SILENE SPALDINGII (SPALDING S CATCHFLY) POPULATION VIGOR AND COMMUNITY CHARACTERISTICS IN ASOTIN COUNTY, SOUTHEASTERN WASHINGTON TARYN BETH CLARK

Transcription

1 SILENE SPALDINGII (SPALDING S CATCHFLY) POPULATION VIGOR AND COMMUNITY CHARACTERISTICS IN ASOTIN COUNTY, SOUTHEASTERN WASHINGTON By TARYN BETH CLARK A thesis submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE IN NATURAL RESOURCE SCIENCES WASHINGTON STATE UNIVERSITY Department of Natural Resource Sciences AUGUST 2010

2 To the Faculty of Washington State University: The members of the Committee appointed to examine the thesis of TARYN BETH CLARK find it satisfactory and recommend that it be accepted. Linda H. Hardesty, Ph.D., Chair Lisa. A. Shipley, Ph.D. Rich Alldredge, Ph.D. ii

3 ACKNOWLEDGEMENTS Thank you to the many people who helped me throughout the past two years on my project. I would like to extend my appreciation to my advisor Linda Hardesty for her personal and professional encouragement. Thanks to Lisa Shipley for her energy and willingness to help. Thanks also to Rich Alldredge for his patience and support. Thanks to Mark Swanson for his input and ideas. A special thank you goes out to Laura Applegate and Sara Wagoner for sharing in my graduate school experience and making our project fun and memorable. Thank you to Becky Greenwood, Rachel Granberg, and Ellen Miller for their field help and humor. Thanks to Bob Dice, Mel Asher, and the Washington Department of Fish Wildlife staff. Funding was provided by the Washington Department of Fish and Wildlife, the Washington Cattlemen s Association, and Washington State University. Thank you to my family and friends, old and new. I am so lucky to have such a supportive network of friends no matter where I am. Thank you to my sister for always putting up with me. To my mom and dad, thanks for helping me see I could do it all along. iii

4 SILENE SPALDINGII (SPALDING S CATCHFLY) POPULATION VIGOR AND COMMUNITY CHARACTERISTICS IN ASOTIN COUNTY, SOUTHEASTERN WASHINGTON Abstract by Taryn Beth Clark, M.S. Washington State University August 2010 Chair: Linda H. Hardesty Silene spaldingii (Spalding s catchfly) is a perennial forb, endemic to the inland Pacific Northwest and listed as threatened by the U.S. Fish and Wildlife Service. The plant inhabits landscapes subject to historical land-use change and degradation, including bunchgrass, sagebrush-steppe, and ponderosa pine communities. In a recently discovered population in southeastern Washington bunchgrass habitat, we compared sites that contained Silene spaldingii to sites without the plant to determine its growth within suitable habitats on northerly aspects. Silene spaldingii was found between aspects of , on slopes up to 45, and was associated with lower elevations with higher percent ground cover of litter than sites without it. Silene spaldingii was more likely to be absent in sites with higher percent canopy cover of native annual forbs, total exotic species, and exotic annual grasses and forbs. Silene spaldingii was more likely to be absent from sites with higher species richness of exotic annual grasses and forbs, exotic perennial grasses, and total species richness. Canopy cover of exotic annual forbs and species richness were the best predictors of the absence of Silene spaldingii. My research iv

5 suggests that competition for resources by exotic species invasions seems to be a factor explaining the absence of Silene spaldingii from otherwise suitable habitat. Recruitment of new plants by Silene spaldingii is most likely the limiting factor in population growth. Silene spaldingii inhabits mid to late seral native perennial bunchgrass communities with few exotic species, and that appear to be resistant to exotic invasions when undisturbed. Silene spaldingii is an indicator of high quality steppe habitat. The amount of insect herbivory, yellowing of plants in June, and non-reproductive stems were different between pastures, but the reasons were not clear. Monitoring for weed invasions and active management to reduce weed infestations are important steps towards protecting the Silene spaldingii population. Yearly, long-term monitoring of patterns in population, community, and vigor characteristics could help us understand the status of the Smoothing Iron population. v

6 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS... iii ABSTRACT... iv LIST OF TABLES... viii LIST OF FIGURES... iv INTRODUCTION...1 Questions...5 MATERIALS AND METHODS...6 Study area...6 Survey for additional populations...7 Population and plant characteristics monitoring...8 Site characteristics and associated plant community sampling...10 Analytical methods...11 RESULTS AND DISCUSSION...13 Size and extent of the population...13 Community composition and the presence or absence of Silene spaldingii...16 Silene spaldingii vigor in macroplots...22 Timing of monitoring...26 Conclusion...27 LITERATURE CITED...43 APPENDICES...48 DATA FORM FOR JUNE MACROPLOT MONITORING...49 vi

7 DATA FORM FOR JULY MACROPLOT MONITORING...50 SPECIES LIST...51 vii

8 LIST OF TABLES 1. Silene spaldingii habitat types within project boundaries Silene spaldingii population estimate at Smoothing Iron Unit Potential suitable habitat at Smoothing Iron Unit Single variable logistic regression tests Correlation of community variables for collinearity Multiple variable logistic regression models Correlation of community variables and Silene spaldingii plant characteristics Correlation of Silene spaldingii plant characteristics Silene spaldingii plant characteristic comparison by pasture Animal scats in macroplots June to July plant comparison...38 viii

9 LIST OF FIGURES 1. Map of Silene spaldingii populations Map of study area Macroplot sampling layout Map of suitable habitat at the Smoothing Iron Unit...42 ix

10 INTRODUCTION Silene spaldingii (Spalding s catchfly) is a perennial forb listed as Threatened by the U.S. Fish and Wildlife Service (USFWS 2001). It is endemic to the inland Pacific Northwest and inhabits bunchgrass, sagebrush-steppe, and ponderosa pine communities (Hill and Gray 2004). The area includes the Palouse Grasslands, the Canyon Grasslands (Washington, Idaho, and Oregon), the Channeled Scablands, the basaltic plateaus of Northeastern Oregon, and the intermontane valleys of northwestern Montana and British Columbia (Hill and Gray 2004; USFWS 2007; Figure 1). Much of the habitat has been developed, converted to agriculture, and used as rangelands. Perceived threats to Silene spaldingii include habitat loss and fragmentation, invasive species, grazing and trampling, altered fire regimes, recreation, herbicides, and climate change (USFWS 2007). The plant primarily grows in deep soil of more mesic sites on northerly aspects dominated by Festuca idahoensis or Festuca idahoensis/pseudoroegneria spicata grasslands between elevations of 365 to 1,615 meters (USFWS 2007). Invasion by exotic species into the semi-arid grassland steppe habitat of Silene spaldingii is a widespread problem. Invasions may be most detrimental to rarer species which rely on specific habitat requirements, both community and environmental, further fragmenting and reducing naturally small populations (Kephart and Paladino 1997; Menke and Muir 2004). Competition for water, light, nutrients, and space are exacerbated when exotic species invade native communities. Silene spaldingii established adult plants may be less susceptible to competition from invasives because they have a long tap root able to access deep soil water (Kephart and Paladino 1997; Menke and Muir 2004). But, germination and growth of new 1

11 recruits (seedlings) may be adversely affected because invasives more readily alter the resources close to the surface seedlings compete for, such as light and water (Menke and Muir 2004). Silene spaldingii exhibits three growth forms (rosette, non-reproductive stem, and flowering stem) and a dormant stage (Lesica 1997). Vegetative growth begins in spring, with basal rosettes evident in late May to early June. Basal rosettes that are apparent in June may wither and die back during July and do not produce stems in the same year (Gray and Hill 2006). A plant that produces a stem may either produce a vegetative stem only, or reproductive stem during that growing season. One plant may send up more than one structure, including a combination of a rosette(s) and stemmed structure(s) (Gray and Hill 2006). The age of a plant does not determine whether the plant produces only a basal rosette, or a vegetative stem, or a reproductive stem, making demographic studies more complex due to the unpredictability of the plant s growth strategy (Hill and Weddell 2003, Hill and Gray 2005). Furthermore, the plant may spend one or more years dormant, confounding the ability to identify a plant s age, or if it has died or gone dormant. Four years is the earliest time frame to accurately assess the age of individual plants in a population (Gray and Hill 2006). Monitoring both in June and again in July provides more accurate estimates of the population due to new plants arising by July (stems and reproductive plants) and rosette plants withering by the same duration (Gray and Hill 2006). In one demography study using a June and July census at Craig Mountain south of Lewiston, Idaho, more plants grew in 2005 than Yet, if only a July census had been conducted the results would have shown more plants grew in 2004 than 2005 (Gray and Hill 2006). Silene spaldingii grows from a long taproot, sometimes greater than 85 cm (Menke and Muir 2004). Stem heights may be up to 60 cm tall, with stems bearing 4 to 7 pairs of opposite 2

12 leaves, characteristic of the family Caryophyllaceae to which it belongs (Lesica 1997, Hill and Gray 2004). The entire plant is covered in sticky gland-tipped hairs. Flowers are inconspicuous, usually cream, light pink, or green colored with five clawed petals inside sepals forming a calyx tube (Hitchcock et al. 1964). Only a small, two mm portion of the corolla claw is visible, making it a diagnostic feature (Hitchcock et al. 1964; Hill and Gray 2004; Gray and Hill 2006). The orientation of fertilized flowers changes from perpendicular to the stem, to pointing upward making it easy to observe (Gray and Hill 2006). When only a rosette of Silene spaldingii is available, the leaves have characteristic retrorse (bent backward or curved downward) hairs along the petioles which may be used to identify the seedlings (Gray and Hill 2006). Dormancy in Silene spaldingii appears to be linked to both internal and external factors resulting in enhanced fitness for the plant. When compared to plants that had a previous year of vegetative growth, plants that had a period of dormancy were more likely to flower following dormancy (Lesica and Crone 2007). The plant reproduces only by seed with episodic recruitment events (Gray and Hill 2006). Gray and Hill (2006) suggested that favorable environmental factors (precipitation) may play a more important role in initiation of reproduction (germination and breaking dormancy) than seed production. In a 5 year study in Montana, Silene spaldingii exhibited dormancy at a rate of 10%, with lower dormancy and more flowering following presumably favorable environmental conditions (Lesica and Steele 1994). Dormancy was more common following flowering or less beneficial precipitation patterns the previous year (too wet or too dry) suggesting dormancy as a resource conservation strategy for the plant and a response to environmental stresses (Lesica and Crone 2007). 3

13 Dormancy complicates monitoring plant populations. Repeated measures on permanent plots allow botanists to accurately assess a species population when the species exhibits dormancy as a normal part of its life cycle (Lesica and Steele 1994). Lesica and Steele (1994) recommended repeated sampling for 3 or more consecutive years, every 5 20 years (depending on a plant s perceived lifespan) to reduce error from short-term variation in growth forms, climate, and dormancy patterns. Disturbance likely plays an important role in the survival of Silene spaldingii populations. Fire has been shown to significantly increase the recruitment of seedlings in the initial years following the burn that removes the litter layer (Lesica 1999). A dense litter layer is thought to inhibit germination and seedling survival of some plants (Facelli and Pickett 1991; Golberg and Werner 1983; Lesica 1999). Fire warms the soil and increases available nutrients, both of which may contribute to increased germination and recruitment (Daubenmire 1968). Grazing also has potential to remove litter (Lesica 1999). However, disturbances may contribute to exotic weed expansion in areas with infestations because invasive species can exploit the conditions created by fire or soil disturbance, and may recover or establish more quickly than natives. Grazing and animal movements have the potential to move invasive species seeds through attachment to an animal s fur or through fecal deposition. Silene spaldingii does not have a formally defined hierarchy of subpopulations, populations, and metapopulations (Lichthardt and Gray 2003). Instead, clusters are defined as groups of plants clumped together with a distribution typical of plants without wind-dispersed seed, with very few plants between clusters (Lichthardt and Gray 2003). The main pollinator is Bombus fervidus, the Golden Northern bumble bee, which exhibits a home range of 4

14 approximately 1.6 km 2. In turn, Silene spaldingii plants further than 1.6 km apart are considered to exist within separate populations based on separate pollination/reproduction groups. In September 2008 a field survey was conducted for Silene spaldingii within cattlegrazed pastures in the Smoothing Iron and Pintler Creek Units of the Blue Mountain Wildlife Area, in Asotin County, WA. The area is managed by the Washington Department of Fish and Wildlife (WDFW). Because a large population of Silene spaldingii was found at the Smoothing Iron Wildlife Unit, WDFW decided not to graze those pastures during the 2009 rotation until further surveys and monitoring determined the extent of the population, the condition of the habitat, and requirements necessary to ensure the survival of Silene spaldingii (WDFW 2009). This project addressed WDFW s need to survey and monitor Silene spaldingii populations and habitat on their lands. It also quantified Silene spaldingii habitat based on environmental attributes and plant community composition. Research questions 1. What is the population distribution of Silene spaldingii in the Smoothing Iron and Pintler Creek Units? 2. What environment or community variables are associated with Silene spaldingii habitat? 3. Do Silene spaldingii plant characteristics have relationships with community and environmental variables, or among themselves? 4. Does plant vigor differ by pasture? 5. Are two monitoring periods necessary to get an accurate population count? 5

15 MATERIALS AND METHODS Study area The study occurred in the Blue Mountain Wildlife Area Complex managed by the Washington Department of Fish and Wildlife (WDFW), in Asotin County, Washington. The Pintler Creek Unit (Figure 2), 8 km southwest of Asotin, WA, was divided into three pastures roughly based on the associated drainages Pintler Creek (Owl pasture), Kelly Creek, and Ayers Gulch. The total study area was approximately 1,732 ha with a mean elevation of 396 m, characterized by canyons with steep slopes up to 60%. Average annual precipitation was approximately 33 cm, most of which falls from October to June, with dry, hot summers. Grassland steppe communities included Pseudoroegneria spicata, Festuca idahoensis, Poa cusickii, and Poa secunda grasses which comprised the majority of the upland area, with few shrubs scattered throughout. In general, the area included a high occurrence of Bromus tectorum and associated non-native plants (personal observation). Natural Resources Conservation Service (NRCS) ecological sites in the Pintler Creek Unit included Cool Stony 15+ precipitation zone (PZ), Cool Loamy 9-15 PZ, and Dry Stony 9-15 PZ, with some Stony 9-15 PZ and Loamy 9-15 PZ (NRCS 2004). The Smoothing Iron Unit (Figure 2) was located 21 km southwest of Asotin, WA, between the North and South Fork Asotin Creek drainages. The unit was divided into six pastures. The total area comprised approximately 1,012 ha and with elevation reaching approximately 1,250 m on the upper ridges based on data in ArcGIS (ESRI, ). Average annual precipitation approaches 51 cm based on the Anatone weather station. Steep slopes and grassland steppe communities dominated, but with a higher shrub component than the Pintler Creek Unit. Common shrubs included Symphoricarpos albus and Rosa woodsii, with 6

16 some Pinus ponderosa found in more mesic locations. Non-native grasses and forbs occurred, but with less frequency than at the Pintler Creek Unit (personal observation). NRCS ecological sites on the Smoothing Iron Unit predominantly included Cool Loamy 15+ PZ and Dry Stony 15+ PZ with some Very Shallow 15+ PZ and Loamy 15+PZ (NRCS 2004). Survey for additional populations Surveys for Silene spaldingii within project boundaries were incomplete and suitable habitat was thought to exist in additional locations at Smoothing Iron and possibly the Pintler Creek Units (Gray 2008). Because known Silene spaldingii populations occurred within Pastures 3, 4, and 5, we surveyed these locations first to become familiar with the plant and its suitable habitat starting on June 6, Additional locations in those pastures, plus Pastures 2 and 6, were surveyed where northeast to northwest aspects occur with suitable grassland steppe communities dominated by Festuca idahoensis and Pseudoroegneria spicata. Areas searched included those supporting species often found with Silene spaldingii such as Koeleria macrantha, Poa cusickii, Arnica sororia, Besseya rubra, Erigeron corymbosus, Frasera albicaulis, Geum triflorum, Hieracium albertinum, Penstemon glandulosus, Solidago missouriensis, Artemisia ludoviciana, Symphoricarpos albus, and Rosa woodsii (Washington Natural Heritage Program 2009). Plants associated with the Smoothing Iron Unit s Silene spaldingii population also include Lithospermum ruderale, Heuchera ssp., and Calochortus macrocarpus var. maculosus (Gray 2008). Table 1 displays possible habitat types within each wildlife unit based on The Washington Natural Heritage Program (WNHP) Silene spaldingii habitat types. 7

17 In areas of known Silene spaldingii occurrences at Smoothing Iron, we walked transects systematically back and forth across the landscape following the hillside contour. Intervals between contours were roughly between 1-3 m, depending on the ease of looking into the vegetation while looking upslope and the steepness of the topography. We recorded locations of clusters with a Global Positioning System (GPS) device as they were found. In areas where suitable habitat should exist but was not yet documented, we walked a wider zigzag pattern. If we found plants, the area was combed in more detail, following the contour method described above. Plants were marked with pin flags or with flagging tape tied loosely around the stem to avoid duplicate counting, and to mark the survey s progress. We returned to some of the earliest surveyed locations as the season progressed because growth made larger plants easier to see amongst the grasses and shrubs. At Pintler Creek we conducted surveys for Silene spaldingii in August Northerly aspects with suitable habitat were surveyed in Ayers Gulch, Owl, and Kelly Creek pastures. The same method of a wide zigzag pattern was walked through the areas. In addition, the WDFW rangeland ecologist was also looking for Silene spaldingii while performing utilization surveys on north aspects during the summer of Population and plant characteristics monitoring Based on the plants located during the survey, we identified clusters within Pastures 2, 3, 4, and 5. Three clusters were randomly chosen from each pasture for a total of 12 monitoring sites. Modifying the method used by Lichthardt and Gray 2003, we established a 10 m x 10 m permanent macroplot within each cluster (Figure 3). A random starting coordinate to place the lower left corner of the macroplot was chosen ensuring that the plot stayed within the cluster. 8

18 The macroplot borders were temporarily flagged at each corner to determine if the macroplot had at least five visible Silene spaldingii plants (the criteria set as the minimum to keep the macroplot location). The macroplot was moved to another random location until it met the minimum criteria. We staked the four corners of each macroplot with 2 m rebar. To ensure the transect boundaries were permanent, each of the 10 belt transects had a 30.5 cm aluminum nail staked on each end. A temporary measuring tape was secured along each transect by the aluminum nails with metal staples. We mapped all the Silene spaldingii plants in the macroplot to determine the identity of a plant from one year to another so that any dormancy that occurred could be accounted for. The lower left corner of the macroplot started at 0 m, 0 m and measured to 10 m going uphill and 10 m perpendicular to uphill (mimicking the X, Y coordinate system). Coordinates of each plant location in meters and centimeters, along with the plant characteristics were recorded from June 18 to July 2, 2009 on a data form (Appendix 1). The characteristics included growth form (stem, rosette, or both), number of stems, number of rosettes, number of rosette leaves, the longest rosette leaf length in centimeters, herbivory (insect or mammal), yellowing or withering of the plant, and notes. We conducted a second set of observations from 28 to 30 July 2009 on a second data form (Appendix 2) to observe changes in the plants as the season progressed. Characteristics recorded during the second round included the longest stem height, number of buds and unfertilized flowers, number of fertilized flowers, number of capsules, whether the plant was alive, dead, or absent, and notes. These plant characteristics describe the vigor, or health and ability to reproduce, of the Silene spaldingii population. In addition, for each macroplot the site name, pasture, date, recorder(s)/observer(s), elevation, aspect, GPS locations, additional plants 9

19 found and growth form, scat of elk, deer, and cattle, disturbances and animal trails, and any other relevant site notes were recorded. Site characteristics and associated plant community sampling Within each macroplot, cover, species richness, frequency, and density estimates were recorded during the first sampling period to represent the communities that support Silene spaldingii. The three transects we randomly chose to gather data from were 2, 5, and 8. Every meter along the transect, excluding 0 and 10 to reduce the effect of the plot boundaries, a pin flag was dropped and all species intercepted recorded. Litter and ground cover were also recorded when hit by the pin flag. This method is similar to the Line-point Intercept (LPI) method (Herrick et al. 2005). The LPI method provides percent canopy cover estimates for plants, and percent ground cover for litter and soil surfaces. Six microplots of size 0.5 m x 1.0 m were randomly chosen to be placed at 2 m and 7 m in the same transects as the LPI measurements (see Figure 1). We counted all species present in the microplot to estimate species frequency. Additionally, the species richness (a count of every species present) was recorded for the entire macroplot. These methods quantify plant community composition and site characteristics associated with Silene spaldingii. To compare habitat associated with Silene spaldingii plants to potential habitat that is considered suitable but does not contain the plant, additional data from monitoring plots from a concurrent study at the Smoothing Iron Unit were used. We defined potential suitable habitat as those areas occurring on northerly aspects with Festuca idahoensis and/or Pseudoroegneria spicata as the dominant grasses, with or without a shrub component. This definition was based on field survey observations and literature review. The 10 plots chosen that fit this criteria were 10

20 in Pastures 2, 3, 4, and 5, with aspects within the range of 285 to 75 and on slopes between 0 to 45, the same range where Silene spaldingii plants were found during the survey and macroplot monitoring. An aspect and slope layer were created from an 1/3 Arc Second, or 10 m pixel resolution, Digital Elevation Maps of the area obtained from The National Map Seamless Server (USGS 2010) using the program ArcGIS. GPS locations of the plots were mapped in ArcGIS and then used with the aspect and slope layers to extract the values of those plots. Additionally, the suitable slope and aspect layers were used to create a habitat map of the Smoothing Iron Unit based on pasture boundaries. None of the plots chosen by our criteria contained Silene spaldingii, and are referred to from now on as the absence plots. Both the plots with Silene spaldingii (presence plots) and the absence plots were found within the bounds of the suitable habitat criteria. The absence plots were also sampled for cover and species richness along each transect using the same methods (LPI and species richness) as the macroplots (those containing Silene spaldingii), with the main difference being the shape and size of the plots. The absence plots have three, 50 m transects arranged at angles of 120 between them, forming a larger circular plot. Analytical methods Population estimates were based on survey counts and corrected for sightability (missed plants during the survey) per pasture using a double sampling. Based on the difference in the number of plants previously found (flagged or tagged during the survey) and the number of additional plants discovered during the macroplot Silene spaldingii mapping, an estimate of the number of plants missed was calculated (double sampling, Morrison et al. 2008). A proportion was calculated per macroplot by taking the number of extra plants found in the macroplot/the 11

21 number of plants originally found during the survey (marked with flags or tape). We used these proportions to estimate the low, median, high, and mean population per pasture by adding the amount missed to the amount found. Survey results were mapped in ArcGIS and the range of slope, aspect, and elevation determined. It should be noted that the pasture boundaries are approximates due to fences being moved and reconstructed, and boundaries drawn in ArcGIS when they were unable to be mapped directly on the ground. We used logistic regression in the program SAS (PROC LOGISTIC - Statistical Analysis Software 9.2, , SAS Institute, Inc., Cary, NC) to determine which individual site variables were associated with the presence or absence (α = 0.05) of Silene spaldingii. The variables used included percent cover data, species richness, slope, aspect, and elevation. The percent cover data included separate variables for total canopy, litter, and soil surfaces. Both the plant percent cover data and the total species richness data were divided into exotic and native categories, and then again into plant functional groups. Plant functional groups are species categorized by native or exotic status, life cycle (perennial, biennial, or annual), and plant type (grass, forb, shrub). Because the communities contained few biennials, they were included in the perennial functional groups for analysis purposes. To determine the multiple logistic regression model(s) for the presence or absence of Silene spaldingii, we first correlated (PROC CORR SAS 9.2) the significant (α = 0.05) individual variables. We considered them collinear if the Pearson correlation coefficient was > 0.50 and significant if P < All uncorrelated variable combinations were included in multiple logistic regression analyses. The criteria for model selection were based on the model with the largest log-likelihood chi-square (α = 0.05) and a model s chi-square improvement 12

22 statistic (Manly et al. 2002). Additionally, each model s variables had Wald statistics of P < 0.10 to be considered significant. To determine whether Silene spaldingii plant characteristics had any relationships with community or environmental variables we used Pearson correlation (PROC CORR SAS 9.2). We also correlated Silene spaldingii plant characteristics with themselves to see relationships amongst plant variables. We used a paired t-test (PROC TTEST SAS 9.2) to compare the total amount of plants found per macroplot in June and July to determine if one sampling period would yield the same density of plants. To explore plant vigor across pastures we used a twoway ANOVA (PROC MIXED SAS 9.2) with pasture as a fixed effect, and macroplot as a random effect to help account for variance across macroplots within a pasture. It should be noted that we did not have replication of treatments by pasture in one season, but rather intended for this analysis to check for differences between plant characteristics by pasture. The purpose was to describe the current status of plant vigor across pastures to provide information, but not to test a treatment effect. We used the least significant difference test (Fisher s LSD, Kuehl 2000) to check which means were different if the ANOVA was significant (α = 0.05). RESULTS AND DISCUSSION Size and extent of the population The survey of the Smoothing Iron Unit yielded a total of 5,879 Silene spaldingii plants in Pastures 2, 3, 4, and 5 (Table 2). Four rosette plants that were not positively identified as Silene spaldingii were found in Pasture 6, and more likely were Silene oregana which had stemmed plants in the same area. Silene spaldingii was not found in the same location as other Silene species (Silene menziesii and Silene oregana), possibly indicating they do not grow 13

23 together because of overlapping niches. Ninety-four plants were found outside the pasture boundaries but adjacent to Pasture 2. Double sampling proportions were different per macroplot mostly due to the detection of basal rosettes during the initial survey (proportions listed in Table 2). We counted and estimated using double sampling the most plants in Pasture 4. The total estimate for Silene spaldingii plants not dormant in 2009 was over 10,000. We found suitable habitat at the Smoothing Iron Unit on slopes from 4 40, aspects between , and at elevations of 750 1,215 m. The range of elevation Silene spaldingii was found at included all elevations at the Smoothing Iron Unit. Therefore, elevation was not used to calculate suitable habitat in this study. We added 5 to each end of the aspect and slope ranges to account for error for the purpose of estimating potential suitable habitat, expanding them to 0 45 for slope and for aspect. Pasture 4 had the most suitable habitat (Figure 4 and Table 3) estimated at ha. The suitable habitat estimate for Pasture 1 was 20 acres, which is high based on experience on the ground. Almost no potential habitat was available based on field surveys of Pasture 1, due to error in the pasture boundary where it crossed into CRP (Conservation Reserve Program) land at the top of the ridge. Even though Pasture 4 had the most Silene spaldingii plants and the largest suitable habitat, Pasture 2 had the highest estimated density of plants/ha compared to Pasture 4 s plants/ha (Table 3). Pasture 5 had the lowest density of plants at plants/ha, with Pasture 3 just slightly higher at plants/ha. In general, much of the suitable habitat in Pasture 5 was either invaded by exotic grasses or had high cover of shrubs and trees, both possibly contributing to the exclusion of Silene spaldingii (T. Clark, personal observation). Pastures 3 and 4 had northerly facing slopes forming draws and steep slopes where there was more tree and shrub cover than grass cover, and less Silene spaldingii was found. Additionally, 14

24 because Silene spaldingii can be dormant in any given year, this estimate is only applicable to plants that had above-ground growth during the 2009 season. Finally, none of these suitable habitat estimates take into account the variation in soils or ecological sites (the NRCS soil data and ecological site descriptions are not detailed enough for Silene spaldingii habitat), or solar radiation intensity at differing aspects, slopes, and elevations. However, these estimates are useful for creating an initial description of the population and informing managers about the extensive habitat in this area to be considered in future management to conserve this species. As of 2007, the total estimate of Silene spaldingii plants (above ground growth) was only 28,750 in the U.S (UWFWS 2007). At that time, only ten populations had greater than 500 plants, with the largest population of over 10,000 plants at The Nature Conservancy s Dancing Prairie Preserve in Montana (UWFWS 2007). The second largest population was at Garden Creek, Idaho, with approximately 4,000 plants where many previous and ongoing Silene spaldingii studies occur (UWFWS 2007). The Smoothing Iron Unit population is a substantial addition to the total remaining Silene spaldingii population. We did not find any Silene spaldingii plants at the Pintler Creek Unit. Very little habitat surveyed seemed suitable and much of it was infested with weedy species such as Bromus tectorum, Bromus arvensis (synonym: Bromus japonicus), and Vicia villosa. The northerly aspects surveyed did contain other species of Silene, including Silene menziesii, another Silene (probably Silene douglasii or Silene scouleri), and Silene alba (exotic). Silene spaldingii may not co-occur with other species of Silene. The survey was done late in the season, so it may be worthwhile to survey another year earlier in the season. 15

25 Community composition and the presence or absence of Silene spaldingii A total of 124 plant species were observed in the presence and absence plots (Appendix 3). The three species unique to the presence plots were Silene spaldingii, Happlopappus ssp., and Acer glabrum. Twenty-seven species were unique to the absence plots, including 13 exotics. A larger breadth of native species composition may indicate a wider range of sites. This could be due to more variability in community composition and environmental characteristics, and/or could reflect the larger plot size of the absence plots, allowing for more species to be included. Many of the native species unique to the absence plots were observed growing with Silene spaldingii during the survey at additional locations outside the macroplots. It may be that those species are less commonly found growing with Silene spaldingii. Percent cover Silene spaldingii was absent from sites with higher percent canopy cover of native annual forbs, total exotics species, exotic annual grasses, and exotic annual forbs (Table 4). Silene spaldingii presence was associated with higher percentages of litter. Percent cover for total canopy, native cover, native perennial forbs and grasses, native shrubs, exotic perennial forbs, basal plants, duff and embedded litter, moss, and soil were not different between presence and absence plots. Exotic annual forbs was the most predictive percent cover variable (χ 2 = , P < ). Species richness Silene spaldingii was absent from plots with higher total species richness due to a similar number of natives species but more exotics (Table 4). Silene spaldingii absence was associated with a higher count of exotic annual forbs, exotic annual grasses, and exotic perennial grasses species. There was no difference between presence and absence plots for species richness 16

26 of natives, native annual forbs, native perennial forbs, native shrubs and trees, and exotic perennial forbs. Lower elevations better predicted the presence of Silene spaldingii, while slope and aspect did not differ between presence and absence plots. Lower species richness of exotic annual forbs was the most predictive variable for Silene spaldingii presence (χ 2 = , P < ). Many of the single variable plant community characteristics were auto-correlated (r > 0.50, P < 0.05, Table 5). Species richness of exotic annual grasses was positively correlated with species richness of exotic annual forbs, and percent cover of native annual forbs, exotic species, and exotic annual forbs. Exotic annual forbs percent cover was positively correlated with percent cover of exotic species and exotic annual grasses, and species richness of exotic annual forbs, which was also correlated with percent cover of exotic species. Exotic perennial grasses species richness was positively correlated with elevation (m). The collinear variables were not put into models together. Multiple variable logistic regression did not yield any model with a larger loglikelihood chi-square than the model with single variable species richness of exotic annual forbs (Table 6). Three models had larger likelihood chi-squares than the single variable exotic annual forbs percent cover, but were not significantly better based on the chi-square improvement statistic (i.e. the improvement statistic of the model containing exotic perennial grasses (SR), litter (%), and total species (SR) over the single variable exotic annual forbs (%) was χ 2 = 1.766, df = 2, P = 0.414). Many models predicted the presence or absence of Silene spaldingii, but overall the best predictive model was still a single variable one. Exotic annual forbs was the most important plant functional group for explaining the presence or absence of Silene spaldingii at a site. Some of the most common in this study were 17

27 Alyssum alyssoides, Arnenaria serpyllifolia, Camelina microcarpa, Draba verna, Holosteum umbellatum, Lactuca serriola, Myosotis stricta, Sisymbrium altissimum, and Veronica arvensis. Invasions from exotic annual forbs are generally assumed to increase competition for light, space, nutrients, and water in the spaces between the understory of the grasses and shrubs that makeup the matrix structure of a community. In this manner, exotic annual forbs compete with native forbs for these niches. Exotic annual forb growth can be dense in the understory and choke out other species, acting much like the exotic annual grasses of the Bromus genus in the study area. Because Silene spaldingii has a long tap-root that enables it to exploit deeper soil water than annuals with shallower root systems, it is possible that competition from exotic annual forbs has the most effect on new Silene spaldingii recruits (Menke and Muir 2004). Annuals in general complete their life-cycle earlier in the season when some perennials are just germinating, causing light to become the limiting factor in spring (Dyer and Rice 1999). Annuals at a site can alter the availability of water by drying out the topmost layers of soil. The lack of soil surface moisture, along with reduced space and light, could retard the germination of Silene spaldingii seedlings, and affect survival in the first growing season (Menke and Muir 2004). Weeding of exotic annuals increased the rates of seedling establishment and survival to reproduction of a rare annual forb, Erodium macorphyllum (Gillespie and Allen 2004). The rare forb s differing growth rates were compared when planted with native perennial grasses versus exotic annuals, and not surprisingly the strongest suppression of the forb was exerted from the exotic annual pots (Gillespie and Allen 2004). Another form of competition between exotic species and natives is the amount of viable seed that is produced, and thus available for germination, establishment, and reproduction, 18

28 especially when many native species are considered seed-limited and have a reduced seedbank from prior invasion, site environmental change, and isolation (Turnbull et al and Seabloom et al. 2003). Silene spaldingii s dormancy, years as a rosette, and years as a vegetative stem are all years when it does not produce any seeds. Invasive annuals tend to produce large seed crops. Plots with Silene spaldingii had higher percent litter cover that may contribute to water infiltration. Litter may also reduce drying in the topmost soil layers by protecting it from heat. As litter decomposes it adds nutrients back to the site. Too much litter, particularly dense or deep litter layers, has been shown to reduce germination and seedling survival (Facelli and Pickett 1991; Golberg and Werner 1983; Lesica 1999). Even though plots with Silene spaldingii had more litter cover, the litter layers were not unusually deep. There is probably an optimum amount of litter that is beneficial to Silene spaldingii, and having too much or too little litter cover changes the site dynamics to some extent. Percent cover of exotic annual forbs is highly correlated with percent cover of exotic annual grasses, including the weeds Bromus tectorum and Bromus arvensis (Bromus japonicus) which have the ability to invade and replace native bunchgrasses. The majority of the sites with higher percent exotic annual forb cover also had a higher presence of exotic annual grasses (exotic annual grasses and forbs were not both included in logistic regression models due to high auto-correlation, Table 5). On the ground, the combination of these plants would produce an environment hostile to any species with a competitive disadvantage, and further exacerbate the drying, shading, and crowding of the understory. The change in conditions due to exotic annual invasions further facilitates other exotic species that have alternate phenologies to invade the altered community. Not only do these species not support Silene spaldingii, but they could 19

29 indicate that the site is moving toward an irreversible threshold of domination by exotic annual weeds (Stringham et al. 2003). The shift from a native perennial dominated community to one with higher annual composition represents a fundamental change in the structure and productive potential of a site. Late successional stages, or seres, are communities dominated by a few species which tend to return to the same composition following minor disturbances. Intact native perennial grasslands have been shown to resist the invasion of exotic species by maintaining their structure and function and remaining competitively resistant (Borman et al. 1991). Because Silene spaldingii absence was associated with higher percent cover of native annual forbs, the seemingly small shift toward more annuals, even native ones, may indicate a fundamental shift, or movement towards a shift, in the community structure. This shift could be considered a transition to another phase (community type) within a state, or group of related community types (Westoby et al. 1989; Bestelmeyer et al. 2009). It also suggests that Silene spaldingii is associated with mid- to late-seral, perennial dominated grassland and shrublands, containing numerous perennial forbs, with little annual presence. These communities represent high quality steppe grassland/shrubland habitat. The absence of Silene spaldingii was also associated with exotic perennial grasses. Bromus inermis (smooth brome) and Poa pratensis (Kentucky bluegrass) are exotic perennial grasses cultivated for CRP, pasture, and in lawns. Both are rhizomatous sod-forming grasses that can invade bunchgrass communities and displace native plants (USDA 2010). In invaded native communities the species can be dense and exclude not only forbs but also perennial bunchgrasses. Once established, the large root system of smooth brome, and rhizomatous habit of both grasses make it hard to remove these species from the native community (USDA 2010). 20

30 The exotic perennial grasses were strongly correlated with higher elevations, and higher elevations were associated with the absence of Silene spaldingii (Table 6). CRP fields of smooth brome are planted on the ridgetops at the Smoothing Iron Unit, at the highest elevation in the area. Smooth brome was invading from the CRP fields on the ridges into the native bunchgrass habitat below it. Even though higher total species richness was associated with the absence of Silene spaldingii, the majority of the difference in total species between absence and presence plots was due to the number of exotic species. The number of exotic species found with Silene spaldingii averaged 6.3, while sites without Silene spaldingii the number of exotic species averaged This accounts for almost all the difference (6.5) between total species richness at presence sites ( x = 34.8) and absence sites ( x = 41.3). Interestingly, the total species richness of natives was about the same on both types of sites, even though there were a larger number of exotics in the absence sites. This likely indicates that even though there is an exotic species component to those sites, overall they have not lost their native species and continue to maintain some resistance against shifting to an annual dominated system. While absence is most highly associated with higher cover and species richness of exotic annual forbs, these plants may be difficult to identify in the field. Some of these plants are tiny, many complete their lifecycle early in the season and then dry up, and others look like many other tiny plants. Year to year climatic variability highly influences the phenology of annuals and necessitates field identification of dead plant skeletons. These factors make using exotic annual forbs as indicator species difficult to an untrained eye. Native annual forbs present the same issues. Exotic annual grasses may be small and experience variability as well, but many people are familiar with these species (e.g., Bromus tectorum) and with minimal training could 21

31 identify the smaller number of species included in this grouping. Exotic perennial grasses are fewer as well, and most have easy to distinguish characteristics. Counting species richness of exotics could be a quick method if the observers know plant identification to the species level. Line point intercept (LPI) is a fast method of estimating cover, when compared to cover plots, but also requires species level differentiation. LPI has the added advantage of quantifying litter ground cover, which was a significant indicator of presence of Silene spaldingii. Silene spaldingii vigor in macroplots The number of rosettes/plant was negatively correlated with native perennial shrub cover and native perennial grass species richness (Table 7). Leaves/rosette was positively correlated with species richness for total species, natives, and native annual forbs. Stem height was positively correlated with native perennial grass cover, while fertilized flowers/plant was positively correlated with cover of total species and natives. The number of fertilized flowers/plant and capsules/plant were negatively correlated with moss cover and positively correlated with species richness of exotic annual grasses, as were the total reproductive parts/plant. Percent insect herbivory was positively correlated with northeast aspect and the percent of June plants absent in July was negatively correlated with exotic annual grass species richness. Finally, percent yellowing in June was negatively correlated with native perennial forb cover and percent withering in June was negatively correlated with litter, and duff and embedded litter cover. The negative relationship of rosettes/plant with percent cover of native perennial shrubs could be attributed the shade a shrub creates or the amount of resources it utilizes which a rosette plant has to compete for. Rosettes are unable to grow vertically to capture light. Similarly, the 22

32 negative correlation of rosettes/plant to the species richness of native perennial grasses could either indicate increased competition from the grasses, or increased shade, or that those sites have some other property that is more beneficial to grass species than the rosettes. The positive relationship of leaves/rosette with species richness of native annual forbs (and the correlated total species and native species richness) indicates rosettes do better (produce more leaves) when they grow in diverse communities with many native species, including annuals. These communities may have less competition for the Silene spaldingii plant because a community composition high in native annuals means less competition for resources later in the season, and less overlap in root depth. Percent cover of all species, natives, and native perennial grasses are all correlated with each other and have positive relationships with measures of reproductive vigor (stem height and fertilized flowers/plant). A community dominated by native perennial grass and forb cover likely represents the preferred habitat of Silene spaldingii, especially plants that are able to successfully reproduce. High percent cover of native perennial forbs, litter, and duff and embedded litter all are associated with a reduced chance of a plant yellowing or withering in June, a precursor to its chance of dying or being absent before flowering in July. The presence of litter may help maintain site moisture and reduce heat load during the growing season. Less clear is why moss cover is negatively associated with fertilized flowers/plant and capsules/plant. This deserves further inquiry. The positive relationship between plant reproductive potential and exotic annual grasses species richness is less intuitive than its association with native species richness. It is possible that the sites that have the most fecund plants (highest reproductive output) are the richest sites capable of supporting some exotic species, without significant resource competition. 23

33 This has been observed in other studies where an otherwise high quality native site with intact perennial grasses resists invasion and conversion to annual stands, even though it has exotic annuals present (Harris 1977; Borman et al. 1991). Therefore, the presence of these exotic annual grasses does not necessarily mean a degraded condition exists, especially since the total cover of these grasses was low (mean exotic annual grass cover in the macroplots was 5%). Alternatively, the sites that have greatest fecundity could represent macroplots that are starting to experience the effects of invasion, and the plants that are surviving are disproportionately the oldest, most established plants. These mature plants should be more capable of competing and persisting in degraded conditions longer than rosettes or younger/smaller stemmed plants, and may persist for years due to Silene spaldingii s long lifespan. The total number of plants/macroplot and percent dead in July were positively correlated with the percent absent in July (Table 8). Stem height, percent yellowing in June, flowers and buds/plant, and total reproductive parts/plant were all negatively correlated with percent absent in July. Rosette plants were negatively correlated with stem plants and percent stems withering in June, but positively correlated with percent dead in July. Stem height, number of flowers and buds/plant, and total reproductive parts/plant were all positively correlated with each other. Flowers and buds/plant and total reproductive parts/plant were positively correlated with the number of stems/plant. Percent insect herbivory was negatively correlated with stemmed plants and stems/plant. The number of leaves/rosette was negatively associated with percent withering, rosettes withering, and stems withering in June. Percent rosettes withering in June was positively correlated with percent plants withering in June as well as percent dead in July. Finally, percent stems withering in June was positively correlated with withering in June and number of stem plants. 24

34 The relationships amongst plant characteristics are not surprising. The pattern of flowers and buds/plant and total reproductive parts/plant being positively correlated with the number of stems/plant and stem height has been shown in previous studies (Menke and Muir 2004). It is possible that fertilized flowers/plant and capsules/plant were not significantly correlated with the same stem characteristics because it was too early in the season for the majority of the flowers to be fertilized and progress into capsules. Previous work also shows that rosette plants and some stems tend to disappear by the time stemmed plants are flowering midsummer (Gray and Hill 2006). Our results corroborate this and show that rosettes made up a majority of the percent dead in July, especially rosettes that started withering in June, but also that the greater the number of leaves a rosette had the less likely it is to wither in June. This suggests that the greater the number of leaves a rosette has the longer it will maintain above ground growth. This means it may be able to produce larger roots to capture more water and store more energy for future flowering years. Similarly, shorter stem plants were more likely to be absent in July and plants with fewer stems/plant had more insect herbivory. These characteristics may affect whether or not a plant is strong enough to produce flowers and ultimately seeds. The percent of insect herbivory on all plants was higher in pasture 5 than pastures 3 and 4 (Table 9). The percent of insect herbivory on stemmed plants was higher in pasture 5 than all other pastures. More than 90% of plants had insect herbivory in pasture 5. The aspects of all the macroplots in pasture 5 were the farthest east, ranging from Insect herbivory was positively correlated with aspect (Table 7). The percent of all plants yellowing in June was lower in pasture 3 than all other pastures. Similarly, the percent of stemmed plants yellowing in June was lower in pasture 3 than in pastures 4 and 5. The percent of non-reproductive stems was 25

35 higher in pasture 5 than 3 and 4. It is not clear why pasture 5 had more insect herbivory and more non-reproductive stems. Pasture 5 sites were drier than macroplots on more northerly aspects, and had steep slopes. Also, it is not clear why pasture 3 had less yellowing plants in June. It is possible a combination of environmental characteristics contributed to these differences, but none were significant when correlated, except for aspect and insect herbivory. The rest of the Silene spaldingii plant vigor variables were not significantly different by pasture. Elk and mule deer inhabit the area and were seen in all pastures in the summers of 2007, 2008, and Cow, deer, and elk scats were present in every plot in pasture 2, which was grazed by cattle in 2007 and 2009 (Table 10). Pasture 5 was similar, minus deer pellets in one macroplot. Pasture 3 had two macroplots with cattle pats, two with elk pellets, and none with deer pellets. Pasture 4 had no cattle pats, and only one macroplot with scat of elk or deer, and it had both. Pasture 2 was grazed by cattle in 2009 before plants were found and monitoring plots begun. Pasture 5 was grazed by cattle in 2007 and 2008, and pasture 3 in Pasture 4 has not had cattle grazing for more than 6 years. The presence of elk and deer pellets suggests use and choice of the same areas where Silene spaldingii grows. Cattle pats in the macroplots are a result of intentional grazing of those pastures. Timing of monitoring The comparison of June to July total plant counts was significant with a mean difference of 6.5 more plants found in June than July (t = 3.90, P = , two-tailed; Table 11). Therefore, monitoring needs to be conducted in June (or comparable phenologocial stage) to find all the plants growing that season. But, monitoring only in June would miss the later flowering. To monitor for population trends and reproductive success, it is necessary to monitor twice a 26

36 season, or it may be possible to adjust July counts by an estimate from June. Monitoring twice annually would account for patterns in dormancy and accurately assess new seedling recruits and the age of newly established plants after 4 years (Gray and Hill 2006). CONCLUSION The large population of Silene spaldingii at the Smoothing Iron Unit is an indication of the high quality native grassland and shrubland that occupied the area. Habitat in neighboring canyons of similar community composition, aspects, and slopes may yield undiscovered populations. The presence of exotic plants in the area represents a potential threat, and managing weed infestations should be of highest priority for protecting the Silene spaldingii population. Monitoring for invasion of exotic plants into native communities is an important step toward protecting the population. Increases in exotic annuals may indicate a trend in the integrity of the community. Yearly monitoring of plant vigor characteristics would provide more concrete evidence for relationships to community composition. Results from one year of data are useful, but results could vary in the future due to yearly variation in growing conditions. Growing conditions in 2009 were considered favorable. Although cattle grazing occurred during the study, one year of data do not provide evidence to draw any conclusions about grazing and Silene spaldingii. Long-term monitoring of patterns in population, community, and vigor characteristics could help us understand the ecology of populations of Silene spaldingii. 27

37 Table 1. Silene spaldingii habitat types (Washington Natural Heritage Program 2009) and ecological sites (USDA, NRCS 2004) within project boundaries. Habitat Types Ecological Site Smoothing Iron Pintler Creek Festuca idahoensis Symphoricarpos albus Cool Loamy 15+ PZ, Cool Loamy 9-15 PZ Yes Possibly Festuca idahoensis Rosa spp. Cool Loamy 15+ PZ, Cool Loamy 9-15 PZ Yes Possibly Festuca idahoensis Koeleria macrantha Cool Loamy 15+ PZ, Cool Loamy 9-15 PZ Yes Possibly Festuca idahoensis Pseudoroegneria spicata Cool Loamy 15+ PZ, Loamy 15+ PZ, Cool Loamy 9-15 PZ Yes Yes Pinus ponderosa Festuca idahoensis Cool Loamy 15+ PZ Yes No Pinus ponderosa Symphoricarpos albus Cool Loamy 15+ PZ Yes No 28

38 Table 2. Silene spaldingii population counts and estimates by pasture at the Smoothing Iron Unit. The proportion of plants missed during survey/plants found in macroplots is listed in parentheses. The equation for population estimates = (# Plants * proportion) + # Plants. Pasture # Plants Low Median High Mean (0.4) 762 (2.92) 2,134 (3.3) 2,339 (2.21) 1, (0.84) 1,755 (1.24) 2,137 (2.2) 3,053 (1.43) 2, ,576 (0) 3,576 (0.14) 4,064 (0.83) 6,556 (0.32) 4, (0.47) 1,182 (0.5) 1,208 (1.5) 2,013 (0.82) 1,467 Total 5,879 7,275 9,542 13,961 10,259 29

39 Table 3. Potential suitable habitat and Silene spaldingii density estimates per pasture at the Smoothing Iron Unit based on aspect and slope. Pasture Ha Plants/ha Total

40 Table 4. Separate logistic regression tests of site variables and Silene spaldingii presence (n = 12) and absence (n = 10) in plots. Canopy and ground cover percent estimates based on the LPI method. Slope, elevation, and aspect values derived from ArcGIS. Bold values are significant at P < Presence plots Absence plots Variable Mean SD Mean SD χ 2 P Canopy cover (%) Total species Natives Native annual forbs Native perennial forbs Native perennial grasses Native shrubs Exotics <0.001 Exotic annual forbs < Exotic annual grasses Exotic perennial forbs Exotic perennial grasses N/A b, c Ground cover (%) Basal plants Litter <0.01 Duff and embedded litter Moss Soil Species Richness (SR) Total species Natives Native annual forbs Native perennial forbs Native perennial grasses N/A b Native shrubs and trees Exotics N/A b Exotic annual forbs < Exotic annual grasses Exotic perennial forbs Exotic perennial grasses Environment Slope ( ) Elevation (m) 1, , Aspect ( ) a a Aspect was treated as a continuous variable past 360 (i.e. 20 = 380 ). b The validity of the model was questionable, or quasi-complete separation of points detected. Errors in SAS. 31

41 Table 5. Correlation of community variables amongst themselves used to assess collinearity. Pearson correlation coefficients are listed in the matrix. Bold r values are significant at the level indicated by the superscript letter. Only significant variables with r > 0.50 are shown. Canopy cover variables in the table are followed by (%) and species richness variables are followed by (SR). Exotic species (%) Exotic annual forbs (%) Exotic annual forbs (SR) Exotic annual grasses (SR) Exotic perennial grasses (SR) Native annual forbs (%) b Exotic species (%) c c Exotic annual forbs (%) c c Exotic annual grasses (%) c c a Litter (%) Total species (SR) a Exotic annual forbs (SR) c c a Exotic annual grasses (SR) a Exotic perennial grasses (SR) Elevation (m) b a P < 0.01, b P < 0.001, c P <

42 Table 6. Multiple logistic regression models for Silene spaldingii presence (n = 12) and absence (n = 10) in plots. Only significant single variables were used in the models (Table 4). Only the combinations with a maximum likelihood and complete separation of data points, where each variable in the model had a Wald statistic P < 0.10, are shown. Single variable models are displayed to show how they compare to the multiple variable models. Models are ordered by likelihood ratio (χ 2 ). Variable 1 Variable 2 Variable 3 χ 2 P Wald P a Exotic annual forbs (SR) < <0.05 Exotic perennial grasses (SR) Litter (%) Total species (SR) <0.001 <0.10 Exotic annual grasses (SR) Exotic perennial grasses (SR) < <0.05 Exotic species (%) Litter (%) < <0.10 Exotic annual forbs (%) < <0.01 Exotic annual grasses (SR) Litter (%) <0.001 <0.05 Native annual forbs (%) Litter (%) <0.001 <0.10 Exotic species (%) Exotic perennial grasses (SR) <0.10 Litter (%) Total species (SR) <0.05 Exotic annual grasses (%) Litter (%) <0.05 Native annual forbs (%) Exotic annual grasses (%) <0.10 Exotic annual grasses (%) Elevation (m) <0.10 Exotic perennial grasses (SR) Total species (SR) <0.10 Exotic annual grasses (%) Exotic perennial grasses (SR) <0.10 Litter (%) Elevation (m) <0.10 Exotic species (%) <0.001 <0.05 Exotic annual grasses (SR) Total species (SR) <0.10 Exotic perennial grasses (SR) Litter (%) <0.10 Native annual forbs (%) <0.05 Exotic annual grasses (SR) <0.05 Exotic annual grasses (%) <0.05 Exotic perennial grasses (SR) <0.05 Litter (%) <0.01 <0.05 Total species (SR) <0.05 Elevation (m) <0.10 a Wald statistic for each variable was equal to or less than the listed P-value; the greatest P-value is shown. 33

43 Table 7. Correlation of significant community variables, shown with Pearson correlation coefficients, with Silene spaldingii plant characteristics from macroplot data. Bold r values are significant at the level indicated by the superscript letter. Rosettes/ plant Lvs./ rosette Stem height (cm) % Insect herb. % Yellow. in June % With. in June % Absent in July Fert. flowers/ plant Caps./ plant Total repro. parts/ plant Total species (%) a Natives (%) a Native perennial shrubs (%) Native perennial grasses (%) Native perennial forbs (%) a b b Litter (%) a Duff and embedded litter (%) a Moss (%) a a Total species (SR) a Natives (SR) a Native perennial grasses (SR) Native annual forbs (SR) Exotic annual grasses (SR) a b a a a a Aspect ( ) a a P < 0.05, b P < 0.01, c P <

44 Table 8. Correlation of significant Silene spaldingii plant characteristics amongst themselves as Pearson correlation coefficients. Bold r values are significant at the level indicated by the superscript letter. Stem plants Stems/ plant Leaves/ rosette Stem height (cm) % Ros. with. in June % Stems with. in June % Absent in July % Dead in July Flowers and buds/ plant Total plants a Rosette plants c a a Stem height (cm) a a % Yellowing in June % Insect herbivory % With. in June % Rosettes with. in June % Stems with. in June a a a a b c b a a a % Dead in July a Flowers and buds/plant b b Total repro. parts/ plant a a b b a P < 0.05, b P < 0.01, c P <

45 Table 9. Silene spaldingii plant characteristics in macroplots by pasture. F and P-values from pasture effect of twoway ANOVA (plot effect not shown and not significant for any characteristic). Different superscript letters following means are significantly different (post-hoc LSD, P < 0.05). Pasture Mean SD Mean SD Mean SD Mean SD F P Total plants Stem plants Rosette plants Mixed plants % Stem plants % Rosette plants % Mixed plants Stems/plant Stem height (cm) Rosettes/plant Leaves/rosette Rosette leaf length (cm) % Animal herbivory N/A 1 % Insect herbivory ab a a b % Stem insect herbivory a a a b % Rosette insect herbivory % Yellowing in June a b a a % Stems yellowing in June ab a b b % Rosettes yellowing in June % Withering in June % Stems withering in June % Rosettes withering in June % Absent in late July % Dead in July Flowers and buds/ plant Fertilized flowers/ plant Capsules/plant Total reproductive parts/plant % Non-reproductive stems 5.48 ab a a b Non-normal and not corrected by transformation. 2 Natural log transformation used to satisfy normality. 36

46 Table 10. Presence of ungulate scat in macroplots. Age of scat not distinguished (i.e. present or past season). Years grazed by cattle Pasture Site Cow Mule deer Elk 2 1 yes yes yes 2007, yes yes yes 2 3 yes yes yes 3 1 yes no no yes no yes 3 3 no no yes 4 1 no no no none 4 2 no yes yes 4 3 no no no 5 1 yes no yes 2007, yes yes yes 5 3 yes yes yes 37

47 Table 11. Plants present in June, but absent in July. T and P-value from paired t-test, two-tailed. Pasture Plants absent in July SD Mean across pastures t = 3.90 P =

48 Figure 1. Inland Pacific Northwest distribution of known Silene spaldingii populations as of 2007 (adapted from the Recovery Plan for Silene spaldingii, USFWS, 2007). 39

49 Figure 2. Map of study areas southwest of Asotin, WA. Pasture boundaries shown with black lines. 40

50 T 0 T 1 T 2 T 3 T 4 T 5 T 6 T 7 T 8 T 9 Downslope 1 m Microplot frame 0 m 0 m Baseline Figure 3. 10m x 10m Macroplot layout with 0.5m x 1.0m microplots nested within (modified from Lichthardt and Gray 2003). 41

51 Figure 4. Map of potential suitable habitat based on aspect and slope at the Smoothing Iron Unit. 42