null matrices under the SS algorithm for all years at p=!0.004. However, those with ‘invaded’ status had significantly higher C-scores than predicted by null matrices for 2001 only (SES = 3.95, p=<0.001); C-scores did not differ significantly from those predicted by null matrices for the other three years (p="0.07). Absolute index values in C-score analyses under the RKT algorithm differed quantitatively from those under the SS algorithm, but the underlying patterns of statistical significance remained the same for all comparisons (Table 3.2).   Figure 3.2: Standardized effect sizes generated from C-score analysis, comparing 30,000 simulated null matrices to the observed matrix for each year and status. ‘Intact’ plots (black bars) for all year classes had significantly higher C-scores than expected by chance at p=!0.004. ‘Invaded’ plots (white bars) had significantly higher C-scores in 2001, but not so in 2002, 2003 and 2005. Standardized effect sizes between 2.0 and -2.0 indicate non-significant deviations from predictions at p=!0.05.   35 Table 3.2: Statistical comparisons of C-score analyses using the SS and RKT algorithms. All statistically significant, directional results were repeated using both algorithms, although absolute values differ. ‘Intact’ plots are in bold font within years; ‘invaded’ plots are in regular font.   Year  Plot status SES and p-value for C- score analysis under SS algorithm SES and p-value for C- score analysis under RKT algorithm ‘Intact’ SES = 3.15, p=0.003 SES = 5.01, p=<0.001 2001  ‘Invaded’ SES = 3.95, p=<0.001 SES = 4.15, p=<0.001 ‘Intact’ SES = 2.95, p=<0.001 SES = 3.61, p=<0.001 2002  ‘Invaded’ SES = -0.59, p=0.62 SES = -0.34, p=0.69  ‘Intact’ SES = 3.16, p=0.004 SES = 3.96, p=<0.001 2003  ‘Invaded’ SES = 1.54, p=0.07 SES = 1.71, p=0.06  ‘Intact’ SES = 4.33, p=<0.001 SES = 7.65, p=<0.001 2005  ‘Invaded’ SES = 1.29, p=0.11 SES = 1.61, p=0.08   Number of species combinations   Analysis of the number of species combinations showed different patterns for ‘intact’ and ‘invaded’ plots between 2001 and 2005 (Table 3.3). ‘Intact’ plots had significantly fewer unique species combinations than predicted by null matrices using the SS algorithm for all years (p=!0.004). However, ‘invaded’ plots only had significantly fewer combinations than predicted by null matrices for 2001 (SES = -2.64, p=0.01); the number of unique combinations did not differ significantly from those predicted by null matrices for the other three years (p="0.06). Absolute index values in comparisons made using the RKT algorithm differed quantitatively from those using the SS algorithm, but underlying patterns of statistical significance remained the same (Table 3.3).  Table 3.3: Statistical comparisons of species combinations analysis using SS and RKT algorithms. All statistically significant, directional results were found under both algorithms, although absolute values differ. ‘Intact’ plots are in bold font within years; ‘invaded’ plots are in regular font.   Year  Plot status SES and p-value for C- score analysis under SS algorithm SES and p-value for C- score analysis under RKT algorithm  ‘Intact’ SES = -3.13, p=0.003 SES = -3.30, p=0.002 2001  ‘Invaded’ SES = -2.64, p=0.01 SES = -2.69, p=0.009  ‘Intact’ SES = -3.01, p=0.005 SES = -2.93, p=0.004 2002  ‘Invaded’ SES = -0.94, p=0.91 SES = -0.96, p=0.89  ‘Intact’ SES = -3.67, p=<0.001 SES = -3.61, p=<0.001 2003  ‘Invaded’ SES = 1.95, p=0.06 SES = 1.82, p=0.21         ‘Intact’ SES = -3.07, p=0.004 SES = -3.93, p=<0.001 2005  ‘Invaded’ SES = -1.93, p=0.16 SES = -1.21, p=0.68   36 Species-specific cheatgrass associations   Six grass species showed co-occurrence patterns of interest with cheatgrass (Table 3.4). Of particular interest was A. cristatum; this species had a positive co- occurrence relationship with cheatgrass in 2001 (it was found 2.2 times relatively more often in ‘invaded’ than in ‘intact’ plots), but this had become strongly negative by 2005 (it was found 1.8 times relatively more often in ‘intact’ plots). Three species (P. spicata, S. cryptandrus and S. comata) had substantially positive co-occurrence relationships with cheatgrass throughout the study, although they were less positively associated with cheatgrass in 2005 than they were in 2001. Stipa occidentalis was the only species with a positive co-occurrence relationship in both years to increase that relative positive association by 2005 (+1.2 in 2001 to +2.1 in 2005). However, also of interest was P. pratensis; this species was negatively associated with cheatgrass in 2001 (it appeared 1.4 times relatively less often in ‘invaded’ plots) but was positively associated in 2005 (Table 3.4). The 5 species that had positive co-occurrences in 2005 were all native, and 4 of these (P. pratensis, P. spicata, S. cryptandus, S. comata) were perennials.  Table 3.4: The 6 species that displayed patterns of interest in co-occurrence relationships with cheatgrass. These species were either found substantially more or less frequently in association with cheatgrass (‘invaded’ plots) than in ‘intact’ plots in either 2001 or 2005, or displayed negative patterns in 2001 and positive patterns in 2005. Bold font is used for species found less frequently in ‘invaded’ plots; regular font is for species found more frequently. Values represent the number of times more or less frequently the species was found in ‘invaded’ plots. Letters in brackets after species names show whether the species is native or introduced (N or I), annual or perennial (A or P).  Grass species in pair-wise C-score comparison  (status) 2001  2005 Poa pratensis  (N,P) -1.4 +1.3 Pseudoroegneria spicata (N,P) +4.2 +2 Sporobolus cryptandus  (N,P) +2.9 +2 Stipa comata  (N,P) +3.1 +1.3 Agropyron cristatum  (I,P) +2.2 -1.8 Stipa occidentalis  (N,A) +1.2 +2.1       37 Consideration of environmental gradients Precipitation   Annual and early-year precipitation in Penticton varied among years of the study (Figure 3.3). Annual precipitation ranged from 426.7mm in 2004 to 197.3mm in 2002, while early-year precipitation ranged from 100.5mm in 2003 to 50.1mm in 2002.  Figure 3.3: Annual and early-year (January – April) precipitation in Penticton (black bars are annual data, patterned bars are early-year data).  Spatial distributions of common grass species  Spatial distribution patterns of three of the four common species (Bromus tectorum, Pseudoroegnaria spicata, Stipa comata) showed no significant correlations with distance from transect start points (p=!0.11). Sporobolus cryptandrus was significantly correlated with distance in 2005 (R2=0.09, p=0.04), but not so in the other years (p=!0.14).        38 Discussion  Grass diversity   In this study I have shown that invasion status had a significant effect on grass species diversity between 2001 and 2003; plots ‘invaded’ by cheatgrass had significantly fewer grass species than those that were ‘intact’. However, the pattern was not detected in 2005. This is surprising given that the abundance of cheatgrass had increased noticeably by 2005 when compared to previous years in the Park Rill pasture (Stephens et al. 2009). Because cheatgrass is reported to suppress and extirpate native species within communities (Blank 2008, Belnap & Sherrod 2009), it was expected that grass diversity would have been reduced at least as much as was seen in earlier years.  The major concerns with introduced plant species decreasing diversity include the possibility of reduced productivity of ecosystems as well as negative impacts on a host of other key processes at both the community and ecosystem level (Gordon 1998).  Davies and Svejcar (2008) focused on the introduced annual grass medusahead, Taeniatherum caput-medusae in Oregon; species richness was found to be significantly higher in non- invaded communities, and this translated into significantly greater biomass production of native perennial species, which are used for habitat and nutrition by native wildlife. Extrapolating invasion patterns from one habitat to another is difficult due to differences in community processes that can in turn impact the time scale on which invasion and its subsequent effects operate (Theoharides & Dukes 2007). Therefore, this result should not be taken as evidence that cheatgrass will fail to suppress other species in the Park Rill pasture in the future. Also, diversity patterns evident at one spatial scale cannot always be extrapolated to larger scales of the same landscape (Crawley & Harral 2001, Willis & Whittaker 2002, Sandel & Smith 2009).  Co-occurrence index comparisons   Co-occurrence analyses provided some support to suggest that cheatgrass might have been exerting an effect on community assembly processes. ‘Intact’ community C- scores were higher than predicted by simulated null matrices, which is expected of  39 communities that are non-randomly structured by biotic and/or environmental processes (Stone & Roberts 1990, Sanders et al. 2003, Reshi et al. 2008). In contrast, ‘invaded’ community C-scores did not differ from null predictions in 2002, 2003 and 2005. Because the 2001 C-score was significantly higher than predicted, these results hint that plots with cheatgrass were showing greater disassembly in the latter years of the study. Co-occurrence analyses of the number of species combinations in observed community matrices yielded relatively similar results as the C-score analyses; plots ‘invaded’ by cheatgrass appeared to be structured randomly, whereas ‘intact’ communities showed patterns suggestive of non-random species assemblages. This is because the observed number of species combinations in ‘intact’ communities was lower than those predicted by null matrices for all years, whereas ‘invaded’ communities showed no significant deviations from null model predictions in 2002, 2003 and 2005. In structured communities certain species are more likely to be found growing together than others. It is thus expected that some species will rarely or never be found growing together if competition dictates relationships and/or environmental requirements ensure realized niches do not overlap. Independent studies considering co-occurrence patterns have assessed a wide variety of plant and animal communities; these have ranged from plants on British Columbia islands (Burns 2007) to ant assemblages in California (Sanders et al. 2003). These studies have often produced conflicting results regarding the probability of assembly patterns following generalized rules, which are typically indicated by highly variable patterns of co-occurrence between different species. However, a comprehensive meta-analysis reported that communities, in general, appear to be associated non-randomly (Gotelli & McCabe 2002). This analysis used 96 studies and portioned these into 8 distinct taxonomic groups. Six of these groups demonstrated non-random co-occurrence patterns. A total of 13 plant-based studies were included and these showed the second strongest effect size of the 8 taxonomic groups. All co-occurrence analyses reached the same conclusions in terms of statistical significance regardless of whether the SS or RKT algorithm was used. Both algorithms have their supporters and detractors (Gotelli & Entsminger 2001, Manly & Sanderson 2002, and references therein), so the fact that results from this study were congruent with both suggests the overall conclusions are robust. Having said this, because the study was  40 merely observational, it is also conceivable that the observed patterns of community assembly were influencing the distribution of cheatgrass, rather than the other way around. It is possible that cheatgrass can only invade communities that are randomly structured; these communities would likely be more easily invaded due to reduced competition. However, this seems less probable as cheatgrass is capable of invading virtually any arid habitat throughout western North America, some of which are very biologically diverse (Mack 1981, Ortega & Pearson 2005).  Cheatgrass-specific co-occurrence analyses and environmental gradient considerations  Species that were found relatively less frequently in ‘invaded’ than ‘intact’ plots (those that deviate by a substantial negative ratio) are speculated to either suffer in competition with cheatgrass or require different environmental conditions; substantial positive ratios should therefore indicate the species that are either likely to compete successfully and/or require similar environmental conditions. This is because species that are found in ‘invaded’ plots relatively more often are likely to be growing with cheatgrass because they are the species that are able to co-occur (and compete) with this introduced plant without being extirpated, and/or local environmental conditions are suitable for both. Attributing these positive and negative relationships to biotic or environmental processes is difficult without comprehensive data on the relative homogeneity of environmental parameters within these transects. However, these species are very likely the ones that need to be tested in competition-based experiments designed to characterize the exact nature of biotic relationships and thus identify strong competitors in this habitat. Annual and spring rainfall varied among the years of this study, but there was no period of intense drought. Although cheatgrass abundance increased in 2005 following high rainfall the previous year, evidence from diversity analyses does not suggest it resulted in suppression of other species per se as grass diversity actually increased in ‘invaded’ plots compared with previous years. The possibility that small-scale environmental heterogeneity was responsible for significantly influencing co-occurrence patterns cannot be ruled out, but species distribution analyses did not suggest this was the case; three of the four common species showed no significant trend in their distribution  41 over transects while the other (Stipa comata) only showed a weak correlation for one year. Although far from strong evidence, distribution of the common species would have been expected to show greater spatial variation if transects had varied significantly in their environmental characteristics. To speculate, it seems that species with a positive association with cheatgrass are likely to be good competitors and thus the ones on which to focus to prevent invasion. Those that had a negative association, however, are likely to be weaker competitors – particularly those whose relationships became progressively negative during the study as cheatgrass abundance increased. For example, in 2001, the introduced crested wheatgrass (Agropyron cristatum) was positively associated with cheatgrass (it appeared relatively more frequently in ‘invaded’ than ‘intact’ plots), but in 2005 it occurred relatively less frequently in association with cheatgrass than in ‘intact’ plots. Crested wheatgrass is often described as having superior competitive ability to many other grassland species (Young & McKenzie 1972, Fansler et al. 2007), which is why it has been frequently used in restoration attempts in areas invaded by cheatgrass. However, relatively few studies provide compelling evidence that the species is able to effectively compete with cheatgrass. For example, Ott et al. (2003) found that by heavily seeding crested wheatgrass into areas in sagebrush and pinon-juniper landscapes in Utah, it became the dominant species following fire. However, cheatgrass was still dominant in close proximity and it too increased during the three-year study. Francis & Pyke (1996) established greenhouse competition experiments with crested wheatgrass, another grass of the same genus (Agropyron desertorum), and cheatgrass. Although crested wheatgrass was the better of the two Agropyron species in competing with cheatgrass, it had lower biomass and tiller production when cheatgrass density was increased, and of the three, cheatgrass was always the best competitor regardless of mixture. So, in conjunction with the results from these co-occurrence analyses, there is a suggestion that crested wheatgrass might have been overestimated in its ability to suppress cheatgrass. Western needlegrass (Stipa occidentalis) was the only species whose association patterns with cheatgrass were positive in both years but more positive in 2005. Callaway et al. (2005) found that this species was especially resistant to the competitive effects of Centaurea maculosa, spotted knapweed, which is another introduced plant that has  42 spread rapidly throughout North America. This was the only one of five native species whose growth increased significantly when exposed to the presence of knapweed root exudates and activated carbon following prior exposure to the species. Callaway et al. (2005) interpreted this as a possible evolution of tolerance. If this hypothesis is correct, western needlegrass might also become more resistant to the competitive effects of cheatgrass when the two species grow together. Kentucky bluegrass (Poa pratensis) was negatively associated with cheatgrass in 2001 (it was found relatively less frequently in ‘invaded’ plots), but became positively associated in 2005. This is a potentially important result; Kentucky bluegrass co-occurred with cheatgrass relatively more frequently in these environments when the abundance of cheatgrass increased. As such, if it is a competitor, there is some hope that it too will aid the success of restoration programs designed to prevent further proliferation of cheatgrass. This seems plausible because Kentucky bluegrass has been shown experimentally to restrict the development of cheatgrass roots; Bookman and Mack (1982) found that when grown together, cheatgrass roots fail to develop laterally due to complex interactions with those of Kentucky bluegrass. Less specifically, Kentucky bluegrass has also been implicated in dominating grassland communities that feature other introduced species (Miles and Knops 2009), which suggests it is a relatively competitive species once established. Three species occurred in association with cheatgrass substantially more often than in ‘intact’ communities across all years. These were bluebunch wheatgrass (Pseudoroegnaria spicata), sand dropseed (Sporobolus cryptandrus) and needle-and- thread grass (Stipa comata). Needle-and-thread grass, from the same genus as the positively associated western needlegrass, has already been shown to compete relatively well with cheatgrass. Melgoza & Nowak (1991) found that root production of this species did not differ significantly whether it was in monoculture or in the presence of cheatgrass for two years following a fire in Nevada. The authors also noted that production did not differ between those plots and plots that had been burned 12 years previously, indicating that cheatgrass had not begun to out-compete needle-and-thread grass in that time. Given that cheatgrass is known to spread rapidly following fire (Rowe & Brown 2008), this result is of particular importance from a restoration perspective.  43 Sand dropseed has been noted as a species that is relatively tolerant to extended disturbances, such as grazing (Fuhlendorf & Smeins 1997). Thus, it is a good candidate to be successful in the harsh environment of the Okanagan Valley, where grazing, fire and introduced species impact other species more negatively. This also makes it of interest as a species with restoration promise. Of particular interest, however, is bluebunch wheatgrass. For some time it has been preferentially selected as cattle forage (Rickard et al. 1975), and because it is also drought tolerant and was found in diverse grasslands prior to the spread of cheatgrass and other introduced species, it has been aggressively seeded in restoration programs (Larson et al. 2000). It is considered so valuable that recent work has focused on the isolation and selective breeding of weed-resistant strains for this purpose (Fu & Thompson 2006). If cheatgrass is a competitor there is thus great hope that bluebunch wheatgrass might prove effective in resisting cheatgrass invasion and spread. Bluebunch wheatgrass has several characteristics that make it a good competitor; it grows in tall, dense bunches that exclude some smaller species, and it germinates rapidly in cool experimental regimes. Despite this optimism, most studies suggest bluebunch wheatgrass is at a competitive disadvantage with cheatgrass. For example, Aguirre & Johnson (1991) showed that seedling growth was reduced when bluebunch wheatgrass was grown in the presence of cheatgrass, and that cheatgrass could grow twice as many roots in the first 45 days of growth. However, bluebunch wheatgrass could be effective at resisting invasion when it is mature – it might just be that it does not compete particularly well when young. As a potential management tool, the promotion of native species such as those found in positive associations with cheatgrass in this study is likely to be preferable to using other rapidly growing introduced bunchgrasses, that have traditionally been used in attempts to restore invaded plant communities (Mazzola et al. 2008).       44 Conclusions   In this study I have shown that grass species diversity in an Okanagan Valley pasture was affected by the presence of cheatgrass; ‘invaded’ plots had significantly lower diversity between 2001 and 2003. Cheatgrass had increased its abundance by 2005 (Stephens et al. 2009) and community-wide co-occurrence analyses showed results that were interpreted in a way that ‘invaded’ communities displayed patterns characteristic of disassembly from 2002 onwards. This was because community co-occurrence appeared random, whereas ‘intact’ communities displayed non-random patterns of assembly. Although the possibility of cheatgrass distribution being affected by community structure rather than the other way around cannot be ruled out, it seems less likely as one of the world’s most invasive plant species would still have been expected to invade competitively structured communities – even if it was less abundant in them. Species- specific co-occurrence patterns involving cheatgrass and other grass species showed that some species had positive associations (appeared relatively more often in ‘invaded’ than in ‘intact’ plots) and others had negative associations. Species distribution analyses were not conducted at a sufficiently detailed level to rule out the possibility that environmental heterogeneity was responsible for the observed co-occurrence patterns, but they at least provided some support that this was not the case. Thus, it is speculated that the species that occurred relatively more frequently with cheatgrass than in ‘intact’ plots are likely to be the most effective competitors. Kentucky bluegrass, bluebunch wheatgrass, western needlegrass, needle-and-thread grass, and sand dropseed could therefore be of great value in restoration programs.          45 Chapter 4: General conclusions  Summary  Cheatgrass, Bromus tectorum, has invaded millions of hectares of arid grassland throughout western North America since its introduction over 150 years ago (Mack 1981). It is a major threat to ecosystem diversity and function as it can quickly dominate habitats and extirpate native species (Bradley & Mustard 2006), and is not favoured by ranchers whose cattle obtain less than adequate nutrition and can be injured by eating the plant (Morrow & Stahlman 1984). Once established it is also hard to remove, making restoration of environments incredibly difficult. This is because cheatgrass possesses a number of traits that make it a successful invader and strong competitor; it produces high quantities of seeds (Monaco et al. 2003), recovers particularly well after fire (Rowe & Brown 2008), and can germinate under a wide range of conditions (Meyer et al. 1997). Nevertheless, studies have characterized the potential success of management plans involving the use of herbicides (Pellant et al. 1999), as well as seeding other species that compete effectively with cheatgrass in specific habitats (Ott et al. 2003). However, these successes appear to be context dependent; discrete habitats feature unique networks of biotic and abiotic associations and good competitors in some habitats do not always compete effectively in others. As a result, knowledge of the environmental and biotic factors influencing the dynamics of cheatgrass invasion and establishment on a site-by-site basis is extremely valuable from a management perspective. Because cheatgrass recently increased in abundance and distribution in pastures of the Okanagan Valley (Stephens et al. 2009), it is imperative that studies are designed to assess how restoration programs should be implemented most effectively in these regions. In this study, I therefore conducted analyses of observational data that attempted to isolate some of the most important environmental and biotic factors influencing the abundance and distribution of cheatgrass in Okanagan grasslands. In Chapter 2, I first asked whether cheatgrass abundance was correlated with (1) proximity to pine trees, (2) diversity of other plant species, and (3) aspect from pine trees. Firstly, I found a highly positive association between cheatgrass abundance and proximity to pine trees; abundance decreased significantly as distance from pine trees  46 increased. Secondly, a highly negative association was found for cheatgrass abundance and the diversity of other plant species; abundance decreased significantly as the diversity of other plant species increased. Thirdly, no significant associations were found for cheatgrass abundance and aspect. Following the discovery that cheatgrass abundance was positively associated with proximity to pine trees, I asked whether (1) soil pH and (2) soil moisture also varied with proximity to these trees. I found highly significant associations for both. Firstly, soil pH increased significantly as distance from pine trees increased. Secondly, soil moisture also increased significantly as distance from pine trees increased (although this association was only found at 4 of 5 study sites). Following the discovery that cheatgrass abundance was negatively associated with the diversity of other plant species, I asked whether abundance was associated with the proportion of (1) native bunchgrasses and (2) other introduced plant species comprising the other plant species. I found significant associations for both, but these were relatively weak and were not found at all study sites. Where found, cheatgrass abundance increased significantly as the proportion of native bunchgrasses increased, while cheatgrass abundance decreased significantly as the proportion of other introduced species increased. In Chapter 3, to further enhance the knowledge of species-specific relationships involving cheatgrass in this region, I conducted grass species diversity and extensive co- occurrence analyses in a pasture previously monitored by Stephens et al. (2009). This attempted to discern whether plots ‘invaded’ by cheatgrass differed in terms of the richness and identity of species that were present from those that were ‘intact’ (lacked cheatgrass), and whether this was suggestive of different patterns of community assembly. In doing this, I used the presence-absence modelling software Ecosim 7.0 (Gotelli & Entsmiger 2006) to compare observed community matrices with 30,000 simulated null matrices. I then asked which species were found in association with cheatgrass relatively more or less frequently by comparing the percentage of ‘intact’ and ‘invaded’ plots that featured each species in 2001 and 2005. The key results were that (1) grass species diversity was lower in ‘invaded’ plots. (2) Community-wide co-occurrence indices such as Stone and Roberts’ (1990) C-score  47 and the number of species associations (Pielou & Pielou 1968) showed different community assembly patterns for ‘intact’ and ‘invaded’ plots; these suggested that ‘intact’ communities featured non-random segregation of species whereas ‘invaded’ communities appeared to be structured randomly as predicted by theory (Reshi et al. 2008). Species-specific co-occurrence analyses showed that Kentucky bluegrass (Poa pratensis) bluebunch wheatgrass (Pseudoroegnaria spicata), western needlegrass (Stipa occidentalis), sand dropseed (Sporobolus cryptandrus) and needle-and-thread grass (Stipa comata) were all found in association with cheatgrass relatively more often than in plots lacking cheatgrass. This suggests these are the species whose biotic relationships with cheatgrass are likely to be most influential in determining its subsequent invasion in Okanagan grassland ecosystems.                      48 Suggestions for future research   The key results of these analyses lead to a number of further research opportunities, which, if pursued, will further understanding of how cheatgrass dynamics are affected by the environmental and biotic components of Okanagan grasslands.  1: Can selective herbicide application in close proximity to pine trees reduce cheatgrass abundance?  Having shown a positive correlation between cheatgrass abundance and proximity to pine trees, it is important to assess whether selective herbicide application near the base of pine trees will have a population level effect on cheatgrass abundance. Selective application will reduce associated economic costs while reducing the likelihood of non- target species being negatively impacted.  2: Can cheatgrass abundance be reduced by selective applications of compounds that manipulate soil pH and available soil moisture?  Because soil pH and soil moisture were shown to decrease with proximity to pine trees, it is important to assess whether these environmental differences explain any of the variation in cheatgrass abundance.  Selective application of compounds that increase soil pH and soil moisture could therefore potentially decrease cheatgrass abundance, providing another management tool for those charged with restoring invaded habitats.  3: Can cheatgrass abundance be reduced by seeding species that co-occur with it more frequently than expected in situ?   In 2005, when cheatgrass abundance had increased, five grass species were found to co-occur relatively more frequently with cheatgrass than in ‘intact’ plots. These species were Poa pratensis, Pseudoroegnaria spicata, Stipa occidentalis, Sporobolus cryptandrus and Stipa comata. Because positive co-occurrence can indicate facilitation as well as competition, it is imperative to ascertain the precise nature of these relationships. By seeding these species with cheatgrass and comparing abundance and growth rate  49 indices with monoculture controls, it will be possible to characterize the precise nature of these species-specific relationships, and ask whether these species can reduce cheatgrass abundance.                              50 Management implications   Analyses conducted in this research used observational data and conclusions were therefore based largely on correlations. However, highly significant associations between cheatgrass abundance and environmental and biotic factors are indicative that these factors are at least partially responsible for the observed abundance patterns in Okanagan grasslands. If manipulated experiments confirm these results, the selective use of herbicide and artificial seeding of species that will compete with cheatgrass most effectively are likely to be adopted as first choice, economically viable options for those charged with restoring these critically important habitats.                       51 References  Aguirre, L. & Johnson, D.A. (1991). Influence of temperature and cheatgrass competition on seedling development of two bunchgrasses. Journal of Range Management 44: 347-354.  Aragon, C.F., Albert, M.J., Gimenez-Benavides, L., Luzuriaga, A.L. & Escudero, A. (2007). Environmental scales on the reproduction of a gypsophyte: a hierarchical approach. Annals of Botany 99: 519-527.  Baker, W.L., Garner, J. & Lyon, P. (2009). Effect of Imazapic on cheatgrass and native plants in Wyoming big sagebrush restoration for Gunnison Sage-grouse. Natural Areas Journal 29: 204-209.  Belnap, J. & Sherrod, S.K. (2009). Soil amendment effects on the exotic annual grass Bromus tectorum L. and facilitation of its growth by the native perennial grass Hilaria jamesii (Torr.) Benth. Plant Ecology 201: 709-721.  Bezener, A., Dunn, M., Richardson, H., Dyer, O., Hawes, R. & Hayes, T. (2004). South Okanagan-Similkameen conservation program: a multi-partnered, multi-species, multi-scale approach to conservation of species at risk. Accessed on 29/02/2010 at: http://www.llbc.leg.bc.ca/public/pubdocs/bcdocs/400484/bezener_edited_final_feb_8.pdf  Blank, R. (2008). Biogeochemistry of plant invasion: a case study with Bromus tectorum L. Journal of Invasive Plant Science and Management 1: 226-238.  Bookman, P.A & Mack, R.N. (1982). Root interactions between Bromus tectorum and Poa pratensis: a three-dimensional analysis. Ecology 63: 640-646.  Booth, M.S., Caldwell, M.M. & Stark, J.M. (2003). Overlapping resource use in three Great Basin species: implications for community invasibility and vegetation dynamics. Journal of Ecology 91: 36-48.  Bradley, B.A. & Mustard, J.F. (2006). Characterizing the landscape dynamics of an invasive plant and risk of invasion using remote sensing. Ecological Applications 16: 1132-1147.  Burns, K.C. (2007). Patterns in the assembly of an island plant community. Journal of Biogeography 34: 760-768.  Callaway, R.M., Ridenour, W.M., Laboski, T., Weir, T. & Vivanco, J.M. (2005). Natural selection for resistance to the allelopathic effects of invasive plants. Journal of Ecology 93: 576-583.  Callaway, R.M. (1995). Positive interactions among plants. The Botanical Review 61: 306-349.  52  Callaway, R.M., Nadkarni, N.M. & Mahall., B.E. (1991). Facilitation and interference of Quercus douglasii on understory productivity in central California. Ecology 72: 1484-1499.  Chambers, J.C., Roundy, B.A., Blank, R.R., Meyer, S.E. & Whittaker, A. (2007). What makes Great Basin sagebrush ecosystems invasible by Bromus tectorum? Ecological Monographs: 77: 117-145.  Chave, J., Muller-Landau, H.C. & Levin, S.A. (2003). Comparing classical community models: theoretical consequences for patterns of diversity. American Naturalist 159: 1-22.  Christian, J.M. & Wilson, S.D. (1999). Long-term ecosystem impacts of an introduced grass in the northern Great Plains. Ecology 80: 2397-2407.  Clements, D. & Young, J. (2007). Experiencing the success and failures of rangeland restoration/revegetation. Society for Range Management Meeting, Nevada. Accessed on 14/01/2010 at: http://www.ushrl.saa.ars.usda.gov/research/publications/publications.htm?seq_no_115=1 98822&pf=1  Corbin, J.D. & D’Antonio, C.M. (2004). Competition between native perennial and exotic annual grasses: implications for an historical invasion. Ecology 85: 1273- 1283.  Cox, R.D. & Anderson, V.J. (2004). Increasing native diversity of cheatgrass-dominated rangeland through assisted succession. Journal of Range Management 57: 203-210.  Crawley, M.J. & Harral, J.E. (2001). Scale dependence in plant biodiversity. Science 291: 864-868.  Crone, E.E., Marler, M. & Pearson, D.E. (2009). Non-target effects of broadleaf herbicide on a native perennial forb: a demographic framework for assessing and minimizing impacts. Journal of Applied Ecology 46: 673-682.  Crowe, C. (2008). British Columbia: the grasslands before us. BC Grasslands: Magazine of the grasslands conservation council of British Columbia. Winter 2008/2009 Special Edition, p8-9.  Currie, P.O., Volesky, J.D., Hilken, T.O. & White, R.S. (1987). Selective control of annual bromes in perennial grass stands. Journal of Range Management 40: 547- 550.  D’Antonio, C.M. & Thomsen, M. (2004). Ecological resistance in theory and practice. Weed Technology 18: 1572-1577.  53  D’Antonio, C.M & Vitousek, P.M. (1992). Biological invasions by exotic grasses, the grass/fire cycle, and global change. Annual Review of Ecological Systematics 23: 63-87.  Davies, K.W. & Svejcar, T.J. (2007). Comparison of medusahead-invaded and non- invaded Wyoming big sagebrush-steppe in southeastern Oregon. Rangeland Ecology Management 61: 623-629.  Davies, K.W., Svejcar, A.J. & Bates, J.D. (2009). Grazing can help western rangelands recover from fire. Oregon Beef Producer 22: 12-14.  Davis, M.A, Grime, J.P. & Thompson, K. (2000). Fluctuating resources in plant communities: a general theory of invasibility. Journal of Ecology 88: 528-534.  Davison, J.C. (2007). Imazapic provides 2-year control of weedy annuals in a seeded Great Basin fuelbreak. Native Plants Journal 8: 91-95.  De Wit, M.P., Crookes, D.J. & Van Wilgen, B.W. (2001). Conflicts of interest in environmental management: Estimating the costs and benefits of tree invasion. Biological Invasions 3: 167-178.  Di’Tomaso, J.M. (2000). Invasive weeds in rangelands: species, impacts and management. Weed Science 48: 255-265.  Diamond, J.M. (1975). Assembly of species communities. 342-444. In: Cody, M.L. & Diamond, J.M. (eds.) Ecology and Evolution of Communities. Harvard University Press, Cambridge.  Downey, P.O., Williams, M.C., Whiffen, L., K, Auld, B.A., Hamilton, M.A., Burley, A.L. & Turner, P. J. (2010). Managing alien plants for biodiversity outcomes – the need for triage. Invasive Plant Science and Management 3: 1-11.  Dunning, J.B., Danielson, B.J. & Pulliam, H.R. (1992). Ecological processes that affect populations in complex landscapes. Oikos 65: 169-175.  Eaton, W.D. & Farrell, R.E. (2004). Catabolic and genetic microbial indices, and levels of nitrate, ammonium and organic carbon in soil from the black locust (Robinia pseudoacacia) and tulip poplar (Liriodendron tulipifera) trees in a Pennsylvania forest. Biological Fertility of Soils 39: 209-214.  Epanchin-Niel, R., Englin, J. & Nalle, D. (2009). Investing in rangeland restoration in the Arid West, USA: countering the effects of an invasive weed on the long-term fire cycle. Journal of Environmental Management 91: 370-379.   54 Fansler, V., Mangold, J., Borman, M. & Pyke, D. (2007). Increasing native plant diversity in crested wheatgrass stands. Society for Rangeland Management Meeting. Paper 138.  Fargione, J., Brown, C.S. & Tilman, D. (2003). Community assembly and invasion: an experimental test of neutral versus niche theory. Proceedings of the National Academy of Sciences of the United States of America 100: 8916-8920.  Fernandez, M.E., Gyenge, J.E. & Schilchter, T.M. (2006). Growth of Festuca pallescens in silvopastoral systems in Patagonia. Part 1: positive balance between competition and facilitation. Agroforestry Systems 66: 259-269.  Francis, M.G. & Pyke, D.A. (1996). Crested wheatgrass-cheatgrass seedling competition in a mixed-density design. Journal of Range Management 49: 432-438.  Frost, W.E. & McDougald, N.K. (1989). Tree canopy effects on herbaceous production of annual rangeland during drought. Journal of Range Management 42: 281-283.  Fu, Y.B. & Thompson, D. (2006). Genetic diversity of bluebunch wheatgrass (Pseudoroegnaria spicata) in the Thompson River Valley of British Columbia. Canadian Journal of Botany 84: 1122-1128.  Fuhlendorf, S.D. & Smeins, F.E. (1997). Long-term vegetation dynamics mediated by herbivores, weather and fire in a Juniperus-Quercus savanna. Journal of Vegetation Science 8: 819-828.  Garrett, H.E., Kerley, M.S., Ladymna, K.P., Walter, W.D., Godsey, L.D., Vansambeek, J.W. & Brauer, D.K. (2004). Hardwood silvopasture management in north America. Agroforestry Systematics 61: 21-33.  Getz, H.L. & Baker, W.L. (2008). Initial invasion of cheatgrass (Bromus tectorum) into burned Pinon-Juniper woodlands in western Colorado. American Midland Naturalist 159: 489-497.  Goodrich, S. & Rooks, D. (1999). Control of weeds at a pinyon-juniper site by seeding grasses. USDA Forest Service Proceedings RMRS-P-9: 403-407.  Gordon, D.R. (1998). Effects of invasive, non-indigenous plant species on ecosystem processes: lessons from Florida. Ecological Applications 8: 975-989.  Gotelli, N.J. & Entsminger, G.L. (2006). EcoSim: nullmodels software for ecology. Version 7.0. Acquired Intelligence Inc. and Kesey-Bear. Jericho, VT05465. http://garyentsminger.com/ecosim.htm.  Gotelli, N.J. & Entsminger, G.L. (2001). Swap and fill algorithms in null model analysis: rethinking the Knight's Tour. Oecologia 129: 281-291.  55  Gotelli, N.J. & McCabe, D.J. (2002). Species co-occurrence: a meta-analysis of J. M. Diamond’s Assembly Rules Model. Ecology 83: 2091-2096.  Griffith, A.B. (2010). Positive effects of native shrubs on Bromus tectorum demography. Ecology 91: 141-154.  Haubensak, K., D’Antonio, C. & Wixon, D. (2009). Effects of fire and environmental variables on plant structure and composition in grazed salt desert shrublands of the Great Basin (USA). Journal of Arid Environments 73: 643-650.  Hobbs, R.J. (2000). Land-use changes and invasions. in: Mooney, H. A. & Hobbs, J.R., editors. Invasive species in a changing world. D.C Island Press. Washington.  Hubbell, S.P. (2001). The Unified Neutral Theory of Biodiversity and Biogeography. Princeton University Press, Princeton.  Hubbell, S.P. (2005). Neutral theory in community ecology and the hypothesis of functional equivalence. Functional Ecology 19: 166-172.  Hughes, R, Vitousek, P.M. & Tunison, T. (1991). Alien grass invasion and fire in the seasonal submontane zone of Hawai’i. Ecology 72: 743-746.  Humphrey, L.D. & Schupp, E.W. (2001). Seed banks of Bromus tectorum-dominated communities in the Great Basin. Western North American Naturalist 61: 85-92.  Hutchinson, G.E. (1957). Concluding remarks. Cold Spring Harbour Symposium on Quantitative Biology 22: 415-427.  Kadmon, R. & Shmida, A. (1990). Spatiotemporal demographic processes in plant populations: an approach and a case study. American Naturalist 135: 382-397.  Kota, N.L., Landenberger, R.E. & McGraw, J.B. (2007). Germination and early growth of Ailanthus and tulip poplar in three levels of forest disturbance. Biological Invasions 9: 197-211.  Larson, S.R., Jones, T.A., Hu, Z-M., McCracken, C.L. & Palazzo, A. (2000). Genetic diversity of bluebunch wheatgrass cultivars and a multiple-origin polycross. Crop Science 40: 1142-1147.  Mack, R.N. (1981). Invasion of Bromus tectorum L. into western North America: an ecological chronicle. Agro-Ecosystems 7: 145-165.  Maestre, F.T., Callaway, R.M., Valladares, F. & Lortie, C.J. (2009). Refining the stress- gradient hypothesis for competition and facilitation in plant communities. Journal of Ecology 97: 199-205.  56  Manly, B. & Sanderson, J.G. (2002). A note on null models: justifying the methodology. Ecology 83: 580-582.  Mazzola, M.B., Allcock, K.G., Chambers, J.C., Blank, R.R, Schupp, E.W., Doescher, P.S. & Nowak, R.S. (2008). Effects of nitrogen availability and cheatgrass competition on the establishment of Vavilov Siberian Wheatgrass. Rangeland Ecology and Management 61: 475-484.  McConnell, B.R. & Smith, J.G. (1971). Effect of Ponderosa Pine needle decomposition on grass seedling survival. USDA Forest Service Research Note PNW-155: 1-6.  Melgoza, G. & Nowak, R.S. (1991). Competition between cheatgrass and two native species after fire: implications from observations and measurements of root distribution. Journal of Range Management 44: 27-33.  Melgoza, G, Nowak, R.S. & Tausch, R.J. (1990). Soil water exploitation after fire: competition between Bromus tectorum (cheatgrass) and two native species. Oecologia 83: 7-13.  Meyer, S.E., Allen, P.S. & Beckstead, J. (1997). Seed germination regulation in Bromus tectorum L. (Poaceae) and its ecological significance. Oikos 78: 475-485.  Miles, E.K. & Knops, J.M.H. (2009). Shifting dominance from native C4 to non-native C3 grasses: relationships to community diversity. Oikos 118: 1844-1853.  Moir, W.H. (1966). Influence of ponderosa pine on herbaceous vegetation. Ecology 47: 1045-1048.  Monaco, T.A., Waldron, B.L., Newhall, R.L. & Horton, W.H. (2003). Re-establishing perennial vegetation in cheatgrass monocultures. Rangelands 25: 26-29.  Monsen, S.B. (1994). Selection of plants for fire suppression on semi-arid sites. 363-373. In: Monsen, S.B. & Kitchen, S.G. (1994). Proceedings-symposium on ecology and management of annual rangelands. Boise, ID. General technical report: INT-GTR- 313. USDA Forest Service, Intermountain research Station, Ogden, Utah.  Morales, C.L. & Traveset, A. (2009). A meta-analysis of impacts of alien vs. native plants on pollinator visitation and reproductive success of co-flowering native plants. Ecology Letters 12: 716-728.  Morrow, L.A. & Stahlman, P.W. (1984). The history and distribution of downy brome (Bromus tectorum) in North America. Weed Science 32. Supplement 1.   57 Myers, J.H., Jackson, C., Quinn, H., White, S.R. & Cory, J.S. (2009). Successful biological control of diffuse knapweed, Centaurea diffusa, in British Columbia, Canada. Biological Control 50: 66-72.  Ortega, Y.K., & Pearson, D.E. (2005). Weak vs. strong invaders of natural plant communities: assessing invasibility and impact. Ecological Applications 15: 651- 661.  Ott, J.E, McArthur, E.D. & Roundy, B.A. (2003). Vegetation of chained and non-chained seedings after wildfire in Utah. Journal of Range Management 56: 81-91.  Pejchar, L. & Mooney, H.A. (2009). Invasive species, ecosystem services and human well-being. Trends in Ecology and Evolution 24: 497-504.  Pellant, M., Kaltenecker, J. & Jirik, S. (1999). Use of OUST herbicide to control cheatgrass in the northern Great Basin. USDA Forest Service Proceedings RMRS- P-9: 322-326.  Pielou, D.P. & Pielou, E.C. (1968). Association among species of infrequent occurrence: the insect and spider fauna of Polyporus betulinus (Bulliard) Fries. Journal of Theoretical Biology 21: 202-216.  Pimentel, D., Zuniga, R. & Morrison, D. (2005). Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52: 273-288.  Pollock, K.M., Mead, D.J. & McKenzie, B.A. (2009). Soil moisture and water use by pastures and silvopastures in a sub-humid temperate climate in New Zealand. Agroforestry Systems 75: 223-238.  Reid, C.R., Goodrich, S. & Bowns, J.E. (2008). Cheatgrass and Red Brome: history and biology of two invaders. USDA Forest Service Proceedings RMRS-P-52: 27-32.  Reshi, Z., Rashid, I., Khuroo, A.A. & Wafai, B.A. (2008). Effect of invasion by Centaurea iberica on community assembly of a mountain grassland of Kashmir Himalaya, India. Tropical Ecology 49: 147-156.  Rickard, W.H., Uresk, D.W. & Cline, J.F. (1975). Impact of cattle grazing on three perennial grasses in south-central Washington. Journal of Range Management 28: 108-112.  Rowe, H.I. & Brown, C.S. 2008. Native plant growth and seedling establishment in soils influenced by Bromus tectorum. Rangeland Ecology & Management 61: 630-639.  Sandel, B. & Smith, A.B. (2009). Scale as a lurking factor: incorporating scale- dependence in experiemental ecology. Oikos 118: 1284-1291.  58  Sanders, N.J., Gotelli, N.J., Heller, N.E. & Gordon, D.M. (2003). Community disassembly by an invasive species. Proceedings of the National Academy of Sciences of the United States of America 100: 2474-2477.  Sax, D.F. (2002). Native and naturalized plant diversity are positively correlated in scrub communities of California and Chile. Diversity and Distributions 8: 193-210.  Simberloff, D. (2005). Non-native species DO threaten the natural environment. Journal of Agricultural and Environmental Ethics 18: 595-607.  Simberloff, S. & Von Holle, B. (1991). Positive interactions of non-indigenous species: invasional meltdown. Biological Invasions 1: 21-32.  Smith, D.C., Meyer, S.E. & Anderson, V.J. (2008). Factors affecting Bromus tectorum seed bank carryover in Western Utah. Rangeland Ecology Management 61: 430- 436.  Spira, T. (2001). Plant-pollinator interactions: a threatened mutualism with implications for the ecology and management of rare plants. Natural Areas Journal 21: 78-88.  Stephens, A.E.A., Krannitz, P.G. & Myers, J.H. (2009). Plant community changes after the reduction of an invasive rangeland weed, diffuse knapweed, Centaurea diffusa. Biological Control 51: 140-146.  Stone, L. & Roberts, A. (1990). The checkerboard score and species distributions. Oecologia 85: 74-79.  Thelin, G., Rosengren-Brinck, U., Nihlgard, B. & Barkman, A. (1998). Trends in needle and soil chemistry of Norway spruce and Scots pine stands in South Sweden 1985- 1994. Environmental Pollution 99: 149-158.  Theoharides, K.A. & Dukes, J.S. (2007). Plant invasion across space and time: factors affecting non-indigenous species success during four stages of invasion. New Phytologist 176: 256-273.  Tilman, D. (1988). Plant strategies and the dynamics and structure of plant communities. Princeton University Press, Princeton.  Turnbull, L.A., Manley, L. & Rees, M. (2005). Niches, rather than neutrality, structure a grassland pioneer guild. Proceedings of the Royal Society B: Biological Sciences 272: 1357-1364.  Upadhyaya, M.K., Turkington, R. & McIlvride, D. (1986). The biology of Canadian weeds. 75. Bromus tectorum L. Canadian Journal of Plant Science 66: 689-709.   59 Valliant, M.T., Mack, R.N. & Novak, S.J. (2007). Introduction history and population genetics of the invasive grass Bromus tectorum (Poaceae) in Canada. American Journal of Botany 94: 1156-1169.  Vidra, R.L., Shear, T.H. & Wentworth, T.R. (2006). Testing the paradigms of exotic species invasion in urban riparian forests. Natural Areas Journal 26: 339-350.  Vitousek, P.M., D’Antonio, C.M., Loope, L.L. & Westbrooks, R. (1996). Biological invasions as global environmental change. American Scientist 84: 468-478.  Whittaker, A., Roundy, B., Chambers, J., Meyer, S., Blank, R.R., Kitchen, S.G. & Korfmacher, J. (2008). The effect of herbaceous species removal, fire, and cheatgrass (Bromus tectorum) on soil water availability in sagebrush steppe. USDA Forest Service Proceedings RMRS-P-52.  Willis, K.J. & Whittaker, R.J. (2002). Species diversity – scale matters. Science 295: 1245-1248.  Woods, M. & Moriarty, P.V. (2001). Strangers in a strange land: the problem of exotic species. Environmental Values 10: 163-191.  Yavitt, J.B. & Smith Jr., L. (1983). Spatial patterns of mesquite and associated herbaceous species in an Arizona desert upland. American Midland Naturalist 109: 89-93.  Young, J.A. & Allen, F.L. (1997). Cheatgrass and range science: 1930-1950. Journal of Range Management 50: 530-535.  Young, J.A & McKenzie, D. (1982). Rangeland drill. Rangelands 4: 108-113.                 

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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.24.1-0071034/manifest

Comment

Related Items