Small Mammal associations with habitat structure in riparian zones of a managed ICH Forest; Revelstoke, B.C. by ISABELLE HOUDE B.Sc, McGill University, 1990 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Forestry) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April 1997 ©Isabelle Houde, 1997 In p resent ing this ; thesis in partial fu l f i lment of the requ i remen ts fo r an advanced d e g r e e at the Universi ty of British C o l u m b i a , I agree that t h e Library shall make it f reely available fo r re ference and study. I fu r ther agree that permiss ion f o r ex tens ive c o p y i n g ' of this thesis f o r ' scholar ly pu rposes may be g ran ted by the head o f m y d e p a r t m e n t or by his o r her representat ives. It is u n d e r s t o o d that c o p y i n g o r pub l i ca t i on o f this thesis fo r f inancial ga in shall n o t be a l l o w e d w i t h o u t m y w r i t t e n permiss ion . D e p a r t m e n t o f ^f'fyr'Ckl The Univers i ty o f British C o l u m b i a Vancouver , Canada Date /ft? DEr6 (2/88) ABSTRACT I investigated small mammal communities and habitat associations within riparian and upland areas of a class S2 stream in Revelstoke, B.C. Sampling sites were distributed among three types of habitat: old-growth ICH forests, 5-10-year-old clearcut, and clearcuts with 60 to 90m riparian buffers. I assessed small mammal populations and habitat associations along a transriparian gradient starting flush with the stream and going upland perpendicular to the stream (300m). Six small mammal species or genera were captured. Deer mice (Peromyscus maniculatus), western jumping mice (Zapus princeps), voles (Microtus spp.j, shrews (Sorex spp), red-backed voles (Clethrionomys gapperi), and short tailed weasels (Mustela erminea) all occupied clearcut sites. M. erminea was not captured within the riparian buffer strips areas, nor within the old-growth habitat. Small mammal species richness varied with sampling-site but was generally higher within the clearcut and buffered habitats where higher percent cover of shrubs and volume of downed wood were found. The relative abundance of small mammals differed among sampling-sites. Small mammal density and abundance were higher within the heterogeneous buffered habitat (composed of both clearcut and riparian buffer zones) followed by the clearcut and old-growth habitats. Herbs (including ferns), species richness and vegetative percent cover differed among sampling-sites. There was no distance effect on vegetation species richness. Percent cover of shrubs differed significantly with distance from stream; Percent cover of herbs was greater ii within old-growth and riparian buffer strip areas than in clearcut zones. Sampling-sites contained a wide range of microhabitats, so it was not possible to discern a continuous gradient or trend in habitat structure. There was no distance effect on the pooled small mammal populations. P. maniculalus dominated all habitats. Sorex were most abundant within clearcut areas and were positively correlated with distance from the stream in the buffered habitat. Microtus spp. and Z. princeps were most abundant within old-growth and riparian buffer strips. Z. princeps capture rates were negatively correlated with distance from stream in clearcuts and old-growth habitats. C. gapperi was found only in riparian buffer strip zones when these were present within a clearcut. My results suggest that heterogeneous habitat (clearcut with riparian buffer) increases the overall small mammal abundance and diversity. Presence of riparian buffer strips allow the density of old-growth obligate species to remain high. From my results it was not possible to suggest an ideal riparian buffer width to contain quality habitats for small mammal communities. Buffers of 60-90m wide appeared adequate to retain species diversity. Because small mammal communities seemed related to the microhabitat rather than the habitat at a larger scale, an investigation of the microhabitat associations would provide better understanding of small mammal population parameters. iii TABLE OF CONTENTS ABSTRACT " TABLE OF CONTENTS iv LIST OF FIGURES vi LIST OF TABLES viii ACKNOWLEDGEMENTS ix 1. GENERAL INTRODUCTION 1 1.1 STUDY SITE 3 1.2 METHODS 4 1.2.1 Sampling sites 4 1.2.2 Small mammal live trapping 5 1.2.3 Vegetation and habitat sampling 7 2. SPECIES COMPOSITION, ABUNDANCE, AND HABITAT ASSOCIATIONS .". 11 2.1 rNTRODUCTION 11 2.2 METHODS 14 2.2.1 Vegetation and downed wood 14 2.2.2 Small mammal populations 16 2.2.3 Population density estimates 19 2.3 RESULTS 22 2.3.1 Vegetation and downed wood 22 2.3.2 Small mammal populations 31 2.3.3 Population density estimates 32 2.4 DISCUSSION 38 iv 3. TRANSHIPARIAN GRADIENT AND HABITAT ASSOCIATIONS 41 3.1 INTRODUCTION 41 3.2 METHODS 43 3.3 RESULTS 46 3.3.1 Vegetation 46 3.3.2 Small mammal populations 53 3.4 DISCUSSION 64 4. CONCLUSIONS AND RECOMMENDATIONS 68 LITERATURE CITED '. 70 v LIST OF FIGURES FIGURE IPOSITION OF SAMPLING GRIDS IN RELATION TO THE EDGES OF THE ADJACENT HABITAT (NOT TO SCALE). GRIDS WERE AT LEAST 400M APART 6 FIGURE 2 TRANSECT LAYOUT IN EACH GRID FOR THE SAMPLING YEARS 1994 AND 1995 6 FIGURE 3 VEGETATION SAMPLING PLOT DESIGN 9 FIGURE 4 DOWNED WOOD LINE-INTERCEPT SAMPLING ON 3 LINES OF 1 1.2M LONG 10 FIGURE 5 TOTAL VEGETATION SPECIES RICHNESS RECCORDED IN ALL SAMPLING-SITES. OG=OLD-GROWTH, BUF=BUFFERED HABITAT ( B U F S T P + B U F C C ) , BUFSTP=BUFFER STRIP, B U F C C = C L E A R C U T ADJACENT TO RIPARIAN BUFFER, AND CC=CLEARCUT 24 FIGURE 6 RELATIVE FREQUENCY OF THE VEGETATION COMPOSITION IN EACH HABITAT: OLD-GROWTH (FIG.6A), BUFFER STRIP (FIG.6B), CLEARCUT ADJACENT TO RIPARIAN BUFFER (FLG.6C), CLEARCUT (FIG.6D). SEE TABLE 1 AND 2 FOR ABBREVIATIONS 25 FIGURE 7 VEGETATIVE PERCENT COVER PER PLOT WITHIN EACH SAMPLING SITE. T H E AVERAGE PERCENT COVER PER HABITAT IS ALSO GIVEN: SHRUBS (FIG.7A), HERBS (INCLUDING FERNS; FIG.7B). OG=OLD-GROWTH, BUF=BUFFERED HABITAT, BUFSTP=BUFFER STRIP, B U F C O C L E A R C U T ADJACENT TO RIPARIAN BUFFER, CC=CLEARCUT 28 FIGURE 8 SMALL MAMMAL RELATIVE ABUNDANCE'. NUMBER OF ANIMALS CAPTURED PER TRAP: 1994 (FIG.8A), 1995 (FIG. 8B). T H E AVERAGE PER HABITAT IS GIVEN. OG=OLD-GROWTH, BUF=BUFFERED HABITAT, BUFSTP=BUFFER STRIP, BUFCC=CLEARCUT ADJACENT TO RIPARIAN BUFFER, CC=CLEARCUT 34 FIGURE 9 JOLLY-SEBER POPULATION DENSITY ESTIMATES OF SMALL MAMMALS OVER TIME FROM 1995 CAPTURES: OLD-GROWTH HABITAT (FIG. 9A), CLEARCUT HABITAT (FIG.9B), .BUFFERED HABITAT (FIG.9C AND D) 35 FIGURE 10 AVERAGE POPULATION DENSITY OF SMALL MAMMAL PER HA CALCULATED WITH THE JOLLY-SEBER MODEL FROM 1995 CAPTURES: AVERAGE DENSITY PER SAMPLING SITE (FIG. 10A), AVERAGE DENSITY PER HABITAT TYPE (FLG. 10B). OG=OLD-GROWTH, BUF=BUFFERED HABITAT, BUFSTP=RIPARIAN BUFFER STRIP, BUFCC=CLEARCUT ADJACENT TO RIPARIAN BUFFER, C O C L E A R C U T 36 VI FIGURE 11 VEGETATION SPECIES RICHNESS AND DISTANCE FROM STREAM: OLD-GROWTH-SITE 1(FIG. 1 1A), OLD-GROWTH SITE 2 (FIG. 11B), BUFFERED SITE 1 (FIG. 11c), BUFFERED SITE 2 (FIG. 11D), CLEARCUT SITE 1 (FIG. H E ) , CLEARCUT SITE 2 (FIG. 11F) 48 FIGURE 12 PERCENT VEGETATIVE COVER PER PLOT WITHIN THE OLD-GROWTH SAMPLING SITES AS A FUNCTION OF THE DISTANCE FROM STREAM: PERCENT SHRUB COVER (FLG. 12A), PERCENT HERB COVER (FLG. 12B) 50 FIGURE 13 PERCENT VEGETATIVE COVER PER PLOT WITHIN THE BUFFERED SAMPLING SITES AS A FUNCTION OF THE DISTANCE FROM STREAM: PERCENT SHRUB COVER (FlG. 13A), PERCENT HERB COVER (FlG. 13B) 51 FIGURE 14 PERCENT VEGETATIVE COVER PER PLOT WITHIN THE CLEARCUT SAMPLING SITES AS A FUNCTION OF THE DISTANCE FROM STREAM! PERCENT SHRUB COVER (FlG. 14A), PERCENT HERB COVER (FlG. 14B) 52 FIGURE 15 SMALL MAMMAL SPECIES DISTRIBUTION AND COMPOSITION IN RELATON TO THE DISTANCE FROM STREAM. CAPTURE DATA FROM 1994 55 FIGURE 16 SMALL MAMMAL SPECIES DISTRIBUTION AND COMPOSITION IN RELATION TO THE DISTANCE FROM STREAM. CAPTURE DATA FROM 1995 56 FIGURE 17 SMALL MAMMAL RELATIVE ABUNDANCE IN 1994 AND DISTANCE FROM THE STREAM IN EACH SAMPLING SITE: OLD-GROWTH HABITAT (FIG. 17A), BUFFERED HABITAT (FIG. 17B), CLEARCUT HABITAT (FIG. 17c)... 57 FIGURE 18 SMALL MAMMAL RELATIVE ABUNDANCE IN 1995 AND DISTANCE FROM THE STREAM IN EACH SAMPLING SITE: OLD-GROWTH HABITAT (FIG. 18A), BUFFERED HABITAT (FIG. 18B), CLEARCUT HABITAT (FIG. 18C)... 58 FIGURE 19 NORMALIZED ABUNDANCE OF SMALL MAMMAL SPECIES AMONG HABITAT WITH 1994 CAPTURES: OLD-GROWTH (FIG. 19A), RIPARIAN BUFFER (FIG. 19B), BUFFERED CLEARCUT (FIG. 19C), CLEARCUT (FIG. 19D). P. MANICULATUS WAS EXCLUDED BECAUSE ITS ABUNDANCE WAS TOO HIGH IN COMPARISON WITH THE OTHER SPECIES THEREFORWE HIDING THE PATTERNS OF ABUNDANCE FREQUENCIES 60 FIGURE 20 NORMALIZED ABUNDANCE OF SMALL MAMMAL SPECIES AMONG HABITAT WITH 1995 CAPTURES: OLD-GROWTH (FIG.20A), RIPARIAN BUFFER STRIP (FIG. 20B), BUFFERED CLEARCUT (FIG. 20c), CLEARCUT (FIG. 20D) 61 FIGURE 21 SMALL MAMMAL CAPTURE RATES AND DISTANCE FROM STREAM WITHIN EACH HABITAT: Z. PRINCEPS (FIG.21 A), MICROTUSSP?. (FIG.21B), SOREXSPP. (FIG 21c), C. GAPPERI (FIG. 2ID ) 62 vii LIST OF TABLES TABLE 1. GROUPING OF THE SHRUB SPECIES RECORDED ACCORDING TO THE SOIL MOISTURE, NITROGEN LEVEL, AND FAMILY. GROUPS ARE THE VARIABLES IN ANALYSIS. T H E SOIL MOISTURE AND NITROGEN LEVEL ARE INDICATED WITHIN THE COLUMN "SOIL" WHERE: N=NITROGEN RICH, N=NITROGEN POOR, M=NITROGEN MEDIUM, 1=WATER SHEDDING SITE, 2=FLOOD PLAIN AND WATER RECEIVING SITE, 3=VERY MOIST TO WET, 4=FRESH TO VERY MOIST, 5=FRESH TO MOIST, 6=DRY TO MIST, 7=VERY DRY TO MODERATELY DRY, 8=DISTURBED SITE 17 TABLE 2. GROUPING OF THE HERB SPECIES RECORDED ACCORDING TO THE SOIL MOISTURE, NITROGEN LEVEL, AND FAMILY. GROUPS ARE THE VARIABLES USED FOR DATA ANALYSIS. SEE EXPLANATION ON COLUMN "SOIL" IN DESCRIPTION OF TABLE l 18 TABLE 3 SHRUB GROUPS AS DEFINED IN TABLE 1 AND 2 THAT WERE SIGNIFICANTLY DIFFERENT (UNIVARIATE A N O V A ) . T H E AVERAGE PERCENT COVER PER PLOT IN EACH SAMPLING SITE IS GIVEN AS WELL AS THE P VALUE CALCULATED BETWEEN HABITATS AND BETWEEN SAMPLING-SITES. BUF=RIPARIAN BUFFER STRIPS AND ADJACENT CLEARCUTS 29 TABLE 4 HERB GROUPS AS DEFINED IN TABLE 1 AND 2 THAT WERE SIGNIFICANTLY DIFFERENT (UNIVARIATE A N O V A ) . T H E AVERAGE PERCENT COVER PER PLOT IN EACH SAMPLING SITE IS GIVEN AS WELL AS THE P VALUE CALCULATED BETWEEN HABITATS AND BETWEEN SAMPLING-SITES. BUF=RIPARIAN BUFFER STRIPS AND CLEARCUTS 30 TABLE 5 JOLLY TRAPPABILITY CALCULATED FOR EACH SAMPLING SITE 37 TABLE 6 WEEKLY JOLLY-SEBER POPULATION DENSITIES CALCULATED FOR EACH SAMPLING SITE 37 TABLE 7 SIGNIFICANCE OF EACH REGRESSION SLOPE FOR SMALL MAMMAL CAPTURE RATES AND DISTANCE FROM STREAM. P VALUES AND COEFFICIENTS OF DETERMINATION ARE GIVEN 63 viii ACKNOWLEDGMENTS I thank Dr. M.D. Pitt and Dr. J. Richardson, members of my committee, and Dr. C. Galindo-Leal for their help and advice on my project. Especially, I thank Dr. F. L. Bunnell without who the study could not have been completed and who provided with guidance, assistance, and financial support. I am grateful to Dr. Bruce McLellan who first believed in me and supported my candidacy as a graduate student at the University of British Columbia. Dr. A. Kozak provided statistical advice. The field work was made easier with the encouragement, help, and moral support of Suzan Hall as well as with the hard work and determination of my research assistants: Saskia and Hooban. Thank you to Glenn Sutherland, Ralph Wells, Jean Houde, Serge Houde, Michael Jackson, Dawn Brett, and Stephanie Hunter for providing me with precious computer materials, technical consultation, and encouragement. Special acknowledgment to Jackie Johnson, a very competent, efficient, and patient person who was the greatest help for logistic and administrative problems. So many thanks, hugs and kisses to my family and friends for their encouragement, especially to Dominique, Denyse, Lise, and Samantha for listening to my complaints and feeding me when I was starving. ix 1. GENERAL INTRODUCTION Riparian areas, as denned by the B.C. Ministry of Forest and B.C. Environment (1995b) on page 1 of the Riparian Management Area Guidebook, are "the areas that occur next to the banks of streams, lake and wetlands and include both the area dominated by continuous high moisture content and the adjacent upland vegetation that exert an influence on it. " Riparian areas promote patchiness and structural complexity creating high diversity and good quality habitat (McComb and Hagar, 1993). Furthermore Bunnell and Dupuis (1995) reported that almost 75% of the vertebrate species found in British Columbia use riparian habitat. Riparian zones are in the productive portions of forest in British Columbia where 75% of the watersheds have been partially developed. Some biogeoclimatic zones have no undeveloped watersheds larger than 50 km2 left (Bunnell and Dupuis, 1995). Because riparian areas are important to both the forest industry and wildlife communities, we require a good understanding of riparian ecosystem structure and function. Riparian zones are believed to serve different purposes, including habitat connectivity and patchiness. Their long and narrow features may provide natural routes for wildlife movements and serve as transition zones between the aquatic habitat and the adjacent uplands. They also increase the amount of edge habitat available (McComb and Hagar, 1993) contributing to biodiversity. Bunnell and Dupuis (1995) suggested that riparian zones could be important reservoirs of biodiversity because many species are more abundant and reproductively successful in these areas than in adjacent l uplands. We do not presently know how riparian areas should be managed to maintain all their valued habitat features. To maintain habitat characteristics associated with riparian zones, different widths of buffers strips have been left in managed landscapes. In this study the small mammal communities associated with riparian areas and the adjacent uplands in a managed forest were investigated for a class S2 stream as defined by the Fish-Stream Identification Guidebook (B.C. Ministry of Forests and B.C. Environment, 1995a). The upland communities were considered because some upland species may depend on riparian habitats for part of their life cycle (Kauffman 1988, Gomez 1992, Gyug and Vaartnou 1996). It has been suggested that the integrity of the riparian ecosystem may also depend on and be influenced by the adjacent upland area. (B.C. Ministry of Forests and B.C. Environment, 1995b ). My objectives were: 1) to assess the associations between small mammal communities and selected habitat structures within ICH forest in both riparian and uplands 2) to test for effect of riparian buffer widths on small mammal species richness and abundance and 3) to determine the impact and the role of the transriparian gradient on small mammal population abundance and richness in managed ICH forest. Chapter 1 describes the study area and sampling design. Chapter 2 discusses associations found between the small mammal community and the different types of habitats studied. Chapter 3 discusses relationships and correlations between small mammal populations and habitat structure as a function of the distance from the edge of the stream. 2 1.1 STUDY SITE The study site is situated approximately 75 km north of Revelstoke in southeastern British Columbia. The riparian zones of Downie Creek drainage are included in the Revelstoke TSA (Timber Supply Area). Headwaters of Downie Creek originate in the Selkirk Mountain glaciers and flow into the Revelstoke reservoir. The bottomland area runs for a length of 20 km and is on average 400 m wide. Downie creek is classified as an S2 stream according to the Fish-Stream Identification guidebook (B.C. Ministry of Forests and B.C. Environment, 1995a). An S2 steam is defined as a fish bearing stream from 5 to 20m wide with an average gradient less than 20%. The study area is classified as Wet Interior Cedar-Hemlock (ICHwkl and ICHvkl). All of the sampling sites were selected from the ICHwkl portion of the drainage, in the Wells Grey wet cool interior Cedar-Hemlock variant (B.C. Ministry of Forest, 1992). Within the ICH forest the most common habitat types are riparian forest, upslope forest, avalanche chutes and clearcuts. Among the forested biogeoclimatic zones in the province, vertebrates in the ICH zones show strong association with riparian areas, second only to the CWH zones (Bunnell et ai, 1991). Elevation varies from 400 to 1400 m with a mean annual precipitation of 1214 mm and 121 frost-free days. The riparian bottomlands are made up of a mosaic of habitat types providing a great edge effect. The area contains a rich diversity of flora and fauna supporting an old-growth dependent wildlife community (B.C. Ministry of Forest, 1992). Wolverine have been sighted in the valley as well as many other smaller carnivores and rodents including marten and beaver (Corbett, 1989). 3 The forest is dominated by red cedar (Thuja plicata) and western hemlock (Tsuga heterophylla) >250 year old. Most of the timber removal in the area happened between 1987 and 1991, early serai stands both with and without riparian buffers were left. The remaining riparian buffer strips vary from 30 m to 90 m wide. Following these logging operations, burning and tree planting was prescribed. No other silvicultural treatments were applied within the sample plots. S 1.2 METHODS 1.2.1 Sampling sites Trapping grids of similar size were installed at all of the sampling sites immediately adjacent to the edge of the stream extending upland. The grids had the same width (parallel to the stream) but varied in length depending on the pattern of forest cover. The sampling sites were distributed among 3 types of habitat: undisturbed old-growth, 5 to 10-year-old clearcut, and clearcut with old-growth riparian buffer (60m-90m wide). Each habitat was sampled with 2 grids for a total of 6 sampling sites. Every clearcut area, with or without a riparian buffer, was bordered by patches of old-growth forest on either side. As shown in Figure 1 the grids were positioned to minimise potential edge effects. A distance of at least 75m was always left between the grid and adjacent habitat. Grids were always set apart by a distance of at least 400 m. I sampled during the spring and summer season of 1994 and 1995. The number of transects per grid and their average length was different between the sampling periods of 1994 and 1995. In 1994 the sampling grids were composed of 5 transriparian transects going from the 4 immediate edge of the stream to 150m upslope. These transects were 15m apart. In the second year of data collection, an increase in resources allowed me to extend the transects from 150m up to 300m. Although the transects were longer (Fig.2), their number was reduced to only 3 transects per grids separated by a distance of 30m. These changes were made to attain a better assessment of the small mammal population and habitat structure in the adjacent upland zones of the riparian habitat. 1.2.2 Small mammal live trapping I estimated small mammal relative abundance and species composition with Longworth live traps. The traps were set 15m apart along the transriparian transects with the first trapping station flush with the edge of the stream. Sites were trappe 5 Old growth habitat Buffered habitat _1 Clearcut habitat >400m Figure 1. Position of sampling grids in relation to the edges of the adjacent habitat (not to scale). Grids were at least 400 m apart. V 15m 60 m 150 m 240 m - 300 m 30 m 60 m Transect Sa^etrt 1384 Transect Sayaut Figure 2. Transect layout in each grid for the sampling years 1994 and 1995. 6 1.2.3 Vegetation and habitat sampling I evaluated species composition and percent cover of herbs and shrubs by plot sampling every 30m along the transriparian transects. Ferns were included with the herbs. I recorded vegetation on transects #\,#3, and #5 of the grids in 1994. Vegetation sampling in 1995 was done on the upland part of these same transects. Shrubs and herbs (including ferns) were sampled within circular plots of 0.1 ha and lm2 quadrats, respectively. Tree species, canopy cover and snags were recorded in circular plots of 0.4ha every 60m along the same transriparian transects (Fig.3). I used line-intercept sampling to evaluate the volume of downed wood, employing the reverse triangle method, with 3 lines of 11 2m each, (Fig.4). I recorded the diameter and the decay class of each piece of wood encountered, and calculated the total DW in each sampling plot following Van Wagner (1968): V(m3ma)=(rc2x2(D2))/8xL where: D = diameter of the piece of wood L = length of the transect Downed wood was classified following within five decay classes: class 1 Twigs present, bark intact, hard texture and round class 2 Twigs absent, bark intact or partly missing, hard and decaying texture and round class 3 Twigs absent, traces of bark, partly decayed and round class 4 Twigs absent, bark absent, blocky texture and round to oval class 5 Twigs absent, bark absent, soft texture and oval. 7 I used abiotic factors (slope gradient and position, surface substrate, soil drainage) of each site to classify them according to the guidelines of "A Field Guide for Site Identification and Interpretation for the Nelson Forest Region"(BC Ministry of Forest, 1992). 8 Duwrt te C r e e k © e © © © © © 60 m ~®- Trapping station without vegetation sampling Herb quadrat o Shrub circular plot (0.01 ha) Trees,snag, and downed wood circular plots (0.04ha) I To end + of transect To end of transect I1 To end of transect Figure 3. Vegetation sampling plot design. 9 OowffsecS wood Line intercept sampling 10 2. SPECIES COMPOSITION, ABUNDANCE, AND HABITAT ASSOCIATIONS 2.1 INTRODUCTION Within a managed forest a mosaic of altered habitat is created throughout the riparian zones and uplands. Wildlife populations associated with riparian habitat are then influenced by both the remaining riparian zones and the altered stands. Changes in habitat caused by logging practices have shown a variety of impacts on small mammal community structure. Geier and Best (1980) suggested that when habitat is altered, populations of some species may benefit while others are adversely affected. Studies assessing small mammal abundance and species richness in managed landscapes have shown different impacts on community structure following harvest. Martell (1983) and Gyug and Vaartnou (1996) found that red back voles (Clethrionomys gapperi), which were prevalent in uncut stands, were replaced by deer mice (Peromyscus maniculatus) following harvest. Despite the changes in the species dominance, the total small mammal abundance seemed to remain the same in stands of different ages. Carey and Johnson (1995) found no difference of the community composition among different aged stands in Oregon forests, but reported that old-growth stands were supporting 1.5 times more individuals and biomass than managed forests. Van Home (1981), however, found a higher small mammal density in successional stands of coastal coniferous forest in southeast Alaska. l l Within unmanaged forests, among the studies conducted to assess small mammal population in riparian zones and adjacent uplands, Galindo-Leal et al. (1995) reported that the majority of these indicated a significantly higher number of small mammal species in streamside forests than in associated upland forests. Doyle (1990) and Cross (1985) found that the total population abundance and the species richness was significantly higher in riparian habitats of southwestern forests. Zuleta and Galindo -Leal (1993) found a greater species richness within riparian habitats of coastal B.C. forest but the population abundance was not significantly different. McComb et al. (1993) found no difference of the abundance and small mammal species richness between riparian and upland habitats in the Oregon Douglas fir forests but reported strong species association with each habitat. Anthony et al. (1987) noticed a higher small mammal species richness in riparian habitat compared to zones 20-30m away from the stream in old-growth forests of Western Cascades. Within managed forests, some authors investigated small mammal populations of riparian and upland habitats. Guyg and Vaartnou (1996) observed a greater species richness and total abundance within riparian areas than upland areas of unharvested forests and clearcuts of south central B.C. Among the species captured, Sorex bendirii were limited to riparian habitat, while Microtus pennsylvanicus, M. longicaudus and S. vagrans showed greater abundance within the riparian zones; deer mice were more abundant in upland areas. Within clearcut stands the meadow vole (M pennsylvanicus) population was dominant in riparian zones. Anthony et al. (1987) found no difference in small mammal abundance between habitats, though species richness was greater near stream zones of young, mature, and old-growth forests in the Cascade Mountain of Oregon. Gomez (1992), on the other 12 hand, found no difference in the species richness between stream and upslope habitats. His study was done within 5 forest types of Oregon. For seven of the nine species recorded (Neurotrichus gibbsii, Sorex bendirii, Sorex pacifwus, Microtus longicaudus, Microtus townsendii, Phenacomys albipes, and Zapus trinotatus) abundance was greater within the riparian zones than upslope (Gomez, 1992). The water shrew (Sorex bendirii) was found to be an obligate of riparian areas while Clethirionomys californicus showed a strong association with the upslope habitat. Clearly there is no consistent pattern in small mammal population abundance relative to riparian and upland habitats of unharvested and managed forests . This chapter focuses on the general habitat associations of the small mammal communities within riparian and upland areas of ICH managed forest. 13 2.2 METHODS I analyzed data at the habitat level (old-growth, buffered, and clearcut habitats) and at the sampling site level. There were 6 sampling sites or 2 per habitat, namely: old-growth site #1 and #2, buffered site #1 and #2, and clearcut site #1 and #2. Results from the buffered sampling sites are often separated and presented in two categories: the riparian buffer strip area and the adjacent clearcut. This finer subdivision of the broad habitat reflects marked changes in transects through riparian buffers. 2.2.1 Vegetation and downed wood I recorded vascular vegetation to species with the exemption of the willow family (Salix spp.). To facilitate data analysis both shrubs and herbs were grouped by soil moisture and nutrient requirements. Ferns are included within the herb data. The number of variables was reduced from 95 vegetation species to 39 vegetation groups. Tables 1 and 2 summarize the resulting classification. The soil requirements of each species were derived from "Indicator Plants of Coastal British Columbia" (Klinka and Scagel, 1989). Because the main focus of the study is the relationship between the habitat structure and the small mammal populations rather than with the specific floral species found, the grouping of the plant species does not affect the analysis and interpretation of the results. Techniques such as clustering and principle components were not employed to group the vegetation species because in my 14 study, location or distance from the stream is important. Grouping the vegetation species according to the variable variances would not account for locations. To express the vegetation species richness I used the original species rather than the vegetation groups. The overall vegetation species richness per sampling-site was recorded for shrubs and herbs (including ferns). I tested for differences in species richness between sampling sites with Analysis of Variance (oc=0.05). The ANOVA was run separately for shrubs and herbs. When the means were found to be significantly different, I tested the difference between means with the Bonferroni's comparison of means test (a=0.05). Vegetation composition was in terms of the vegetation groups (Table 1 and 2). I retained all of the vegetation groups that represented more than 1% cover per plot per site and gave their relative frequency in each habitat: old-growth, buffer strip and adjacent clearcuts areas, and clearcut. Frequency distributions are illustrated as dominance: diversity values by scaling the most frequent group as 1.0 for each habitat. I calculated the average percent vegetative cover per sampling plot and present the results for herbs (including ferns) and shrubs within each sampling site. I performed Multivariate Analysis of Variance on vegetative cover data using SAS package with PROC GLM (a=0.05). Shrubs and herbs were tested for sampling site effects (OG 1, OG 2, BUF 1, BUF2, CC 1, CC 2) nested within habitat effect (old-growth, buffered, clearcut). MANOVA tests the equality of vectors of means over several treatments analyzing all of the dependent variables simultaneously, using the assumptions of multivariate normality 15 and equal co variance matrices. Prior to MANOVA, the equality of the means both between habitat and between sites were tested for each dependent variable with Univariate Analysis of Variances (a= 0.05). MANOVA runs under the assumptions of normality and equality of variances. Multivariate normality was tested by a Goodness of fit test (a=0.05) on all of the variables. The variance homogeneity was tested using a Chi-Squared test (a=0.05). Because the data did not meet the assumptions, they were submitted to the angular transformation: arcsinVx. Arcsin transformation is known to create normal distribution from original percentages or proportions which usually form binomial distributions (Zar, 1984). The assumptions of normality and homogeneity of variances were tested again. Manova was run with transformed data. Data on downed wood were treated similarly to the vegetation where the average volume per sampling plot was given for each habitat. I calculated how much of the wood volume was represented per diameter class and decay class. The average number of logs per plot recorded for each diameter class was compared among the habitats studied. 2.2.2 Small mammal populations Small mammal species richness and composition were calculated for each sampling site. Differentiation among some species can be very difficult with live specimens. Individuals from the genera Sorex and Microtus were not identified to species with the exception of the southern red-backed vole (Clethrionomys gapperi). 16 Tablel. Grouping of the shrub species recorded according to the soil moisture, nitrogen level, and family. Groups are the variables used in analysis. The soil moisture and nitrogen level are indicated within the column "soil" where: N=nitrogen rich, n=nitrogen poor, m=nitrogen medium, 1=watersheding site, 2=flood plain and water receiving site, 3=very moist to wet, 4=fresh to very moist 5=fresh to moist, 6=dry to moist, 7=very dry to moderately dry, 8=disturbed site. Var. Common name Latin name Soil RASB Red raspberry Rubus idaeus N 4 WMA Western mountain ash Sorbus scopulina Sitka mountain ash Sorbus sitchensis n 6 ELDER Red elderberry Sambucus racemosa N 4 GOOS Black gooseberry Ribes lacustre N 5 MAPL Douglas maple Acer glabrum N 1,2 CRAN Highbush cranberry Viburnum edule N 4 THIM Thimbleberry Rubus parvWorus N 8 DCLB Devil's club Oplopanax horridus N 3 ALDR Sitka alder Alnus viridis ssp. sinuata Mountain alder Alnus incana ssp. tenuifolia N 4 HAZL Beaked hazelnut Corylus comuta N 6 HUCK Black huckleberry Vaccinium membranaceum n 6 BLUB Oval leaved blueberry Vaccinium ovalifblium n 4 FALS Falsebox Paxistima myrsinites n 6 SASK Saskatoon Amelanchier alnifolia m 6 TWIB Black twinberry Lonicera involucrata N 3 DGW Red osier dogwood Cornus stolonifera 3 CTW Black cottonwood Populus balsamifera ssp. trichocarpa 2 , 4 WILL Willow family Salix spp. TREE White spruce Picea glauca Western redcedar Thuja plicata Douglas fir Pseudotsuga menziesii var. glauca Western yew Taxus brevifolia Western hemlock Tsuga heterophylla 1 7 Total numbers of individuals captured per sampling site were calculated. Results are presented as the number of captures per trap over the total number of trapping-nights. Recaptured individuals are excluded from these values in 1995. The results therefore document the average number of individuals captured once per trap. This approach allowed appropriate per unit comparison of the small mammal relative abundance in each of the habitat types. I compared the relative small mammal abundance between sampling sites with Analysis of Variance (a=0.05) for both trapping years (1994 and 1995) separately. When the means were found to be significantly different I tested the difference between means with the Bonferroni's comparison of means test (a=0.05). 2.2.3 Population density estimates Mark/recapture data were collected from an open population which changed in size and composition over time. Animals were marked during the trapping season of 1995 and density estimates are calculated for that year only; no animals were marked in 1994. Relative abundance of the population was calculated weekly by the Jolly-Seber model applied within each of the sampling sites. Each week represented a trapping period. Analysis was performed with the small mammal programs for mark/recapture data analysis written by C. J.Krebs (1991). This model was developed for open populations and allows estimation of the population size at each sampling time as well as the survival rates and birth numbers between sampling times (Pollock et al. 1990). It can be run with samples that are not at regular time intervals and can be run on data extending over many years. The Jolly-Seber method works 19 on the assumptions that: 1) every individual has the same probability of capture at the time of the ith sample (/' = l,2,...,k); 2) every marked individual has the same probability of surviving from the ith sample to the (/ + 1) sample; 3) individuals do not lose their marks, and marks are not overlooked at capture; and 4) sampling time is negligible in relation to intervals between samples (Krebs, 1989). Prior to density estimation, assumption 1 was evaluated with the Leslie and Chitty test of equal catchability, which is the preferred test when dealing with more than three samples from an open population (e.g., Krebs, 1989). Leslie and Chitty test works under the principle that it is possible to have unequal catchability between marked and unmarked animals but still have equal catchability within the marked segment of the population (Krebs, 1989). Therefore it uses only the marked individuals who were caught twice or more. The null hypothesis of equal catchability was tested using a Chi square distribution (a=0.05). The Jolly trappability estimates were calculated for the small mammal population found within each of the sampling sites. This estimate was chosen over the maximum and minimum trappability estimates following of Krebs and Boonstra (1984) who found that it was the best method after comparing them all. Jolly trappability is defined: Jolly trappability = Z (Total no. of marked individuals caught at time i) (100) Estimated marked population size at time i S 20 where S is the number of sampling times. If the assumption of equal catchability does not hold, trappability is usually overestimated because estimates are weighted by individuals that are caught more often (Krebs and Boonstra, 1984). Trappability was estimated because it gives an index of the accuracy of the relative population density calculated. Accuracy increases with the portion of the population captured. Reported relative densities include all species except Sorex spp. and Mustela erminea. The relative density was not calculated for the species individually because the sample sizes were too small. Density estimates are given for each of the sampling sites separately, as well as for the buffer strip areas and the adjacent clearcut zones, when sample sizes are large enough to use the Jolly-Seber model. The density estimates were calculated weekly or for time periods covering one trapping session. Length of the trapping transects varied among sampling sites. The Jolly-Seber population density was then presented as the population number per ha to facilitate the comparison of the population density among the habitats under study. Number of ha contained within each sampling site included the surface area found between the transriparian transects as well as a 7.5m to 15m strip surrounding the grid which represents half the distance between trapping stations in a transriparian and parallel to the stream fashion. 21 2.3 RESULTS 2.3.1 Vegetation and downed wood Species richness No significant difference in shrub species richness (Fig. 5) was found among the sampling sites (df=5,48, F=l. 132, P=0.356, a=0.05). Herb species richness was significantly different (df=5,48, F=12.376, P=0.000, a=0.05). The Bonferroni's comparison of means showed that the buffered site #1 was significantly different from all other sampling sites. Buffered site #2 differed from old-growth site #1 (a=0.05). Species composition Ferns (group F3 in Table 2), preferring moist or wet nitrogen-rich soils, dominated the old-growth areas along with devil's club (Oplopanax horridus) (Fig.6a). Alder {Alnus spp.) and thimbleberry (Rubus parviflorus) comprised 30 to 40% of the vegetation by relative frequency. All other vegetation groups comprised between 10 to 20% of the relative frequency (Cornus stolonifera, Streptopus spp., Tiarella trifoliata, Lonicera involucrata, Circacea alpina, Corylus cornuta) and are also typical of fresh or wet nitrogen rich soils. Riparian buffer strips zone were dominated by L. involucrata (Fig.6b) but contained several species groups at high relative frequencies (50-60%). O. horridus, R. parviflorus, Cornus canadensis, ferns, and C. stolonifera found in rich nitrogen and fresh to wet soils were 22 among these. Among the 7 vegetation groups composing 20 to 30% of the relative frequency some species (Vaccinium ovalifolium and Equisetum arvense) are known to be associated to nitrogen poor or medium soils. Clearcut habitats were inhabited by vegetation species typical of disturbed sites. Thimbleberry (R. parviflorus) and red raspberry (R. idaeus) represented the highest frequencies within the clearcut zones (Fig.6c and d), including all clearcut sampling areas with and without an adjacent riparian buffer strip. Strong dominance was most clearly expressed in clearcuts and there were fewer vegetation groups within the clearcut areas with high relative frequencies than in the buffered zones. Ferns, Salix spp., and C. stolonifera accounted for 30 to 40% of the frequency. Populus balsamifera, Epilobium angustifolium, and Pteridium aquilinum all ranged between 20 to 30 % of the relative frequency. Vegetative cover Average percent cover of shrubs per plot in all sampling sites varied from 58 to over 100% cover (overlapping cover). Buffered site #2 contained 20% to 43% more shrub cover than all of the other sampling sites (Fig.7a). Herb and fern layers exhibited two general tendencies: the clearcut zones of the buffered and non-buffered habitats contained 30% less herbs than the old-growth and riparian buffer strip sites (Fig.7b). 23 Vegetat ion species r ichness in all si tes Figure 5. Total vegetation species richness reccorded in all sampling-sites. BUF=buffered habitat (BUFSTP+BUFCC), BUFSTP=buffer strip, BUFCC=clearcut adjacent to riparian buffer, OG=old-growth, and CC=clearcut. 24 n co JC o o (0 Q. 3 2 O) 10 cu cn cu > u c a> 3 cr a> > or 4-2dS w s u s IdVlftl SOSlftl 10SIAI 10IA IH10 Id 1NIM NVUO „ a. TIIM | D) aai3 o 'I ama | , Z H I O > sooo 1ZVH eosiAi ai/vu tosw m Moa 3381 IftllHl ua iv a ioa Ed n n ro (A I 3 « a. (0 a 3 2 cn ra a> cn > o >. u c a> 3 CT 9> cu > 1 CU CC T- oo ci (O ci ci ci (0 t o CU I 3 C O a. 3 2 O) c o ra aJ a a> > CO 3 O) JQ CO 3 co CO C O % 2 C D i 33 o ro !5 ro sz o ra cu c o (75 o C L E o o c g ro cu C O cu > CO cu c : sz o £ £ cu A 3 ro CT ,_ CU o cu T J *- c ro co cu T -cu „N ro £ H C O aj LL co m CM 3 U i— to co — O (Q = 2 w 5 a, 2 .S-2 ~ o> tn = i •2 £ S 0) W > "C u_ (Q o a. s = > ™ re a> OS re re 3 U (0 a> o c (A a 3 o I— O) re a> CO > a> >< o CT > re a> o CO 9dS eds 20SI/M 1dW\l 10 IA x o n H NVUO i d cosw uaiv aioa m ama aaia i d s i mo S l V d H I M 1ZVH tOSIAI M10 2dS ai/vu CH10 2HJ.O S 0 0 0 3381 Zd /v\oa asvy Cd INIH1 a. 3 o O) c o en v > T3 (O O) 3 re o co C o re •a re 3 co a> u re o re cu c o lo o Q. E o o tz o co cu O) cu > CO cu tz £2 g P o £ cu n 3 re 8> £ IN •a cz re CO (O CN O d ci ci ci X3 CD cu i 3 g> u. cu > re cu Qi co 8> . = cu CO cu U . CO _CD .a re co CM Data transformation (arcsinVX) did not manage to meet the assumption of multivariate normality and homogeneity of variances. All variables differed significantly from a normal distribution (P=0.0001, oc=0.05) and the equality of variances test was rejected (P=0.0001, a=0.Q5). Failure to meet assumption affects the reliability of the MANOVA results. The data set contained a lot of zero values because species were absent from a plot. There was a signiiicant site effect for both herb cover (df=66, 398, F=3.97, P=0.0001, a=0.05) and shrub cover (df=54, 417, F=3.32, P=0.0001, a=0.05) analysis. All vegetation groups or dependent variables (Table 1 and 2) that were found to be significant between habitat and/or site from the univariate analysis of variances (a=0.05) are shown in Tables 3 and 4. Percent cover per plot in each sampling site and P values calculated for the habitat effect and nested site effect are given. Downed wood The volume of downed wood recorded was highest in the clearcut sites (53.4m3/plot) followed by the buffered (43.4m3/plot) and old-growth habitats (37.3m3/plot). Plots were O.lha. Most pieces encountered on clearcut sites were logs of 7.5 to 40cm diameter, while the pieces found within the riparian buffer strips and undisturbed sites were of either small diameter (7 to 30cm) or large diameter logs (50 to 100cm). Downed wood pieces within the clearcut zones were estimated to be in decay class 3 and 4. Decay class 4 and 5 decay wood dominated for downed wood recorded within the 27 Figure 7a Figure 7b. Figure 7.Vegetative percent cover per plot within each sampling site. The average percent cover per habitat is also given: shrubs (Fig. 7a), herbs (including ferns; Fig. 7b). OG=old-growth, BUF=buffered habitat, BUFSTP=buffer strip, BUFCC=clearcut adjacent to riparian buffer, CC=clearcut. 28 Table 3. Shrub groups as defined in Table 1 and 2 that were significantly different from the MANOVA analysis. The average percent cover per plot in each sampling site is given as well as the P value calculated for the habitat and nested sampling site. BUF :=riparian buffer strips and adjacent clearcuts. Shrub Variables OG 1 %/PLOT OG2 %/PLOT BUF 1 %/PLOT BUF 2 %/PLOT CC 1 %/PLOT CC 2 %/PLOT P values RASB 0.1 0.0 5.6 18.0 14.0 4.9 h: 0.0017 s(h): 0.0006 WMA 0.2 0.0 0.1 0.0 0.1 0.3 h: 0.0001 s(h): 0.0001 GOOS, 3.6 1.6 1.5 2.2 1.7 5.9 h:.01277 s(h): 0.0222 MAPL 0.1 1.3 1.1 0.1 0.4 0.7 h: 0.2594 s(h) : 0.0403 CRAM 0.0 3.2 0.2 0.1 1.2 0.1 h: 0.2688 s(h): 0.0280 THIM 12.1 7.4 27.2 27.6 24.4 30.0 h: 0.0009 s(h): 0.0001 DCLB 15.7 19.4 3.9 5.1 1.6 1.3 h: 0.0001 s(h) : 0.7563 ALDR 10.2 10.2 1.0 1.7 2.0 0.8 h: 0.0001 s(h): 0.7779 HAZL 0.3 8.2 4.4 3.9 1.7 2.5 h: 0.0001 s(h): 0.9545 HUCK 0.5 0.2 0.7 0.1 0.9 0.3 h: 0.3084 s(h) : 0.0154 SASK 0.0 0.1 0.2 0.1 0.0 0.2 h: 0.0005 s(h) : 0.2781 CTW 0.0 0.2 4.3 7.7 1.6 2.9 h:.0.0723 s(h) : 0.0129 WILL 0.4 3.0 2.8 . 13.7 1.9 2.0 h: 0.4145 s(h) : 0.0061 TREE 3.4 11.0 5.0 6.2 6.2 5.5 h: 0.0001 s(h) : 0.1298 h=habi ta : s(h)=site nested wit l i n habitat 29 Table 4. Herb groups as defined in Table 1 and 2 that were significantly different from the MANOVA analysis. The averagepercent cover per plot in each sampling site is given as well as the P valuecalculated for the habitat and nested sampling site. BUF-riparian buffer strips and adjacent clearcuts. Shrub Variables OG1 %/PLOT OG2 %/PLOT BUF 1 %/PLOT BUF 2 %/PLOT CC 1 %/PLOT CC 2 %/PLOT P values F1 0.2 2.7 3.0 5.8 0.7 0.9 h: 0.9468 s(h): 0.0006 F2 0.0 0.1 2.2 2.2 11.2 3.0 h: 0.0017 s(h): 0.0001 F3 43.5 23.6 9.7 12.7 12.4 10.3 h: 0.0268 s(h): 0.3660 SF1 0.0 1.3 2.9 1.2 3.3 0.2 h: 0.0005 s(h): 0.0041 SF2 0.6 0.3 1.7 2.3 1.7 3.3 h: 0.0001 s(h): 0.0006 SF4 0.0 1.0 0.6 0.4 0.1 0.3 h: 0.0014 s(h): 0.0055 SF5 0.0 0.1 0.1 0.0 0.0 0.1 h: 0.2252 s(h): 0.0065 SF6 0.0 0.2 0.8 0.2 0.3 0.3 h: 0.3503 s(h): 0.0120 LL3 0.1 0.3 0.1 0.0 0.2 0.2 h: 0.2940 s(h): 0.0091 MSC3 0.1 2.8 1.2 1.8 0.3 1.1 h: 0.0213 s(h): 0.0017 MSC4 6.4 4.6 2.2 2.2 3.3 1.4 h: 0.0221 s(h): 0.3199 MSCS 0.0 0.1 0.0 0.1 0.0 0.0 h: 0.0001 s(h): 0.0001 VIOL 0.7 1.9 0.9 0.6 0.6 0.6 h: 0.0001 s(h): 0.1491 WINT 0.8 2.3 0.1 0.4 0.0 0.0 h: 0.1726 s(h): 0.0001 OTH2 2.6 2.2 5.4 1.8 4.7 2.3 h: 0.0036 s(h): 0.4743 h=habitat: s(h)=site nested within habitat 30 old-growth habitat. The riparian buffer strip areas were dominated by decay class 4 followed by class 3 decay. 2.3.2 Small mammal populations Species richness and composition Six small mammal species or genera were captured. Deer mouse (Peromyscus maniculatus), Western Jumping mouse (Zapus princeps), vole (Microtus sppj, Shrew (Sorex spp,), Red Backed vole (Clethrionomys gapperi), and short tailed weasel (Mustela erminea) all occupied clearcut sites. M. erminea was not captured within the riparian buffer strips areas, nor within the undisturbed habitats. Small mammal abundance The pattern of small mammal abundance varied from 1994 to 1995. The number of animals captured per trap in 1994 (Fig.8a) was greatest within the old-growth habitat (2.03 animal/trap) followed by the buffered (1.65 animal/trap) and the clearcut habitats (1.19 animal/trap). The following sampling year, 1995, the transriparian transects were extended from 150m up to 300m upslope. The small mammal abundance then showed opposite trends. The population was the highest within the buffered and clearcut areas with 2.78 to2.98 animals per trap (Fig.8a). The lowest abundance was found within the undisturbed habitats with 2.08 animals per trap. The riparian buffer strip zones sustained a lower small mammal 31 abundance (2.50 animal/trap) compared to the adjacent clearcuts (Figure 8b). The ANOVA test run on the sampling-sites showed significant differences in small mammal population abundance among sites during both trapping years (1994: df=5,60, F=l 1.010, P<0.001 and 1995: df=5,108, F=10.420, PO.001, a=0.05). Bonferroni's comparison of means test found that the buffered site #2 differed signiiicantly from all sites except old-growth #2, and the old-growth site #2 differed significantly from buffer site #1 and clearcut site #1 in 1994 (oc=0.05). In 1995 the population abundance was significantly different between the old-growth site #2 and all other sites except the old-growth site #1 (oc=0.05). It is not possible to compare these results between years because the trapping activities in 1994 were interrupted a month and a half earlier (hantavirus) and the animals were not marked that year. Population abundance was calculated from the total trapping data for each year. 2.3.3 Population density estimates The assumption of equal catchability was evaluated for each of the sampling sites (a=0.05). Catchability was found to be equal everywhere except within the clearcut site #2, the old-growth site #2, and the buffered site #2. The Jolly trappability varied between 54.5% and 91.3% (Table 5) with large confidence intervals. 32 Total population densities calculated by the Jolly-Seber model as a function of the time are illustrated in Figure 9 for each of the sampling sites in 1995. Small mammal population density within buffered site #2 seemed to be negatively correlated with time. The weekly data were then aggregated and the average small mammal density for each sampling site was calculated as the average of all weeks. The old-growth habitat sustained the lowest small mammal population density followed by the clearcut and the buffered habitat (Fig. 10). The weekly population densities per ha for each trapping site are summarized in Table 6. The standard error of the population densities were often unusually high for the calculated values within the riparian buffer strip areas and the adjacent clearcut zones as a result of small sample sizes. 33 D- O LL (- o ^ t n u . c a CO CM 0. CM o 3 m 3 Sampl ing sites 13 m - Mean/habitat T- CM o o o o Figure 8a. Figure 8b. Figure 8. Small mammal relative abundance. Number of animals captured per trap: 1994 (Fig. 8a), 1995 (Fig. 8b). The average per habitat is given.OG=old-growth, BUF=buffered habitat, BUFSTP=buffer strip, BUFCC=clearcut adjacent to riparian buffer, CC=dearcut. 34 s LO o i n o i n ( CSI CM eii/Ajjsuap uoi)e|ndod n CD i ZJ O ) i i . cu CJ LO O LO ei|/A)!suap uo;)e|ndod T3 2 g> LL CQ CQ CO eu/Ausuap uoj)e|ndod co CD CD C O i i 0) eu/Ajisuap uo!JB|ndoj o o> 2 3 C O CC CD C I ) LL .Q CO x: % 2 C O t o (/j 2 3 Q . CO O LO CD CD E £ cu E CD > o w co E E co E E "a * r C O CO to o £ C D co • E •-CO 0 3 CO >> * i • t i . Q co S S c 2 g OJ CO 3 3 -a Q . CD a> cb •g LL co o ^ 2 § 2. ro C O CD LL" o co Figure 10a. 14 -I 12 -is CC BUF BUFCC BUFSTP OG Figure 10b. Figure 10. Average population density of small mammals per ha calculated with the Jolly-Seber model from 1995 captures: average density per sampling-site (Fig.10a), average density per habitat type (Fig. 10b). OG=old-growth, BUF=buffered habitat, BUFSTP=buffer strip, BUFCC=clearcut adjacent to riparian buffer, CC=clearcut. 36 CD "(7) O) c a. E ca to sz o CO CD T J CD 3 o CO co n co a. C L co LO 0 ) CO ca cu c CD U c CD n c o o co co csi O CM o o CD csi m co n | l co co CO 00 c E co T - CM 2 2 co co] 32 2 O O . 0 ( 0 g ci od 2 CT) CO o o o CJ) CN LO T f cj) b CN T f LO CO LO T f T f T f LO CO LO 3 T ~ O CL CO £ C CD IS o -g l_ l_ l_ i t : fc 3 3 3 CO CD CD CD co tf) s LO LO* CO t - CN O o o o r- LO oo co j^" cd T f x - CO co o co T — b CN co s s CM CM CM CJ CD CL CO to o CD CD CD fc fc fc 3 3 3 CD CD CD i - CJ) CJ) LO o o T f T " co co co T f co o CJ> oo T - CM 3 3 2 2 co co 03 J D o o as '55 CO e Q. E CO to .c o CO CD T J CD 3 o s to CD tocz CD T J c o ro 3 C L o Q . t_ CD £3 CD CO CD CD CO 0 .O co H 03 8 s -H -H LO LO oo cj> b CN CO CJ) CO CO I— LO CO T -N •* (N •H -H -H -H T f T f CO CO CJ) CJ> CO S CD N O CO T — CO T — -H -H -H -H CO CJ) CN 00 CO 00 T f T— CN T-CO CD CO CO LO x— -H -H O CO CN LO CJ) co •H -H CN CN 00 to Tf N T — OO CN csi •H -H -H LO CN CO O) S O N •H -H T f i -O 00 00 T f Csi CM •H •H CD LO LO CO N CO S csi csi co co •H T f -H T f CD CD CO CO •H +1 T - CO LO T f r-- LO CD CJ) co • 00 •H -H -H +1 CO T — CO LO CD b T f csi co •H -H -H -H T f T - i - CO CD LO LO CO O O) S N b T— csi co •H •» CO T -T f CO CD 00 CO T f LO CO CO 2.4 DISCUSSION The general overview given here includes all of the small mammal species captured. It repoits the abundance, density and species richness of the small mammal population as a whole without treating species separately. Because habitat structure and small mammal populations differed among the sampling-sites investigated most of the analysis treated data at the sampling site level rather than the habitat level. The sampling sites were selected for their similarities in habitat structure but exhibited more differences than expected. Percent vegetative covers of sites within habitats differed significantly. The analysis may not be entirely reliable because the assumptions of MANOVA were not met before or after data transformation (arcsinVX). Herb cover (including ferns) was more abundant within old-growth and the riparian buffer strips zones where the exposure to solar radiation was less and the moisture levels were higher. The higher moisture of the old-growth and buffer strips areas was consistent with the vegetative species composition (Fig.6). Ferns (refer to group F3 in Table 2) and devil's club (O. horridus) dominated the old-growth areas. Riparian buffer strips are believed to experience edge effects and their structure becomes modified from the typical old growth habitat. The edge which is defined as the junction of two different landscape elements (Yahner, 1988) creates a transition zone which in 38 this case is part of the buffer strip areas. Vegetation composition of buffer strips was dominated by L. involucrata, typically found on high nitrogen and wet sites. R. parviflorus, typical of disturbed and high nitrogen sites, represented 60% of the vegetation composition in buffer strips (Fig.6b). That species also dominated the clearcuts but accounted for only 25% of the old-growth vegetation composition. The vegetation composition thus seemed to illustrate a transition stage of the buffer strip habitat structure, where species were generally found at a higher frequency than in the other habitats (Fig.6). Vegetation species richness (Fig. 5) and cover (Fig.7) were too variable to permit general statements. These habitat components were found to be either non-significant among sites or particular to each sampling site. Downed wood is believed to provide an important source of refuge, resting areas, and feeding and breeding sites for wildlife (Bunnell et al. 1991, 1997). Downed wood may favor certain populations like species of the Sorex genus foraging on insects (Craig, 1995). The volumes available, the average diameter, and the level of decay differed and seemed to indicate that riparian buffer strips once again were intermediate between old-growth and clearcut zones. The pattern of small mammal relative abundance seemed to differ between trapping years: (1994 and 1995). Trapping data of 1994 and 1995 were not tested against each other because of the disparity in transect length, duration of trapping season, and marking process. Small mammal relative abundance differed significantly among sites in 1994 (ANOVA: df=5:,60, F=l 1.010, P<0.001, a=0.05). Relative abundance seemed the highest within the old-39 growth habitat. When sampling sites were examined separately, the highest abundance was found within old-growth site #2 (Fig. 8a) and was significantly different from relative small mammal abundance in buffer site #1 and clearcut site #1 (Bonferroni's comparison of means test, a=0.05). Small mammal relative abundance in 1994 was also high within buffered site #2 and significantly different from abundance calculated in all other sampling-sites except old-growth #2. There was no strong difference among habitats in 1994. In 1995, when the captures included the associated upland species, the overall picture changed with the highest overall abundance within buffered habitat and the lowest in old-growth habitat (Fig.8b). Small mammal relative abundance differed significantly among sites (ANOVA: df=5,108, F=10.420, P<0.001, a=0.05). In most cases clearcut areas of buffered sites showed a higher abundance than the buffer strip zones. Only the relative abundance calculated within old-growth site #2 was found significantly different from all other sampling-sites (except old-growth site #1; Bonferroni's comparison of means test, a=0.05). Small mammal relative abundance was found to differ significantly within only one of the six sampling-sites in 1995, and again there were no pronounced differences among habitats. The population density calculated by the Jolly-Seber model followed the overall relative abundance pattern of 1995 captures (Fig.8b). The density seemed slightly higher within the buffered and clearcut habitats followed by old growth habitat (Fig. 10). 40 3. TRANSRIPARIAN GRADIENT AND HABITAT ASSOCIATIONS 3.1 INTRODUCTION The majority of the studies on small mammal populations and riparian habitats compare the riparian communities with separated upland communities (e.g.: Doyle 1990, Cross 1985, Zuleta and Galindo-Leal 1993, Gyug and Vaartnou 1996). From the few studies investigating trannsriparian changes in small mammal communities, McComb etal. (1993) found that deer mice, Pacific jumping mice, shrew moles, and marsh shrew capture rates decreased with distance from the stream while Clethrionomys occidentalis increased at 200m and more from the stream. Gomez (1992) found that along 200m transriparian transects, capture rates of S. pacificus and S. bendirii were higher near streams than farther away. Clethrionomys californicus was more abundant at distances farther from the stream. Studies of transriparian gradients in riparian zones should take into account the adjacent upland portions. It has been suggested that upstream and upslope influences are inseparable from the riparian zone. The integrity of a riparian ecosystem depends on and is influenced by the adjoining upland area (B.C. Ministry of Forests and Ministry of Environment, 1995b). Considering the importance of the uplands to the riparian habitat it must be recognized that narrow riparian buffer strips that do not contain upslope habitat in a managed landscape may not meet the needs of upslope associates, such as C. gapperi (McComb and Hagar, 1993). Very few studies on small mammals and riparian habitat were able l:o suggest proper riparian buffer width to sustain the population investigated. Gomez 41 (1992) and McComb et al. (1993) suggested that streamside corridors wider than 150m should be sufficient to include the optimal habitat used by C. gapperi. In this chapter I examine the small mammal and vegetative communities along transriparian gradients that extend from the edge of the stream to upslope areas (300m away from stream). Johnson and Lowe (1985) suggested delimiting the inland extent of the riparian areas by functional differences rather than physical or biological definition, because of non-apparent and gradual differences in habitat that may originate farther away from the observed riparian zones. Johnson and Lowe (1985) also considered the transriparian gradient a major factor in controlling the biological biodiversity and structural complexity of riparian zones in the southwestern U.S. My objective is to evaluate the distances or gradient at which small mammal population parameters may change and to relate these to differences in habitat structure. Such data should permit better assessment of the role and effect of riparian buffers in managed landscapes on small mammal communities. The riparian buffer strip width (60-90 m) investigated here is therefore assessed, for its role into sustaining the small mammal population under study. 42 3.2 METHODS I analyzed data at the habitat level (old-growth, buffered, and clearcut habitats) and at the sampling site level. There were 6 sampling sites or 2 per type of habitat, namely: old growth #1 and #2, buffered #1 and #2, and clearcut #1 and #2. Results from the buffered sampling sites are often separated into two sets of data: the riparian buffer strip area, and the adjacent clearcut zone. I performed this subdivision to permit more specific evaluation of habitat associations of the small mammal populations. I recorded vegetation species richness and percent cover as a function of the distance from the edge of the stream going upslope for each habitat and sampling site. The cover and species richness are calculated for both the shrubs and herbs (including ferns). I calculated the vegetative cover from the vegetation group variables and the species richness from the original species. Because the vegetation was sampled every 30m (Fig.3), the cover and species richness per plot is given at 30m intervals starting flush with the stream to the end of the transriparian transects. Sampling site and distance effects on shrub and herb species richness were tested with a two-way ANOVA (ot=0.05). I analyzed the vegetative percent cover the same way with two-way ANOVA (a=0.05). I ran a t-test (a=0.05) on the buffered site data to evaluate if the vegetation from the buffer strips and adjacent clearcuts were drawn from the same populations. 43 I recorded small mammal species richness, composition and relative abundance as a function of the distance from the stream. The traps were set at 15-m intervals along the transriparian transects (Fig.3). The small mammal population parameters were then calculated for every 15-m interval starting flush with the stream going upland. The mammal species richness and composition was plotted for all sampling sites and each trapping year (1994 and 95). I analyzed data on population relative abundance (# animals/trap for all species pooled) for both trapping years separately. I tested the sampling site and distance effects on the population abundance with a two-way ANOVA (a = 0.05). I ran a t-test (oc=0.05) on the buffered site data when the distance effect was significant to evaluate if the small mammal data from the buffer strips and adjacent clearcuts were drawn from the same populations. I looked at the small mammal species and genera individually. I calculated the normalized abundance of each species per type of habitat (old-growth, buffer strip, adjacent clearcut, clearcut) for both trapping years. Normalized abundance was calculated using total number of animals captured (excluding recaptures) per trapping unit and trap/night within each habitat. Frequency distributions of normalized abundance are illustrated by scaling the most abundant species as 1.0 for each habitat and having the other species standardized according to that most abundant species. Deer mice were not included with the frequency distribution of 1994 small mammal normalized abundance. Deer mice dominated the small mammal population and when included, did not allow detection of differences among the other small mammal species that were captured in very low numbers. 44 I present plots of the small mammal species capture rates along the transriparian gradient with the exception of P. maniculatus and M. erminea. These graphs use capture data of the second trapping year only (1995) because capture data in 1994 are very low for all of the species other than P. maniculatus. The trapping season was short in 1994 and perhaps not representative. Deer mice were not included because they dominate the small mammal population. Analyses of the entire small mammal population are greatly influenced by this species and are a good indication of its abundance and distribution among the habitats (See Chapter 2). Short tailed weasel data were insufficient to produce representative plots. Graphs were produced for all habitats (old-growth, buffered clearcut, and clearcut). Linear distributions of the capture rates of small mammal species with distance from the stream among the habitats were developed. The significance of each slope of the capture rate over the distance from the stream was tested using a two-tailed student's / test on a Simple Linear Regression drawn from the functional relationship: Yi = a + (3Xi (a=0.05). The coefficient of determination (r2) or the proportion of total variation in Y accounted for by the fitted regression also designated as the strength of the straight-line relationship (Zar, 1984) was calculated for each regression. 45 3.3 RESULTS 3.3.1 Vegetation Species richness Species richness of shrubs and herbs (including ferns) are given along transriparian gradients for each sampling site (Fig. 11). Herbs species richness within the old-growth site #2 was reduced by half from 150m going upland . The slope changed from level to 40% at 150m away from the stream affecting the habitat characteristics. The herb species richness within buffered site #1 increased from 90m on, where the transition from the buffer strip to the adjacent clearcut area occurred (Fig. 1 lc). This clearcut zone contained some temporary streams that might have influenced the herb layer composition. Site and distance effects were not significant for shrub species richness (df=5,40, F=1.222, P=0.198 and df=8,40, F=1.473, P=0.317, a=0.05) nor was a distance effect significant for herb species richness (df=8,40, F=1.438, P=0.211, a=0.05), but the site effect was found significant (df=5,40, F=13.279, P<0.001, a=0.05). Percent vegetative cover None of the habitats and sampling sites demonstrated a specific trend in the distribution of percent cover of shrubs and herbs with distance from the stream (Fig. 12, 13, 46 and 14). A sharp increase (3.5 times higher) of shrub cover from 180m and greater occurred within the undisturbed site #1 (Fig. 12a). This portion of the sampling site included a depression with greater soil moisture that favoured the growth of shrubs typical of high nitrogen soil namely O. horridus, Alnus spp., and R. parviflorus. There was a significant site and distance effect from the RCB design analysis (df=5,40, F=3.46, P=0.015 and df=8,40, F=2.224, P=0.006, a=0.05) for shrub percent cover. No significant distance effect was found for herb cover but the site effect was significant (df=5,40, F=5.884, PO.001, df=8,40, F=0.991, P=0.457, a=0.05). A f-test run on shrub cover showed that the means from the buffer strip and adjacent clearcut populations were not significantly different and therefore drawn from the same population (df=9, t=2.15, and df=8, t=0.52, a=0.05). 47 Figure 11a. Vegetation species richness in old-growth site 2 Distance from stream (m) Figure 11b. Figure 1 1 . Vegeta t ion species r ichness and d is tance f r o m the s t ream: o ld-growth site 1 (F ig.11a) , o ld-growth site 2 (Fig. 11b), buf fered site 1 (F ig .11c) . 48 Vegetation species r ichness in buffered site 2 SHRUBS HERBS TOTAL Buffer strip Distance from stream (m) Figure 11d. Vegetat ion species r ichness in clearcut site 1 T- T- CM CM CM Distance from stream (m) Figure 11e. Vegetation species r ichness in clearcut site 2 SHRUBS HERBS TOTAL Distance from stream (m) Figure 11 f. F igure 1 1 . Vegeta t ion species r ichness and distance f r o m the s t ream: buf fered site 2 (Fig.11d), c learcut site 1 (Fig. 11e), c learcut site 2 (F ig.11f) . 49 160 -i 0M 30M 60M 90M 120M 150M 180M 210M 240M Distance from stream (m) Figure 12a. Figure 12. Percent vegetative cover per plot within the old growth sampling-sites as a function of the distance from the stream: percent shrub cover (Fig. 12a), percent herb cover (Fig. 12b). 50 140 1 § § , CM 8 § 2 § ° 1 T- T- CM CM CM Buffer strip Distance from stream (m) Figure 13a. Figure 13 b. Figure 13. Percent vegetative cover per plot within the buffered sampling-sites as a function of the distance from the stream: percent shrub cover (Fig.13a), percent herb cover (Fig. 13b). 51 160 E CC 1 race 2 Distance from stream (m) Figure 14 a. Figure 14 b. Figure 14. Percent vege ta t i ve cove r per plot wi th in the c learcut samp l ing s i tes as a funct ion of the d is tance f r o m the s t ream, a) percent shrub cover , b) percent herb cover . 52 3.3.2 Small mammal populations Species richness and species composition In 1994 the number of small mammal species captured along the 150-m transects within the old-growth and clearcut sampling sites varied between 2 and 3 . Members of Sorex spp. and M. erminea were absent from the old-growth. All of the species and genera captured were found within the clearcut habitat (P. maniculatus, Microtus spp., C. gapperi, Z. princeps, Sorex spp., M. erminea, Fig. 15). Small mammal species richness was generally higher in the buffered site #2 ranging between 3 and 5 species captured. Buffered site #1 had the poorest species richness of all sites with only 1 or 2 species at most distances (Fig. 15). OnlyM erminea was absent from the buffered habitats. In 1995 the small mammal species richness in relation to the distance from the stream followed two different trends. In half of the sampling sites including old-growth site #2, buffered site #2, and clearcut site #1, the average number of taxa found for most distances varied between 1 and 3. Remaining sites (old growth site #1, buffered site #1, clearcut site #2) had a higher species richness ranging between 3 and 5 species and genera at all distances. M. erminea and C. gapperi were absent from the old-growth habitat. All the species and genera captured were present within the buffered and clearcut habitats (Fig. 16). 53 Small mammal abundance In 1994 the site effect was found significant for small mammal abundance (Fig. 17) with a Randomized Complete Block design (df=5,50, F=l 1.094, PO.001). There was no significant distance effect for small mammal abundance (df=10,50, F=T.046, P=0.421, a=0.05). In 1995 (Fig. 18) there was no significant effect of the distance on the small mammal abundance (df=18,90, F=0.825, P=0.667, o=0.05). The site effect was found significant (d£=5,90, F=10.116, PO.001, cx=0.05). Gradient and species Normalized small mammal abundance frequency was calculated for all of the species and genera captured. P. maniculatus dominated all habitats for both trapping years. Sorex spp. were found dominant, after the deer mice, in most of the clearcut areas (1994 and 1995). Microtus spp. reached its highest normalized abundance within old-growth habitat for both trapping years and within buffer strip habitat (1994). Normalized abundance frequencies for this genus were higher within the non-buffered clearcut areas than within the buffered ones. If a riparian buffer was present, the frequency was higher within the buffer strip than within the adjacent clearcut area Z. princeps dominated the old-growth and buffer strip areas after P. maniculatus in 1995 (Fig. 20) while its highest abundance frequencies were found within all clearcut areas in 1994 (Fig. 19). The normalized abundance of the species was higher within buffered clearcuts than within the non-buffered ones for both trapping years. C. gapperi 54 (3 O CM O O CM U-D m o a CM a o TJ-CD 01 c «-.2 f= 8 £ £ co 8 ' C 3 <* Q. C CO .2 O I E C CO "d 10 CO CD CD £ O ** CO p M — CO