Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Influence of large-scale food supplementation on diversity of rodent communities Kohler, Chris Mark 1993

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

Item Metadata


831-ubc_1994-0048.pdf [ 1.12MB ]
JSON: 831-1.0075240.json
JSON-LD: 831-1.0075240-ld.json
RDF/XML (Pretty): 831-1.0075240-rdf.xml
RDF/JSON: 831-1.0075240-rdf.json
Turtle: 831-1.0075240-turtle.txt
N-Triples: 831-1.0075240-rdf-ntriples.txt
Original Record: 831-1.0075240-source.json
Full Text

Full Text

INFLUENCE OF LARGE-SCALE FOOD SUPPLEMENTATION ON DIVERSITY OF RODENT COMMUNITIES by CHRIS MARK KOHLER B.Sc, The University of New Brunswick, 1987  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN FORESTRY  in THE FACULTY OF GRADUATE STUDIES (Department of Forest Sciences)  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA OCTOBER, 1993  © Chris Mark Kohier, 1993  In  presenting  this thesis  Department of  c’es+  in  partial  fulfilment  of the  requirements  for an  advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  ScLeiAeS  The University of British Columbia Vancouver, Canada  Date OcevJ  DE-6 (2/88)  Il  11  ABSTRACT  Rodents were monitored bi-weekly to determine seasonal responses of rodent diversity to large scale supplemental feeding. Sunflower seeds were added by helicopter to thinned, fertilized, pole-sized lodgepole pine stands in the Montane Spruce ) 1 (MSdfl  -  Engelmann Spruce-Subalpine Fir (ESSFdC ) biogeoclimatic zone (transition 7  area) of southern interior British Columbia. My objectives were: 1) to determine the rate of seed consumption; 2) to determine the response of rodent communities to supplemental feeding; and 3) to evaluate methods for expressing species diversity. Forty-eight 1-rn 2 seed plots were used to estimate seed consumption rate. Consumption rates did not differ among stands, or between inside (where trap bait was present) and outside small mammal trap grids. Seed consumption rate increased as summer progressed, and was faster during the second year than the first year of seeding. The Simpson index indicated that diversity decreased in seeded declining evenness caused by increased deer mouse  areas,  as a result of  (Perornyscus inanicularus)  population density. Species richness was maintained on seeded areas, and hence log series alpha expressed higher diversity on one of the two seeded areas, when compared to its respective control. I reviewed literature to assess the merits of using the Shannon-Wiener and Simpson indices, Shannon-Wiener and Simpson evenness, and log series alpha. For my purposes, the Simpson index was the most useful for expressing changing dominance, and log series alpha was have low  sensitivity  most useful for  reflecting richness. Both indices  to sample size and are highly discriminatory. The Shannon  Wiener index and Simpson and Shannon-Wiener evenness were used for comparing results among other studies.  111  TABLE OF CONTENTS Page ABSTRACT  ii  .  LIST OF TABLES  v  LIST OF FIGURES  vii  ACKNOWLEDGMENTS  ix 1  GENERAL INTRODUCTION  CHAPTER 1. INFLUENCE OF LARGE-SCALE FOOD SUPPLEMENTATION ON DIVERSITY OF RODENT COMMUNITIES INTRODUCTION  2 2  Literature Review  2  Objectives and Hypotheses  4  METHODS AND MATERIALS  6  Description of Study Area  6  Experimental Design and Seed Distribution  6  Seed Monitoring  7  Rodent Populations and Diversity  8  RESULTS  10  Seed Consumption  10  Rodent Populations and Diversity Populations Pooled Relative Abundances and Diversity Bi-weekly Diversities  13 13 18 23  DISCUSSION  32  Experimental Design  32  Seed Consumption  32  iv Rodent Populations  .  33  Diversity of Rodent Communities Control A- Treatment B Comparison Control C- Treatment D Comparison  34 34 37  General Discussion  38  CONCLUSIONS AND RECOMMENDATIONS  43  CHAPTER 2. DIVERSITY: CRITIQUE AND FORMULA SELECTION  45  INTRODUCTION  45  BACKGROUND  45  ASSESSMENT OF SOME POPULAR DIVERSITY MEASURES  47  SPECIES ABUNDANCE MODELS  50  INDICES BASED ON THE PROPORTIONAL ABUNDANCE OF SPECIES  51  FOCUSING THE DIVERSITY CONCEPT  57  LITERATURE CITED  59  V  LIST OF TABLES Page Table 1. ANOVA results showing no significant difference in seed consumption rate among sunflower seed treatment grids B and D, or inside and outside small mammal grids. These data were collected on 15 May and 14 June 1992, in the MSd 2 biogeoclimatic zone 1 ESSFdC -  11  near Ewer Creek (Vernon B.C.)  Table 2. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are MNA) for rodents on control grid A. (CHIP PERO  =  M.LONG  Table 3.  deer mice; RBV =  =  redback voles; PHEN  long-tailed voles; ZAP  =  =  =  chipmunks;  heather voles; 14  jumping mice)  Bi-weekly population estimates (Jolly-Seber, except first and last*,  which are MNA) for rodents on treatment grid B. (CHIP PERO  =  M.LONG  deer mice; RBV =  =  redback voles; PHEN  =  chipmunks;  heather voles; 15  long-tailed voles)  Table 4. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are MNA) for rodents on control grid C. (CHIP PERO  =  M.LONG  deer mice; RBV =  =  redback voles; PHEN  long-tailed voles; ZAP  =  =  =  chipmunks;  heather voles;  jumping mice)  16  Table 5. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are MNA) for rodents on treatment grid D. (CHIP  chipmunks;  vi PERO  =  M.LONG  deer mice; RBV =  =  redback voles; PHEN  =  heather voles; 17  long-tailed voles)  Table 6. Effect of sunflower seed treatment on total number of animals (N), number of species (R), and evenness (E). These are average values for 1991, compared to those of 1992 (NC  no change)  41  vii  LIST OF FIGURES Page Figure 1. Rate of sunflower seed germination and consumption by animals. Data were collected during 1991 and 1992, in thinned, fertilized, 20-28 year old lodgepole pine stands in the MSdm ESSFdC, -  biogeoclimatic zone of south central interior British Columbia  12  Figure 2. Rank abundance plots for grid A control and grid B food supplemented communities. Summer data are pooled from mid-June to mid-August for 1991 (and late May to 20 July for 1992), fall data are pooled from midAugust to mid-October 1991 (and 20 July to mid-October for 1992)  19  Figure 3. Rank abundance plots for grid C control and grid D food supplemented communities. Summer data are pooled from mid-June to mid-August for 1991 (and late May to 20 July for 1992), fall data are pooled from mid August to mid-October (and 20 July to mid-October for 1992)  20  Figure 4. Diversity of rodents in treated and control grids. Pooled population data for summer and fall. Summer data are pooled from mid-June to mid August for 1991 (and late May to 20 July for 1992), fall data are pooled from mid-August to mid-October (and 20 July to mid-October for 1992).. 22  Figure 5. Simpson indices of diversity and evenness for the control A-treatment B comparison. Data were collected bi-weekly during 1991 and 1992  24  vi” Figure 6. Simpson indices of diversity and evenness for the control C-treatment D comparison. Data were collected bi-weekly during 1991 and 1992  26  Figure 7. Shannon-Wiener indices of diversity and evenness for the control Atreatment B comparison. Data were collected bi-weekly during 1991 and 1992  28  Figure 8. Shannon-Wiener indices of diversity and evenness for the control Ctreatment D comparison. Data were collected bi-weekly during 1991 and 1992  29  Figure 9. Log series alpha indices of diversity for both study replicates. Data were collected bi-weekly during 1991 and 1992. Numbers near data points represent richness  30  Figure 10. Hypothetical rank abundance plots illustrating the typical shape of four species abundance models: geometric series, log series, log normal and broken stick. Abundance of each species is plotted on a logarithmic scale against the species rank  48  Figure 11. K-dominance curves ranking species for grid C control and grid D food supplemented communities, on July 5, 1991, and July 19, 1992. Species are ranked from most to least dominant  55  ix  ACKNOWLEDGMENTS  I was encouraged when Dr. Alton Harestad, Dr. Peter Marshall and Dr. Hamish Kinirnins became my supervisory committee. I thank Dr. Tom Sullivan, my supervisor, for his patience and guidance on both personal and scientific levels. Thanks to Don Purdy from the B.C. Ministry of Forests, Vernon, for his flexibility on sunflower seeding dates. All help in the form of time, materials and equipment, greatly improved the quality of this field research. Funding was obtained through an NSERC operating grant to Dr. Sullivan, and several contracts from the B.C. Ministry of Forests Silviculture Section, Kamloops Region. There were contributions from a wide range of field assistants, and all of their help was greatly appreciated. The Bolton family opened their doors, providing an ideal base for my project at the Bolton Ranch which saved time and money. Not only were they very hospitable, but also their ranch provided us with countless, on-location, recreation and leisure opportunities. Special thanks to all Boltons, and especially Eleanore’ s cookies.  GENERAL INTRODUCTION  There has been increasing concern about the degradation of our global environment, and especially about the issue of decreasing biological diversity (biodiversity). Forestry practices which alter the environment are constantly under attack for various reasons, and they are blamed for decreasing biodiversity (Westman 1990, Kimmins 1992). This claim is not always true, and frequently is not supported by data. Accusations can result from a misinterpretation of statistics, because of the large variety of ways they can be used to describe diversity (Hunter 1990, Westman 1990, Kimmins 1992). This study was designed to explore the validity of diversity as a concept for guiding present and future changes in forest management practices. Chapter 1 shows how supplemental feeding, as a forest stand protection technique, affects small mammal diversity at the community level. Chapter 2 contains a review of conventional methods of calculating diversity indices. Included with this ‘critique and formula selection chapter?, are various opinions on the validity of these measures, and reasons for selection of the Simpson, Shannon-Wiener, and log-series alpha indices to summarize my data.  2  CHAPTER 1  INFLUENCE OF LARGE-SCALE FOOD SUPPLEMENTATION ON DIVERSITY OF RODENT COMMUNITIES  INTRODUCTION  Adding sunflower seed to the forest as a diversionary food is a relatively new silvicultural technique which reduces red squirrel (Tamiasciurus hurJsonicus) de-barking damage to juvenile lodgepole pine stands (Pinus conrorta var. iatfoiia) (Sullivan and Kienner 1993). However, little is known of its effects on mammalian species diversity. Past evidence indicates that enrichment of all parts of the resource spectrum will increase diversity, while an increase in only part of the resource spectrum will allow superior competitors to take advantage and exclude other species, thus decreasing diversity (MacArthur 1972). The present study was designed to determine the population density (abundance) response of each species in a small mammal community to this resource enrichment, and to express the data in the context of diversity.  LITERATURE REVIEW  Information on how supplemental food influences species diversity at the small mammal community level (large scale) is scarce (Szaro and Saiwasser 1989). Abramsky (1978) reported an increase in species diversity when supplemental food was  3 added to a small mammal  community.  This supports MacArthur (1972) who suggested  that: 1) an increase in the abundance of all resources caused by an overall increase in •production may lead to higher species diversity; and 2) an increase in only a part of the total resource spectrum may lead to a decrease in species diversity through dominance by those species that are competitively superior in exploiting the augmented resources. Abramsky (1978) holds that because he supplied “large amounts” of seed, a response in one group of organisms (i.e.  immigrating Dipodomys ordii) would not have  prevented a similar response in other groups by intertaxon competition. This, he suggests, supports MacArthur’s prediction #1 above. I do not agree, and offer that it actually rejects prediction #2. In this case, the “increase in part of the resource spectrum” resulting from addition of a non-native seed is an increase in resource diversity and spectrum in MacArthur’s (1972) model. Further complicating Abramsky’s interpretation is his experimental technique, whereby he added a single supplemental food plot after monitoring his control and increased production plots for 3 years. This is simple and temporal pseudoreplication (Hurlbert 1984). Whether an increase in a small part of the resource spectrum will decrease diversity should depend on: 1) size, behaviour and other interacting mechanisms of invading species (if any), and those already present and: 2) how their niche parameters overlap with those of other resident species. Sullivan (unpubi.) found no difference in diversity between control and treatment areas supplemented with 2 applications of sunflower seed at a rate of 22.7 kg/ha. A 2- to 3-fold increase in population density of terrestrial vertebrates is a typical outcome of small scale food supplementation experiments, which indicates that these populations are frequently limited by food supply (Boutin 1990). Food is often supplied on a spatial scale that is relevant to the question of habitat choice by individuals, but not to the questions of population regulation or community dynamics. Increased reproduction and survival usually occur in fed populations, but they do not  4 alter the pattern of population change. Food supplementation on a limited scale (1-2 ha) merely creates small ‘hot spots’ which have an increased production of animals. •  The ability to conduct large-scale (25 ha) experiments on food supplementation was made possible by an operational program using diversionary food to protect forest stands. Aerial application of food, in a uniform distribution over large areas for a twoyear period, was hypothesized to eliminate the population limitations caused by food supply. Based on this rationale, it was decided to investigate how rodent diversity was affected at this large-scale community level.  OBJECTIVES AND HYPOTHESES  The challenge was to determine how to describe changing dynamics of small mammal communities (order Rodentia) supplemented with a high energy food source, compared to dynamics of non-supplemented control communities. This was done by using conventional diversity measures and interpreting the meaning of the diversity statistics in the coitext of current silvicultural and small mammal community paradigms. There were three main objectives: 1) to determine if sunflower seeds were eaten, and, if so, what the consumption rate was and how this rate changed over a two-year period. This provided evidence that any change in diversity of rodents in the treatment area was probably due in part to sunflower seeds being eaten; 2) to transform rodent population data, obtained by live-trapping bi-weekly for two years, into diversity statistics; and 3) to discuss these data in the context of applied and theoretical science. Two main hypotheses were evaluated in the study: 1) that sunflower seed consumption was the same on both supplementally-fed areas. Seed plots were established outside and inside the trapping area (where trap bait was present). Therefore, it was necessary to test the hypothesis that seeds were consumed more  5 slowly where trap bait was present than where no bait was present; 2) and that over time, rodent diversity would increase on food supplementation areas compared to control areas where supplemental food was not added.  6  METHODS AND MATERIALS  DESCRIPTION OF STUDY AREA  Field work was conducted 25 km west north-west of Vernon in the south central interior of British Columbia, between the latitudes of 50° 21.5’ to 50° 22.8’ N, and longitudes of 119° 35.3’ to 119° 36.5’ W, near the northwest end of Okanagan Lake. Site elevation ranged from 1356 rn to 1478 m, in the MSd,T, ESSFdC 2 biogeoclimatic -  zone (transition area) (Meidinger and Pojar 1991). The climate is characterized by warm, dry summers and cold, dry winters with mean July and January temperatures of +16 and -10° C, respectively, and a mean annual precipitation of 40 cm. In mature stands, the predominant coniferous species are western red cedar (Thuja plicata), Douglas-fir (Pseudotsuga nienziesii), subalpine fir (Abies lasiocarpa), Engel mann spruce (Picea engeiniannii)  -  white spruce (P. glauca) hybrid, western larch (Larix  occidentalis), and lodgepole pine. Two lodgepole pine stands (age 21 and 28 years), approximately 36 ha each which had been thinned and fertilized, were selected for large-scale food supplementation. Two other thinned and fertilized stands of 17.8 and 38.8 ha, aged 28 and 20 years, respectively, were used for control replicates. The average d.b.h. of trees on control and treatment blocks ranged from 13.8 ± 0.2 cm to 16.8 ± 0.3 cm.  EXPERIMENTAL DESIGN AND SEED DISTRIBUTION  A randomized block design was used, with two control and two treatment areas. Rodent sampling grids of 4 ha were within the center of each area. All treatment areas were separated by at least 100 m to minimize animal movements between grid areas.  7 Treatment areas were selected on the basis of efficiency of helicopter seeding, rather than on the animal communities that were expected in each area. Treatment blocks had sunflower seed applied once a month from early June to October in 1991 (5 times), and May to October in 1992 (6 times). This provided a liberal amount of food to all rodent species living in the treatment areas. Sunflower seed was applied uniformly over the treatment areas at a rate of 20 kg/ha/month, using a helicopter carrying a cone shaped hopper designed to spread fertilizer or grass seed. Seed survival plots were established to test the hypothesis that sunflower seed was not being eaten at the same rate inside the trap grids as outside, or at the same rate among both supplementally fed treatment areas. Each treatment area 2 seed plots in a square arrangement, 24 of which were within the trap had 48 1-rn grids, the other 24 being in the surrounding seeded area. Seed plots within the trap grids, where seed consumption rate may have been affected by the presence of trap bait (oats and peanut butter), were considered ‘different’ than those plots outside the trap grids. Therefore, the four areas were compared by 2x2 factorial ANOVA, realizing that the in-out of trap grid comparisons were not true replicates because of lack of independence (i.e. they were only separated by 28.6 m).  SEED MONITORING  Six 1-rn 2 seed plots were spaced 28.6 rn apart on each of 8 lines. On each line, 3 plots were in, and 3 were  out of  the trap grid. All seeds landing in these plots during  aerial application were removed, and replaced with 15 seeds whose locations were . The intention was seeds/rn ) marked by toothpicks (aerial application approximated 15 2 to record rate of consumption, and number of seeds germinated on a bi-weekly basis. Three checks were completed during 1991, and 10 checks during 1992. Data stated as  8 ‘rate of consumption’ refers to number of seeds eaten from the date of the most recent seeding, to the sampling date indicated.  RODENT POPULATIONS AND DIVERSITY  In each replicate stand, 225 Longworth live-traps were located at 14.3-rn intervals in a checkerboard pattern on 16 lines (16 x 15). One trap was placed at each station for the duration of the project, so animals would become accustomed to their presence, thus keeping trappability high. Trapping was conducted from May to October during 1991 and 1992 on a bi-weekly schedule. At each trap-session, traps were set on the afternoon of day 1, checked on the morning and afternoon of day 2, and again on the morning of day 3. Traps were baited with oats, peanut butter mixed with sunflower seed oil and a slice of carrot. Cotton was provided for bedding. At each check, all untagged animals were ear tagged with numbered fish fingerling tags. Information on sex, reproductive condition, species, body weight (using Pesola spring scales), and tag number was recorded. This trapping technique provided data for calculating the change in population densities of each species over time, using the Jolly-Seber stochastic model (Seber  1982). Also, minimum number alive (MNA) estimates (Krebs 1966) were made for the first and last week of trapping. The Jolly-Seber estimates were used to infer abundances of each species for calculating diversity statistics. Bi-weekly population density estimates are presented, and pooled abundances are presented as relative abundance graphs. The four pooled periods are summer 1991, fall 1991, summer 1992, and fall 1992. For the control A- treatment B comparison, summer data are pooled from mid-June to mid-August for 1991 (late May to July 20 for 1992), and fall data are pooled from mid-August to mid-October (July 20 to mid October for 1992). The Simpson and Shannon-Wiener indices and log series alpha  9 (described in Chapter 2) were used to estimate diversities using first the pooled abundances, and then the un-pooled, bi-weekly abundances.  10  RESULTS  SEED CONSUMPTION  An application density of 20 kg/ha of sunflower seed is the standard quantity for operational application. Seed consumption will be described in detail, after an explanation as to why it was possible to pool seed consumption plot data from both treatment areas, and plots from inside and outside the small mammal trapping grids. ANOVA (2x2 factorial) was used to test for different consumption rates among both treatment areas, and inside and outside the small mammal trap grids. Visual inspection of the mean consumption rates from the four areas suggested there were no obvious differences, so only the data from May 15 and June 14 of 1992 were statistically analyzed. These data were collected one week following seed applications and are early enough in the season that less than half the applied seed had been eaten. ANOVA showed that there was no signitcant difference (  =  0.05) in consumption  rates among all four areas, at either date (Table 1). The consumption rate one week post-seeding in May 1992 was 6.06 kg/ha/week, which increased to 11.42 kg/ha/week one week after the June seeding (Figure 1). By June 5, 17.04 of the original 20 kg/ha had been eaten, leaving 2.96 kg/ha of uneaten or germinated seeds. Thus, after the June application, 22.96 kg/ha was present, of which 21.59 kg was consumed by July 3. It was not possible to get a 2-week post-seeding check in May, but June, July and August can be compared with 2- and 4-week checks. Their respective two-week rates of consumption were 17.80 (June) to 19.07 (July) to 19.84 kg/ha/2 weeks (August). This two week rate decreased to 19.14 between September 25 and October 12.  11 Table 1. ANOVA results showing no significant difference in seed consumption rate among sunflower seed treatment grids B and D, or inside and outside small mammal grids. These data were collected on 15 May and 14 June 1992, in the MSdfl ESSFdC 2 biogeoclimatic zone near Ewer Creek (Vernon) B.C. P = probability, . = 3.95). 9 () 005 (F -  a) May 15, 1991  Source of variation ALL PLOTS  DF  SS  MS  F  P  3  34.88  GRID (A)  1  2.04  2.04  0.08  0.72  IN/OUT (B)  1  0.17  0.17  0.01  0.93  Ax B  1  32.64  32.64  1.32  0.25  IN CELLS  92  2277.08  24.75  TOTAL  95  2311.96  DF  SS  MS  F  P  b) June 14, 1991  Source of variation ALL PLOTS  3  51.04  GRID (A)  1  12.04  12.04  0.50  0.51  IN/OUT (B)  1  37.50  37.50  1.55  0.22  AxB  1  1.50  1.50  0.06  0.80  IN CELLS  92  2232.58  24.27  TOTAL  95  2283.63  12  * -  kg/ha EATEN  kg/ha ADDED  kg/ha GERMINATED  kg/ha NOT EATEN  25  20 o 15  U  uJ  0  • 10 5  0 7 A  21  1991  1992  Figure 1. Rate of sunflower seed germination and consumption by animals. Data were collected during 1991 and 1992, in thinned, fertilized, 20-28 year old lodgepole pine stands in the MSdfl ESSFd, biogeoclimatic zone of south central interior British Columbia. -  13 The above consumption rates for 1992 can be compared to the 1991 rates. Note that the mid-July 1991 germination rate of 5.52 kg/ha/2 weeks (27.6%, or 41,400 seeds/ha/2 weeks) is substantially higher than the mid-July 1992 rate of 1.07 kg/ha/2 weeks (5.4%, or 1605 seeds/ha/2 weeks), probably because there was much more rain in 1991 than 1992. The germinated seeds were mostly consumed before rooting, while the others were observed to have been clipped by the end. of the four week post-seeding period (data for germinant survival were not recorded). This is evident by noting that all seeds, including germinants of July 26, 1991, were consumed by August 7, 2 weeks later (Figure 1). Comparing 1991 to 1992, the July 26 (1991) 2-week consumption rate of 13.39 kg/ha increased by 5.68 kg/ha to 19.07 kg/ha on July 19, 1992 (slightly earlier in the season). The seed was all consumed by 4 weeks post-seeding during both years. Again, on August 21, 1991, the two-week rate was 18.85 kg/ha, compared to 19.84 kg/ha on August 12, 1992. Germination rates on these dates were comparable at 0.56 (1991) and 0.66 kg/ha (1992). In summary, seeds were eaten at a faster rate as the season progressed, and faster in 1992 than 1991.  RODENT POPULATIONS AND DIVERSITY  Populations  To compute diversity indices on a bi-weekly interval, population densities of each species were estimated (Tables 2-5). This brief section presents absolute abundances of species on the study site (discussed in detail in Sullivan and Kohler [in prep.]). However, the relative abundances are more important for calculating and examining diversity.  14  Table 2. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are chipmunks; PERO = deer mice; RBV MNA) for rodents on control grid A. (CHIP = redback voles; PHEN = heather voles; M.LONG = long-tailed voles; ZAP = jumping mice).  WEEK  CHIP  PERO  RBV  PHEN  M.LONG  ZAP  YEAR 1991 *Jun07  30.0  7.0  10.0  0.0  0.0  1.0  Jun-21  37.7  26.0  12.0  1.0  0.0  2.0  Jul-05  30.9  37.2  10.7  0.0  0.0  2.0  Jul-19  39.9  36.4  17.0  1.0  0.0  3.0  Aug-02  40.2  31.9  14.4  1.0  0.0  0.0  Aug-16  36.2  41.9  15.0  1.0  1.0  0.0  Aug-30  34.5  37.1  12.8  3.0  0.0  0.0  Sep-13  33.8  38.4  9.5  1.0  0.0  0.0  Oct-li  76.8  36.2  16.0  1.0  0.0  0.0  May-22  32.5  26.1  27.0  2.0  0.0  Jun-05  27.1  29.5  8.4  1.0  0.0  0.0  Jun-19  27.1  39.9  13.6  0.0  0.0  0.0  Jun-29  29.2  51.0  8.8  0.0  0.0  0.0  Jul-13  33.5  56.0  16.7  0.0  0.0  0.0  Jul-27  39.1  63.8  7.3  1.0  1.0  0.0  Aug-10  37.9  53.0  8.7  0.0  9.0  0.0  Aug-24  34.1  58.6  9.0  0.0  1.0  0.0  Sep-07  21.9  46.2  12.0  0.0  0.0  0.0  Sep-22  12.0  46.3  21.7  0.0  0.0  0.0  Oct-06  7.0  44.7  11.0  0.0  0.0  0.0  *Qct.i9  1.0  58.0  8.0  2.0  YEAR 1992  0.0  —  0.0  0.0  15  Table 3. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are MNA) for rodents on treatment grid B. (CHIP = chipmunks; PERO = deer redback voles; PHEN = heather voles; M.LONG = long-tailed voles). mice; RBV  PHEN  M.LONG  16.0  0.0  0.0  15.5  18.0  2.0  0.0  18.8  29.0  20.4  0.0  0.0  Jul-19  48.2  53.0  23.9  1.0  0.0  Aug-02  48.6  69.4  21.1  0.0  0.0  Aug-16  23.1  90.7  22.1  0.0  0.0  Aug-30  50.5  103.2  27.4  0.0  0.0  Sep-13  36.4  100.6  32.5  0.0  0.0  Oct-11  24.1  115.5  57.5  0.0  0.0  May-22  23.4  60.0  22.9  1.0  0.0  Jun-05  46.6  106.8  16.6  0.0  0.0  Jun-19  19.0  132.6  20.2  0.0  0.0  Jun-29  19.2  169.5  19.0  0.0  0.0  Jul-13  32.2  177.6  19.7  0.0  0.0  Jul-27  44.3  206.9  13.3  0.0  1.0  Aug-10  46.0  171.6  12.6  0.0  1.0  Aug-24  31.4  176.7  22.6  0.0  1.0  Sep-07  12.0  167.5  34.9  1.0  0.0  Sep-22  0.0  163.9  44.9  0.0  0.0  Oct-06.  0.0  135.5  13.9  0.0  0.0  *Oct.19  0.0  41.0  8.0  0.0  0.0  CHIP  PERO  *Jun07  21.0  11.0  Jun-21  15.0  Jul-05  WEEK  RBV  YEAR 1991  YEAR 1992  16  Table 4. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are MNA) for rodents on control grid C. (CHIP = chipmunks; PERO = deer mice; RBV = redback voles; PHEN = heather voles; M.LONG = long-tailed voles; ZAP = jumping mice). WEEK YEAR 1991  CHIP  PERO  RBV  PHEN  M.LONG  ZAP  -  *May.22  26.0  11.0  4.0  1.0  0.0  0.0  Jun-07  33.6  18.2  8.0  2.0  0.0  0.0  Jun-21  53.2  48.9  10.0  3.0  0.0  0.0  Jul-05  50.4  63.9  11.0  2.0  0.0  0.0  Jul-19  53.4  66.3  15.8  1.0  0.0  1.0  Aug-02  63.3  82.5  20.9  2.0  0.0  0.0  Aug-16  62.3  84.6  15.0  1.0  0.0  0.0  Aug-30  57.9  71.1  9.6  1.0  0.0  0.0  Sep-13  61.1  68.3  13.1  2.0  0.0  0.0  Oct-11  67.5  74.3  33.0  0.0  0.0  0.0  May-22  56.8  33.8  28.9  0.0  0.0  0.0  Jun-05  60.6  33.0  10.8  1.0  0.0  0.0  Jun-19  61.0  51.6  16.4  2.0  0.0  0.0  Jun-29  59.6  62.5  19.8  1.0  0.0  0.0  Jul-13  51.6  70.4  13.9  5.0  0.0  0.0  Jul-27  67.6  93.9  16.9  0.0  0.0  0.0  Aug-10  36.0  81.1  12.0  0.0  0.0  0.0  Aug-24  33.4  64.1  18.0  0.0  0.0  0.0  Sep-07  21.4  51.0  13.7  0.0  0.0  0.0  Sep-22  7.0  54.5  19.0  0.0  1.0  0.0  Oct-06  2.0  45.9  26.2  1.0  2.0  0.0  *Octl9  2.0  29.0  24.0  3.0  2.0  0.0  YEAR 1992  17  Table 5. Bi-weekly population estimates (Jolly-Seber, except first and last*, which are MNA) for rodents on treatment grid D. (CHIP = chipmunks; PERO = deer mice; RBV = redback voles; PHEN = heather voles; M.LONG = long-tailed voles). WEEK YEAR 1991  CHIP  PERO  RBV  PHEN  M.LONG  -  *May22  9.0  6.0  5.0  0.0  0.0  Jun-07  17.0  11.0  5.0  4.0  0.0  Jun-21  19.0  18.9  4.0  15.0  0.0  Jul-05  21.5  39.5  6.3  22.0  0.0  Jul-19  89.3  41.5  6.6  16.5  0.0  Aug-02  63.5  52.2  14.1  17.0  0.0  Aug-16  47.3  65.8  10.2  12.0  0.0  Aug-30  63.7  89.0  7.5  8.0  0.0  Sep-13  85.8  105.2  7.2  10.0  0.0  Oct-11  84.3  102.9  5.3  4.0  0.0  May-22  29.9  38.3  9.3  9.0  2.0  Jun-05  22.4  58.2  12.8  19.2  2.0  Jun-19  21.5  79.8  8.0  16.5  4.0  Jun-29  48.0  113.2  27.5  15.8  1.0  Jul-13  30.9  143.1  10.7  10.9  1.0  Jul-27  161.8  208.7  15.0  13.0  1.0  Aug-10  60.4  185.0  27.5  12.0  1.0  Aug-24  30.0  177.9  15.0  3.0  1.0  Sep-07  12.1  221.7  3.0  3.0  1.0  Sep-22  6.0  179.4  2.0  3.0  1.0  Oct-06  0.0  109.8  4.0  2.0  1.0  *Qct19  0.0  38.0  2.0  2.0  1.0  YEAR 1992  18 The population data indicate that there was a large increase in abundance of deer mice on both treatment grids (Tables 3,5), relative to the controls (Tables 2,4). Deer mice were the most abundant species on all grids. Yellow pine chipmunks (Euranilas wnoenus) were, on average, the second most abundant species on treatment and control areas, but it was difficult to determine the extent to which they responded to supplemental food. The largest resident numbers of chipmunks were on control C (Table 4) and treatment B (Table 3). There were times when treatment D had some apparently unstable but high numbers of chipmunks (e.g. >  160/4 ha; July 27, 1992). Chipmunk abundance on treatment areas fluctuated much  more than that on control areas. On treatment D, northern redback voles (Clethrionomys gapped) and heather  voles (Phenaconiys intermedius) in creased in abundance, and long-tailed voles (Microlus longicauclus) were consistently present (Table 5). These less abundant species were also caught on the other 3 grids. The western jumping mouse (Zapus princeps) was the least abundant species, and was trapped on the control grids only (Tab]es 2, 4).  Pooled Relative Abundances and Diversity  Before discussing bi-weekly diversity, pooled relative abundance and diversity information is presented. In terms of relative abundances of each species, the data were averaged for the summer and fall of each trapping season, providing four successive graphs for each grid (Figures 2 and 3). The largest change in small mammal community structure was increasing dominance by deer mice in the food supplemented areas. The most noticeable change in the control areas was a switch from a community dominated by chipmunks in 1991 to one dominated by deer mice in 1992. The increase in deer mouse dominance in treatment areas was the cause of a  19 CONTROL A  0.4  0.4  z 0., 03  100_V  ZA.P  £101014  0.t  0.4  \  L)  z 0.4  z  0.2  :j r  L-i  ,..  PERO  RBV  0.4 £4 0s  0.4  03  £0_no  100_v  £REN  o.  ;•__tsç  -  0.6  0.4  02  01000  0_WV  CHIP  MI.ONG  P100_N  Figure 2. Rank abundance plots for grid A control and grid B food supplemented communities. Summer data are pooled from mid-June to mid-August for 1991 (and late May to 20 July for 1992), fall data are pooled from mid-August to mid-October 1991 (and 20 July to mid-October for 1992). (CHIP = chipmunks; PERO = deer mice; RBV = redback voles; PHEN heather voles; M.LONG = long-tailed voles; ZAP = jumping mice).  20 (‘n,,yronT  0.6  0,  PEEN  c-)  z  ZAP  0.6  0.6  04  0.4  03  0.2  0  z  -  = II  .  REV  PHVN  PERO  CHIP  PHEN  REV  03  0s  0,  0.2  :  PERO  CHIP  PHEN  I  REV  M.LONG  :L 0.4  0.  PPRO  COOl?  REV  PENN  M.LONG  Figure 3. Rank abundance plots for grid C control and grid D food supplemented communities. Summer data are pooled from mid-June to mid-August for 1991 (and late May to 20 July for 1992), fall data are pooled from mid-August to mid-October (and 20 July to mid-October for 1992). (CHIP = chipmunks; PERO = deer mice; RBV = redback voles; PHEN = heather voles; M.LONG = long-tailed voles; ZAP = jumping mice).  21 decrease in evenness, which is indicated by the increasing concavity of the accompanying curves. In the controls during 1992, naturally increasing dominance by • deer mice had a similar but reduced effect on evenness. The two species (chipmunk and deer mouse) switched numerical positions in the community, having no effect on diversity computational results (Figures 2 and 3). The three species caught consistently on the study area were redback voles, deer mice and chipmunks. There were also three uncommon species appearing rarely to occasionally. The rarest, the western jumping mouse, appeared during the first four trap sessions only, on control grid A, and the fifth session on control grid C. The next rarest species after the jumping mouse was the long-tailed vole which in 1991 was caught only once, on control C (August 16). This species was more common in 1992, and most prevalent on treatment D. Heather voles were caught occasionally on all 4 grids during all four pooled periods (except summer 1991, treatment B). This population was especially prevalent on treatment grid D, where it became more dominant than redback voles during 1991 and summer 1992. Of the three rare species, absolute abundance of long-tailed and heather voles increased most on treatment D. Jumping mice were present only during the wet early summer of 1991. I used average relative abundances to calculate Simpson, Shannon-Wiener and log series alpha diversity indices (Figure 4). The Simpson and Shannon-Wiener indices indicated that diversity on the control grids was fairly constant over time. On treatment D, Simpson indicates decreasing diversity (due to increasing deer mouse abundance), whereas Shannon-Wiener shows an increase due to the summer 1992 addition of the rare long-tailed vole to the community. Immigration of this species is also reflected by the rise of log series alpha from 0.73 to 1 .00. On treatment grid B, this increase in Shannon-Wiener did not occur with the re-appearance of heather voles in summer 1992 or the new appearance of long-tailed voles in fall 1992. During summer 1992 for this grid, there were only 4 species, compared to 5 in treatment D, and the absolute  22  SHANNON-WIENER  o 1.2  LOG SERIES ALPHA 0.2  0 SUMMER 1991  C0NTA  I  FALL 1991 -±-TREATB  SUMMER 1992 -*-[  FALL 1992  TREATD  Figure 4. Diversity of rodents in treated and control grids. Pooled population data for summer and fall. Summer data are pooled from mid-June to mid-August for 1991 (and late May to 20 July for 1992), faIl data are pooled from mid-August to mid October (and 20 July to mid-October for 1992).  23 abundances of heather voles were much lower in treatment grid B than D. ShannonWiener and log series alpha, therefore, being sensitive to an increase in number of rare • species and their abundance, responded more positively for grid D in summer 1992. During faIl 1992, both treatment grids had 5 species which is reflected in the similarity of both the Shannon-Wiener and log series alpha indices for that time.  Bi-weekly Diversities  The general overview presented by pooling the data into four seasonal categories  is now followed by a more detailed examination of diversity, on a bi-weekly basis.  s index T Sinipson Results of Simpson’s indices of diversity and evenness indicated that treatment B was initially more diverse than control A (Figure 5a). On the June 7 pre-seeding trap session, Simpson’s index was 0.556 on control A and 0.659 on treatment B, for a difference of 0. 103 (not on graph). The only measurement available for this comparison was MNA, whereas the remainder of these two diversity curves are based strictly on Jolly-Seber estimates. MNA was used here to indicate the magnitude of change that would be expected after an operational application of supplemental food. The difference decreased to 0.029 four weeks later, after seeding, when control A diversity increased to 0.634. and treatment B increased to only 0.663. After this initial decrease in treatment-control difference following one seeding, the declining trend continued in the treatment relative to the control area, with successive seedings. Evenness was relatively stable in 1991 on control A until after August 30 (Figure 5b). Redhack voles then began decreasing, while the chipmunk population was still increasing, causing a decline in evenness and diversity. Evenness on treatment B began a downward trend after June 5, 1992, but an upward fluctuation was measured  24  3  CONTROL A  TREATMENT B  0.8  I  z0  0.6  0.4  0.2  0  I I I, I 2  I  t  I  I  2  2  7 21 5 19 2 16 30 13 27 11 0 J A J S  N  D  J  F  M  A  III  I I  R2  2  I  II 2  I  *  22 5 19 28 13 27 10 24 7 22 6 A 0 MJ S J 1992  1991  1.2  C’, C,,  1  z 8 . 0 z  7 21 5 19 2 16 30 13 27 11 S 0 J A J 1991  N  D  J  F  M  A  SUNFLOWER SEED  22 5 19 28 13 27 10 24 7 22 6 0 S J A MJ 1992  Figure 5. Simpson indices of diversity and evenness for the control A-treatment B comparison. Data were collected bi-weekly during 1991 and 1992.  25 on September 22, 1992, when chipmunks (probably hibernating) and heather voles were no longer captured. This left only 2 species, creating very low diversity, but relatively high evenness. The patterns of Simpson diversity and evenness on control C and treatment D, were similar to those on control A and treatment B (Figure 6). Trapping began two weeks earlier here than on grids A and B. A slight increase from May 22, 1991, (MNA data, not on graph) in treatment D diversity was evident after the first (operational level) seed application, but there was a greater increase during that period on control C. The June 7 control C diversity was 0.610 and treatment D was 0.689 for a difference of 0.079. Four weeks after the first seeding (July 5), thr ordering of sites had switched, because control C had become more diverse (0.688) than treatment D (0.588). There was a large decline in Simpson diversity on the treatment D grid by July 19, when the chipmunk population increased from 21 the previous trap session to 89, and the less common heather voles decreased from 22 to 16. Diversity returned to the control level by the following trap session as the chipmunk population dropped back to 63 and redback voles increased from 6.6 to 14.0. By October 11, 1991 however, the treatment area (0.543) was substantially lower than the control diversity (0.639), because deer mice increased their domination of the small mammal community. During two declines in the 1992 treatment diversity curve (June 19 and July 13), a large increase in deer mouse abundance and a concomitant decrease in redback vole abundance occurred relative to previous weeks. The largest declining trend in treatment D diversity occurred after August 10, when the relative abundance of 3 of the 5 species present decreased to one-half their previous level, but deer mouse abundance decreased only slightly.  26 CONTROL C  TREATMENT D  v  0.8 >•  z0  0.4  0.2  0  __.L.II  II liii  I  I  I  II  I  II II III_l_IL_]_  7 21 5 19 2 16 30 13 27 11 S 0 A J J 1991  N  0  J  F  M  A  22 5 19 28 13 27 10 24 7 22 6 0 S A J MJ 1992  7 21 5 19 2 16 30 13 27 11 0 S A J J  N  D  J  F  M  A  MJ  1  0.8  rJ  z z z0  0.6  0.4  0.2  0  19 28 13 27 10 24 7 22 6 0 S A J  Figure 6. Simpson indices of diversity and evenness for the control C-treatment D comparison. Data were collected bi-weekly during 1991 and 1992.  27  Shannon-Wiener The Shannon-Wiener function indicated an initial decrease in diversity for all grids, except treatment D after first seeding (Figures 7, 8), as did Simpson’s index. Although treatment B is more diverse than control A (Shannon-Wiener index) before seeding started, by July 5, 1991, it became lower for the duration of the study. The Shannon-Wiener index in both replicates is similar to that of the Simpson index. However, the control C-treatment D comparison was different during 1992, due to a consistent richness value of 5 species in treatment D for all but the last trap session, compared with only 3 to 4 species in control C (Figure 8). The Shannon-Wiener index was sensitive to this richness difference during the first half of the 1992 season. This was also reflected in the pooled relative abundance graphs during this time, where both Shannon-Wiener and log series alpha were higher for treatment D (Figure 3). By August 4, the absolute abundances of chipmunks, redback and heather voles relative to deer mice on treatment D were so low that Shannon-Wiener diversity also declined.  Log series alpha The difference in richness between control C and treatment D is evident when expressed by log series alpha. Log series alpha fluctuates greatly and synchronously for control A and treatment B in early 1991 due to coincidental richness changes occurring on each grid (Figure 9a). Richness (R) for the first 4 trap essions changed in a sequence of 5, 4, 5, 4 on control A and 4, 3, 4, 3 on treatment B. These fluctuations show the very strong dependence of log series alpha on the number of species. Both the treatment and control replicates had a generally decreasing trend in log series alpha until mid-August 1991, with the treatments decreasing more rapidly than the controls.  28 CONTROL A  TREATMENT B  -  2  rd  1.5  z0 z z  1  0.5  0 1992  1991  1.2  zi >08  z  0  Z0.6  ri 0.4 0.2  0  7 21 5 19 2 16 30 13 27 11 0 J S J A 1991  N  D  J  22 5 19 28 13 27 10 24 7 22 6 0 J A S 1992 SUNFLOWER SEED F  M  A  MJ  Figure 7. Shannon-Wiener indices of diversity and evenness for the control A treatment B comparison. Data were collected hi-weekly during 1991 and 1992.  29 C  CONTROL C  TREATMENT D  2  1.S  ‘-.4  z  1  0  0.5  0 7 21 5 19 2 16 30 13 27 11 A J J 0 S 1991  N  D  J  F  M  A  22 5 19 28 13 27 10 24 7 22 6 0 A S J MJ 1992  1  0.8  z >0.6  0 0A U)  0.2  0 7 21 5 19 2 16 30 13 27 11 S 0 J A J  1991  N  D  22 5 19 28 13 27 10 24 7 22 6 0 S A J MJ 1992 SUNFLOWER SEED  J  F  M  A  Figure 8. Shannon-Wiener indices of diversity and evenness for the control Ctreatment D comparison. Data were collected bi-weekly during 1991 and 1992.  30  1.4  1.2  Cl,  0.8  0.6  0.4  0.2  0 1992  1991  1.4  1.2  Cl)  0.8  0.2  7 21 5 19 2 16 30 13 27 11 ON S A J J  1991  0  J  F  M  A  SUNFLOWER SEED  M  J 1992  Figure 9. Log series alpha indices of diversity for both study replicates. Data were collected bi-weekly during 1991 and 1992. Numbers near data points represent richness.  31 Treatment D starts decreasing in diversity at a faster rate than control C, after the first seeding. During 1992 however, treatment D had more species than control C, -  until October 6, and similarly, log series alpha values for treatment D were higher. This is especially noticeable on May 22 when treatment and control alphas were 1. 149 (R=5) and 0.558 (R=3), respectively. The curves for June 5 to July 13, 1992, reflect the rapidly increasing deer mouse population on treatment grid D as its diversity decreases slightly relative to the control (richness did not change for either grid). From July 27 to September 7, treatment richness was 5, while control C was 3, giving D higher diversity again. To a greater extent than in the A-B replicate, treatment D maintained more species than control C during 1992.  32 DISCUSSION  EXPERIMENTAL DESIGN  The precision of the diversity calculations may be considered low, because there are only 2 replicates. Accuracy may also be low, in the conventional sense of design, because there is only one sample unit (trap grid) per treatment area. This precluded randomization of units, which is necessary to insure accuracy. Increased accuracy obtainable from randomization can be imagined as sampling over a range of lodgepole pine stands across the landscape. This scale of project was not possible because of financial and logistical limitations. However, because the 4-ha grids were much larger than necessary to estimate small mammal abundance, accuracy and precision are effectively increased when compared with conventional 1-ha or smaller grids. Large grids also increase the precision of the measuring instruments (trap grids) because the sample will be more representative of relative abundances of species in the small mammal community.  SEED CONSUMPTION  This study was designed to show how rapidly sunflower seeds were consumed in the forest, and to investigate the response of small mammal diversity to this increase in one component of the resource spectrum. Seed plots were used to determine if seeds were consumed, and if so, how fast. The plots were not intended however, for determining what species consumed the seeds. Birds, insects, and mammals other than those caught in live-traps were also utilizing the added resource. Pine siskins  (Garduelis pint(s) and evening grosbeaks  (Goceorhrausres vespertina)  were observed  33 eating seeds, and seeds were observed in black bear (Ursus americanus) feces, and caches (likely red squirrel) excavated by bears. The general trend in both the 1991 and 1992 field seasons was that seeds were consumed at a faster rate following each successive application, and faster in 1992 than 1991. This trend is similar to that of increasing abundances of rodents. Rates should theoretically have declined on a negative exponential rate, as seeds were consumed and became harder to find. If there was a constant supply however, the consumption rate should have increased exponentially because of increasing populations of rodents as summer progressed, until the seasonal population decline in early winter. Four of the species present in the community (chipmunks, deer mice, redback and long-tailed voles) were held in captivity for approximately one month, and survived well on sunflower seeds, carrots and water. Chipmunks, deer mice and red squirrels have been shown to benefit (increased birth rates, survival and immigration) from sunflower seeds used as supplemental foods (Sullivan 1979, Sullivan er al. 1983, Sullivan 1990). Combining this evidence with the live trapping data indicated that all the species present probably benefited from the sunflower seed treatment. It is possible that some species such as deer mice, may specialize on sunflower seeds. This might allow the rarer herbivores such as redback, long-tailed and heather voles, to utilize natural plant products more efficiently because of reduced competition.  RODENT POPULATIONS  The reason why chipmunk population density fluctuated more on treatment areas than on control areas is not known, but it may be related to reduced trappability caused by the presence of sunflower seeds. In another study, provision of sunflower seed increased and maintained Town send ch ipnl un k (Eutamias townsendii) population density 50% above that of control areas in a coastal British Columbia forest (Sullivan er  34  al. 1983). Traps were baited with sunflower seeds which may have improved trappability in that study, but sunflower seeds were restricted to the treatment areas for my study, so high energy food was not available to control communities. If there were such large fluctuations in chipmunk density as indicated, the overall effect of sunflower seed on community structure was obviously different than if no fluctuations occurred. The following two factors may have caused the apparent fluctuations: 1) reduced trappability due to sunflower seeds on the study site, causing disinterest in the trap bait; 2) fluctuations in density caused by different immigration and emigration patterns on treatment, than on control areas. These patterns could have been affected by seed availability, and by an increase in the competitive red squirrel population, causing behavioral changes. Behaviour was likely altered by increased food availability. Alihough average litter sizes for redback, heather and long-tailed voles in the Kananaskis Valley, Alberta, were reported to be 5.2, 4.1, and 3.7, respectively (Norrie and Milr 1989), these species have not been shown to specialize on certain plant foods, suggesting they are flexible in what they eat. It was suggested that the litter sizes were related to differences in social organization and activity patterns, both of which could have been altered by sunflower seeding in my study, for mice, voles and chipmunks.  DIVERSITY OF RODENT COMMUNITIES  Control A- Treatment B Comparison  Although I showed that at least 20 kg/ha/month of sunflower seeds could be consumed, the hypothesis that the index of rodent diversity would increase in food supplemented areas was not supported. Pooled relative abundances, the Shannon Wiener and Simpson’s index, and log series alpha (except 1992, replicate C-D) all  35 indicate that diversity decreased in food-supplemented young lodgepole pine stands in the  MSdlfl  -  ESSFdC,  biogeoclimatic transition zone of B.C.  Simpson’s index is a Type II heterogeneity measure, being most affected by the dominant species. Deer mice were obviously the most successful in converting sunflower seed into increased reproduction (Sullivan and Kohler, in prep.), and thus were by far the most influential factor in causing a decline in the Simpson and Shannon-Wiener indices over time. The diversity curve for treatment grid B showed three lows which reflect this quality of Simpson’s index (Figure 5a). The first on August 16, 1991 occurred when chipmunk abundance decreased to about one-half of the previous trap session value (and thus inflating relative deer mouse abundance). Chipmunks were the second most abundant species at that time, and deer mouse abundance was low compared to 1992 values. In 1992, diversity hit a second low on June 29, when the deer mouse abundance was rising (then at 169 animals) and chipmunk abundance had decreased again. These observations demonstrate the effect of dominance on irnpson’s index. The converse effect on Simpson’s index is evident by looking at two examples where richness had more of an effect. The third low (October 6, 1992) was more attributable to richness which had decreased to 2, from 4 in the previous trap session (September 22). This was buffered by a slight increase in redback vole abundance, which dropped to its lowest level on October 6. On control A (July 13, 1992), Simpson’s index had a major increase from 0.562 to 0.604, which was mostly due to an increase in abundance of the rarest species (then redback voles) from 9 to 17 animals. This indicated that as the most dominant species became more comparable to the rare species abundance (lower density of deer mice on control A than treatment B), rare species abundance had more of an effect on Simpson’s index.  36 High evenness should not be considered high diversity, as shown by comparing the A-B replicate grids (Figure 5b). When one or two animals of a rare species are caught, which increases richness and likely diversity in comparison with the previous sampling period, a decrease in evenness is expressed. Good examples of this occurred on control A from trap session 1 to 6 where there were weekly fluctuations of richness which caused the diversity curve to go up when R increased and down when it decreased. The evenness curve A fluctuated in the opposite manner. For treatment B, the same opposite fluctuations occurred for the first four trap sessions. This effect was most dramatic when abundances of all species were relatively low. Another good example on control A was when R increased from 3 to 5 between July 13 and 27, 1992. Here, one redback vole and one heather vole each were caught. Diversity decreased here for both Simpson and Shannon-Wiener although ShannonWiener is a richness measure (Figures 5, 7). However, evenness decreased much more than the diversity indices. These observations of simultaneously increasing diversity and decreasing evenness contradict the underlying theory of Simpson (1949). An increase in diversity theoretically should be caused by an increase in evenness. For this and reasons given in Chapter 2, evenness will not be considered as a reliable diversity index for interpretation of my data. The trends for Simpson and Shannon-Wiener diversity followed the same sequence, but there were noticeable differences when R changes (Figures 7, 8). If R increases, the accompanying effect on diversity will be slightly more positive with the Type I Shannon-Wiener index, which is sensitive to rare species (relative to Simpson’s index).  37  Control C- Treatment D Comparison  The control C-treatment D replicate followed a trend of decreasing treatment diversity (Simpson) through the 1991 and 1992 trapping. The main difference in this replicate, compared to the A-B replicate, is that the rare species persisted much better in the treatment than in all other grids. There was considerable variation, however, which likely was partly caused by fluctuations in evenness created by non-synchronized litters among species, by immigration, and by inconsistent appearance of rare species. During 1992, for example, the decrease in treatment D diversity was buffered on June 28, July 27, and August 10 by high chipmunk, redback and heather vole abundances, and consistent appearances of long-tailed voles. Simpson and ShannonWiener indices on treatment D were similar to control C during these weeks, even with treatment area deer mice increasing to 206 to 208 anirnals!4 ha by July 27. The early 1992 deer mouse abundance was lower, and chipmunk, heather vole and long-tailed vole abundances were much higher, on treatment D than B. This was the main factor contributing to higher Simpson, Shannon-Wiener, and log series alpha indices for treatment D compared to control C, for at least part of summer 1992. It was not until abundances of voles and chipmunks decreased greatly on treatment D (September 7) that diversity also decreased substantially. It is not known what benefits the rare species on grid D may have obtained from sunflower seeds. Similarly, I do not know if the rare jumping mice appearing on the control grids in 1991 would have persisted during 1992 had sunflower seed been present there. With respect to treatment D, a wet grassy area dominated by aspen (Populus rreniuloides) was where many heather and long-tailed voles were caught.  Possibly this interspersion of habitats provided a refuge for the rare species, from the more seed-oriented deer mice (T. Sullivan, pers. comm.). It could also be that the  38 added food was more preferable than that naturally present, and in this habitat allowed the rare species to benefit. The rare species likely benefited from the supplemental food as indicated by Shannon-Wiener and especially the log series alpha indices. Treatment D diversity (log series alpha) was higher or comparable to control C through the 1992 season until August 24, when heather voles decreased dramatically (12 to 3) and chipmunks and redbacks also decreased. A similar occurrence 6 weeks earlier (July 13) also caused a decrease in the Shannon-Wiener index. Increased prevalence of heather and long-tailed voles in treatment D had an important influence on richness-based indices, which log series alpha expresses much better than Shannon-Wiener. Diversity is higher on the treatment than control at the start of 1992, and stays higher until September 22, when all rare species except jumping mice reappeared on control C, and October 6 when heather and long-tailed voles were still present on the control. On October 6, chipmunks were not caught on the treatment area, but two were caught on the control.  GENERAL DISCUSSION  An additional hypothesis to be tested was that diversity would return to normal shortly after termination of seeding, with a time lag which was to be identified. The field work was terminated before this could be tested, but at another food supplemented site in interior British Columbia (ICH 11 biogeoclimatic zone), 2 of 7 species recorded as significantly increasing in abundance by Sullivan (unpubi.) and Sullivan and Klenner (1993)(red squirrel and meadow voles  [Micrortis pennsyivamcus])  returned to control  levels by August. Sunflower seed was applied at a density of 22.7 kg/ha on four treatment blocks (2 manual and 2 by helicopter), once in each of May and June. This  39 suggests that diversity on my study site would likely return to control levels soon after termination of seeding. The response of the present rodent communities to supplemental food is clear and would have been even more so had trappability been higher in the treatment areas. Changes to diversity however, are less clear, because of the dichotomy between richness and dominance based measures apparent with these data. Less clear are comparative diversities at times when treatment-control pairs were near the same level of diversity. Even if a statistical comparison was used for these periods, another index could likely be used to contradict the conclusion made. The best example was the late 1992 low (and decreasing) treatment area diversities as indicated by Simpson and Shannon-Wiener, but higher diversities as indicated by log series alpha. Among the three indices used, concordance of rankings are most highly correlated between Shannon-Wiener and log series alpha (P < 0.05), but are not significantly correlated between log series alpha and Simpson, or Simpson and Shannon-Wiener (Magurran 1988). In many cases, the Simpson and Shannon-Wiener indices will give inconsistent ranking of diversity among grids (Spearman rank correlation coefficient; Magurran 1988). This occurred with my data, as ShannonWiener and log series alpha expressed treatment D as being more diverse than did the Simpson index during early 1992. By using more than one measure, emphasis can be put on either more species, or a more even distribution of abundances of species, depending on which is regarded as having greater diversity. For my study, log series alpha could have been chosen as the sole indicator in attempts to argue that supplemental food increases rodent diversity. Likewise, Simpson could have been chosen to argue the contrary. Still another measure could have been chosen, such as beta, which expresses greater difference among communities as being more diverse. This would make the treatment more diverse than the control since there are greater differences among the entities (speèies populations) used in the diversity  40 formulae. Beta diversity contrasts differences among habitats, and would express sunflower seeding as increasing diversity, because differences among forest stands are increased across the landscape. By choosing to interpret my data in the context of diversity, value judgments were necessary to determine whether seeding improves or degrades the small mammal community. The reality is that evenness decreased because the more common species responded most positively (Table 6). This supports MacArthur’s (1972) prediction that ‘an increase in only part of the resource spectrum may lead to a decrease in species diversity through dominance by those species that are competitively superior in exploiting the augmented resources’. However, no rare species were excluded by these competitively superior mammals, as predicted by MacArthur’s (1972) model. This indicates that the rare species may have benefited in some way (Table 6). This could have been from decreased pressure on natural foods by the common species, or from the rare species switching to sunflower seeds. The high deer mouse population density resulting from this large scale supplementation of sunflower seed (far above operational stand protection levels; see Sullivan and Klenner 1993), has occurred elsewhere in nature. Densities of 55 deer mice/ha (MNA; Sullivan and Sullivan 1982) were found where seeds were released following harvesting of lodgepole pine stands, and should similarly be found after natural fires (depending on stand age and fire severity) which release large quantities of lodgepole pine seed. Before deciding whether these results indicate that sunflower seeding is a good or bad practice, the treatment must be put into perspective. First, seeding is a technique which benefits both trees (reduces squirrel debarking damage) and animals, so it is a good integrated forestry practice. Second, operational seeding occurs only once/year, in early to mid-May, depending on when the cambium becomes active, so population effects will be less than that observed for my research. Third, a whole array of species  41 Table 6. Effect of sunflower seed treatment on total number of animals (NUM), number of species (RICH), and evenness (EVEN). These are average values for 1991, compared to those of 1992 (NC = no change).  NUM  CONTROL GRIDS  TREATMENT GRIDS  A  B  f 1%  C t 8%  RICH EVEN  4.4 to 3.7  4.0 to 3.6  NC  NC  D  f 50%  t 70%  t  t  3.3 to 3.6  4.0 to 4.9  42 are affected which were not caught in Longworth traps, including birds, fungi, insects, rodents, and large herbivores which eat the germinated sunflowers as they develop into plants. The research area was a multiple land use area which included logging, recreation and agriculture. Cows grazing in the area were observed feeding on sunflower plants growing at the helicopter landing where some seeds were left behind. Another group of animals of importance are the predators. Intuitively, it seems likely that with a greatly increased prey base (total number of animals, Table 6), predators would benefit. Long-tailed weasels (Musrelafrenata) and short-tailed weasels  (Musrela  ermina)  were caught and tagged during field work, but few recaptures  occurred on treatment or control areas, so population estimates could not be made.  43 CONCLUSIONS AND RECOMMENDATIONS  All the species caught on the study site (except heather voles) were held in captivity for approximately I month, and survived on water, carrots, and sunflower seeds. This showed that they would eat the supplemental food. Seed plots showed that the animal community could consume close to 20 kg/ha in one month in early summer, and at least 20 kg/ha in 2 weeks or less, by late summer and early fall. There were two important facts evident from the Jolly-Seber population  estimates. One was that in lodgepole pine stands with supplemental food, deer mice and chipmunk population densities increased the most. However, they also were the two most abundant species naturally (without sunflower seed). The other fact, was that rare species were not excluded from the treatment areas, indicating that they may have benefited from supplemental food. Evidence from other studies indicate that these community alterations would have returned to normal within a few months postseeding. I suspect that with 6 months of feeding, there may be a residual effect the following year caused by improved over winter survival and reproduction. Increases in the absolute abundances of deer mice and chipmunks produced largely skewed relative abundances, causing decreasing diversity when indices incorporating relative abundances of species were considered. The maximum population density occurring for deer mice, has also been found in habitats with no supplemental food. My data suggest that rare species were not excluded by the more common species, but that all species benefited. Seeding creates forest stands with different small mammal community structures across the landscape, possibly increasing beta diversity. These changes should rebound to normal relatively quickly after cessation of seeding.  44 Thus, it seems likely that seeding can safely be used at well above the 1 month/year operational level without a detrimental effect on small mammal populations. Sunflower seed as a supplemental food is likely very beneficial to the animal community, because rodents are the prey base for carnivorous animals. However, carnivores have large home ranges, and hence the effect on them may be much less than that on the local population of rodents. Sampling to include carnivores may be the next step in determining the full effect of sunflower seeding on the ecosystem. With this larger scale investigation, diversity across the landscape (beta) could also be examined.  45  CHAPTER 2 DIVERSITY: CRITIQUE AND FORMULA SELECTION  INTRODUCTION  Browsing through an ecology textbook will reveal techniques for measuring single entity attributes of an ecological community such as population (of a single pre  defined taxocene), and for measuring community parameters which involve more than one entity such as the concepts of niche and diversity. The fact that some diversity measures are a function of number of species present (richness), and evenness with which the individuals are distributed among these species, contributes only partly to the present confusion and debate concerning the validity of diversity as a concept (Huribert 1971, Southwood 1978, Wood 1993), and its measurement (Magurran 1988). Calculating and comparing diversities seems initially like a simple task, but upon close investigation, conclusions become less clear. BACKGROUND  One fact that makes comparisons difficult is that, for some formulae, two varying components are sometimes involved, richness and relative abundance (relative abundance is the same as evenness; and hence a source of confusion) (Magurran 1988). Magurran (1988) identifies richness as the number of taxonornic units in the area of concern. Relative abundance, in the case of species, refers to the absolute number of each species relative to the absolute number of each of the other species in a given  46 community. These two variables change in ways that are not consistently correlated, so when they are combined it becomes difficult to interpret the relevance of the measures. Huribert (1971) points out that the use, by some ecologists, of the separate parameters of richness or evenness to indicate diversity (and other parameters on which he elaborates), has contributed to criticism of diversity as a concept to the extent that he calls species diversity a non-concept. However, Hill (1973) recommended that diversity indices be used if carefully defined in the context of the data being analyzed The second fact that leads to confusion is that diversity measures are value judgements, on which two people are unlikely to agree perfectly. For example, the reality of having two choices (richness and evenness) with variable weighting depending on the formula used, results in a value judgment. Similarly for evenness, there is a more complex, hidden value choice. This problem is discussed by Wood (1993), and relates to the truth that diversity is measured to attempt to determine the number or degree of differences among biological entities (Wood 1993). He explains how degrees of difference can be a function of  point of view,  and that evenness fits this  category. By convention, maximum evenness is considered most diverse, because the probability of interspecific encounters is high (Simpson 1949). Maxinnim evenness is generally not observed in natural systems. There is usually a hierarchy where some species are more or less dominant, with the range (differences) being dependent on factors such as geographical scale of data collection and successional stage. A perfectly even community (all species with the same abundance) has no difference among its entities, so it could be considered least diverse, if greater difference among entities is chosen to be most important (Wood 1993). Considering a perfectly even community, for example, there is:  1) no diversity in actual abundances (for they are identical); and  47  2) maximum diversity in terms of the perception of the observer in the field (Wood 1993).  Here, there is a value judgment placed on evenness as a measure of diversity, because evenness was chosen to form the underlying theory of most diversity indices. This problem is considered in the interpretation of the results of my study. Magurran (1988) disagrees with this criticism, but she also refers to Lloyd and Ghelardi’s (1964) suggestion to compare empirical data with the broken stick model (Figure 10), which represents the most even state of affairs found in nature. Wood (1993) emphasizes that most ecologists do not claim the entities (species, genes, ecosystems etc.) as the diversity, but rather the variety among them is the diversity, and that number and frequency of the attributes are key components of the variety. This confusion, arising from a number of sources, surrounding the concept of diversity, has led many to believe that biodiversity may not be reasonable objective in natural resources management. It is not my intent to discuss this part of the issue which is explained elsewhere (Magurran 1988, Hunter 1990, Burton et a!. 1992, Kimmins 1992). ASSESSMENT OF SOME POPULAR DIVERSITY MEASURES  There are prefixes used which help describe the geographical scale at which measurements of diversity are being made. They are from large to small scale respectively, landscape, beta (across ecosystems), and alpha (within  ecosystem)  1 zone of British Columbia, and is classified diversity. My study was in the lDF-MS as alpha diversity (which is easily confused with the diversity statistic used in Chapter 1, called log series alpha). There is also temporal diversity which retrs to the change  48  100  roken stick 1.  I 0 0 C 0 C  0.1  normal  I I  ‘  0  S  I  S  S  S  \ log  series  0.01  I  geometric series  0.001 species sequence  Figure 10. Hypothetical rank abundance plots illustrating the typical shape of four species abundance models: geometric series, log series, log normal and broken stick. Abundance of each species is plotted on a logarithmic scale against the species rank in order from the most to least abundant species (Magurran 1988).  49 in diversity with respect to natural succession over time (Kimmins 1992). 1 suggest there is another form of temporal diversity which is seasonal. This should be obvious in harsher seasonal environments. For example in this project, density of deer mice continually increases as early spring progresses to fall, then winter mortality occurs, so diversity has seasonal fluctuations within the process of successional change. For example, differences in diversity statistics for food supplemented and control areas both show change over time, such as seasonal temporal diversity. But the statistics also can be used to show how the actual difference between areas changes over two seasons. Whether the comparison changes positively or negatively, is subject to opinion for the above reasons. Wood’s (1993) analysis suggests that what is measured as being less diverse in the sunflower treated areas over time, could either be considered more diverse because there are larger differences (in relative abundances) within the community after seeding, and between the treatment and control areas, or less diverse as evenness decreases. This becomes more clear upon considering the results and discussion sections of Chapter 1. Confusion of different evenness/richness weightings can be avoided by using just N (number of animals), but this ignores very real differences in abundance between species (deer mice and heather voles for example), and also ignores the basic difference of number of species. If number of species (R, same as S; another source of confusion; species density’), was chosen to interpret the present data, Wood’s (1993) measurement would be S-i because there are S-i differences between the entities. This use of only S similarly eliminates ambiguity caused by combining S and relative abundance. This brings up another source of conflict that Wood (1993) and  Bunnell(1990) elaborate on concerning species as an entity. The animals caught 1 Typically species counts are called richness, unless specific areas are censused, which was done for this study, in which case it is called species density (Hurlhert 1971). To avoid further confusion however, the results of species counts in this study will be referred to as richness (R).  50 during field work are identifiable to non-interbreeding species, and are assumed to be unambiguous entities. The Margalef index (Djg=(S-l)/lnN) and Menhinick index (DMfl=SIVN) use S and N, and these parameters are both used in the log series alpha formula, which was used for my analysis (see Magurran [1988] for details of these and other formulae; a computer program for calculating log series alpha is given in Krebs 1989). Log series alpha involves more complex calculation and is more discriminatory (identifies subtle differences better) than the above two measurements. The Margalef index uses S  -  1  which is consistent with Wood’s (1993) expression of differences. Any index such as log series alpha that uses S and N, but not relative proportions, has the problem of masking shifts in relative abundance (if present) when S and N do not change. There are very little temporal data in diversity studies, and as Magurran (1988) points out, having the same N and S values seldom occurs with real data. In my data there are large differences in N among grids, and within grids, over the two-week intervals of measurement. SPECIES ABUNDANCE MODELS  Species abundance models are theoretical species abundance distributions, to which empirical data can be tested for the appropriate ‘fit’. Magurran (1988) states that “a species abundance distribution utilizes all the information gathered in a community and is the most complete mathematical description of the data”. It is useful to learn how empirical data relate to these models, because the models provide a basis for discussing relevant theories such as the type of community competition present. For example, the geometric series (Figure 10) representing a community having only a few very dominant species, is a condition of pre-emption, where a few species have pre-empted a large portion of the niche hyperspace. Results of my field study on  51 sunflower seeding indicated that the rodent community gradually progresses toward this state over a two-year period, although the number of species is not reduced. There are two main reasons why I have chosen not to use the models. One is that the difficult computations require complex computer analyses, and this would have to be done repeatedly over time to identify when the community changed in underlying theoretical distribution. Secondly, it is difficult to justify conforming empirical data to a mathematical or biological theoretical distribution. As Southwood (1978) states: “for the ecologist, the ability of a parameter or index to discriminate between changed conditions may be more relevant than, and is not identical to, the precision of the ‘fit’ of the underlying model”. Similarly, “statistical models in general seem unhelpful, partly on theoretical grounds in that they test no biological criterion, but mainly on practical grounds since rarely do biological data seem to fit statistical models to any reasonable degree and often real data can be fitted to a variety of mathematical curves equally badly” (Lambshead and Platt 1984). Other authors have made similar criticisms, that relative abundance data often fit more than one underlying theoretical curve. The  Q statistic may have been a good measure for the present study because it  measures the interquartile slope of the species abundance curve, and thus does not give weighting to either the rare or dominant species (Magurran 1988). However, during my study no more than 7 species were counted at any one census, so eliminating the upper and lower quartiles (50% of the species) would not leave sufficient data. INDICES BASED ON THE PROPORTIONAL ABUNDANCE OF SPECIES  These indices can be broken into two groups, information statistics (Type I) and dominance indices (Type II), and one has been chosen from each group for analyzing my data. Magurran (1988) recommends using one of these indices in combination with  52 an index which is not based on proportional abundances, such as log series alpha. For using the first group of indices (of which the Shannon-Wiener function is used for this study), one assumption is that the information being tested is analogous to information in a code or message, and answers the question ‘how difficult would it be to predict correctly the species of the next individual collected?’. Shannon-Wiener (H’) is calculated as follows: S  H’= ‘  I  -  p (In p)  where s = number of species, and the quantity p is the proportion of individuals found in the ith species. And evenness is calculated as: E = H’/Hiiax  where H.. is the maximum possible evenness, given S and N.  Two further assumptions of Shannon—Wiener are that individuals are randomly sampled from an infinitely large population (Pielou 1975), and that all species in the community of interest are represented in the sample. BrilloLlin’s index produces similar results to Shannon-Wiener and is recommended if all the species are not known (Pie]ou 1966), and if every individual is censused without replacement (Krebs 1989). Neither of these two conditions applies to my data. Krebs (1989), recognizing Hurlbert’s (1971) and Washington’s (1984) criticisms of Shannon-Wiener, recommends using the index on empirical rather than theoretical grounds. Krebs (1989) also cites Taylor et al. (1976) as stating log series alpha was better than the other indices that he tested, because it reduced the variance in his replicates. Taylor (1984) also found that in testing eight diversity measures on his Rothamsted Insect Survey data, log series alpha was the best discriminator, the next best being Shannon-Wiener. Peet (1974)  53 recommends N 1 =  eH  (a transformation of Shannon-Wiener) as the best Type I  heterogeneity measure. It has been shown that estimates of Shannon-Wiener from field data are biased (Adams and M’Cune [1979] in Krebs 1989), but because in my study comparisons were made among treatments and controls, the bias may be similar in each. The Berger-Parker index (N,ax/N), where Niax is the abundance of the most dominant species and N is the abundance of all species combined, is also subject to bias caused by fluctuations in abundance of the most common species. It would express increasing deer mouse dominance (as occurred in my study) very well, but it is important to also be aware of the number of species present (Magurran 1988, Burton et  al. 1992). Simpson’s index was the third measurement used for this study for the above reasons, and it also has a higher discriminatory power than the Berger-Parker index. It is calculated as follows:  D  =  2 1 p  where D is diversity, and p 1 is proportion of the ith species. Evenness is calculated as: E = D/D, where Dimix is the maximum possible evenness, given S and N.  The reciprocal of Simpson’s index (l/D, used for my analysis) theoretically gives “the probability that two individuals chosen at random and independently from the population will be found to belong to the same group” (Simpson 1949). From this  54 point of view,  lID could be of value as an indicator of the probability of interspecific  encounters.  May (1975) has shown that the underlying species abundance distribution (Figure 10) strongly affects the outcome for Simpson’s index, if more than ten species are collected in the sample. That would be important for this study (but there are < 10 species) because the non-seeded control areas approach the log normal or broken stick distributions (Figure 10). The sunflower seed treated areas start similarly, but progress toward the geometric series. Therefore, the treatment area calculations of Simpson’s diversity would be less influenced by richness as time progressed, which would bias results toward dominance. There is bias inherent in Simpson’s index (toward evenness as a ‘human value’) because the formula is known as a dominance measure, and hence is sensitive to the most dominant species (e.g., deer mice in my study). Because two of the formulae selected for use here, Shannon-Wiener and Simpson, are biased towards richness and dominance, respectively, there is a good chance that they could give inconsistent ordering of communities. Constructing K-dominance curves will detect if this error will occur. A K-dominance curve shows percentage cumulative ranked abundances by species, where K is the species rank on the X axis. If the curves for the areas being compared intersect, there is a high likelihood of nonconcordance of site ranking when using Shannon-Wiener and Simpson indices (Platt et al. 1984). There are many cases where non-concordance of site ranking occurs for the present data (although for reasons already given and to follow, Simpson and Shannon Wiener were both used in Chapter 1). For example, trapping results from July 5, 1991, indicated the community structures of control C and treatment D grids were such that the K-dominance curves did not intersect (Figure 1 la). In this case, the Simpson and Shannon-Wiener indices rank the two grids concordantly, indicating control C is lower in diversity than treatment D (Figures 6a, 8a). Non-concordance of rankings  55  k-dominance curves a)  percent cumulative abundance  1  2  4  3  b)  1  2  3  4  species rank  Figure 11. K-dominance curves ranking species for grid C control and grid D supplemented communities, on July 5, 1991, and July 19, 1992. Species are food ranked from most to least dominant.  5  56 occurs on July 19 however, and the K-dominance curves for that date do intersect (Figure 1 lb). In this case, the Shannon-Wiener index ordered treatment D more diverse than control C (Figure 8a), while the Simpson index gave the opposite ordering (figure 6a). The ability to communicate in a universal language is desirable, and is one positive result of common use of the Shannon-Wiener and Simpson indices, whether the reasons for their use are scientifically sound or not. Using them in my study for comparative purposes is a recommended objective (Magurran 1988). Having chosen formulae which place value on both ends of the relative abundance spectrum (dominant and rare) permits a discussion of the data from both points of view. Log series alpha is preferred over the Shannon-Wiener index for its higher discriminatory abilities, and because of recommendations in the literature for its usage, and non-use of ShannonWiener (Peet 1974, Goodman 1975, May 1975, Alatalo and Alatalo 1977, Southwood 1978, Taylor 1978, Routledge 1979, Magurran 1988). In measuring response of nematode species abundance patterns to pollution, where one or a few species usually become very dominant, the Simpson index, which weights species by their abundance, is preferred (Platt el a!. 1984, Lambshead and Platt 1985). My data are similar and indicate a rapidly increasing deer mouse abundance. For the above reasons I used Simpson’s index for dominance, and log series alpha for a richness index. Despite the very strong increasing dominance of deer mice, log series alpha, with its high discriminatory ability, indicated that one of the treatment areas which had two more species than its control counterpart during most of 1992, similarly had a higher diversity. Using the Simpson and Shannon-Wiener indices show how areas which have very different communities, may be evaluated purely on whether the investigator places more value on richness or dominance. “It is a well known but not sufficiently emphasized observation that numerical diversity indices conceal more than they reveal. However, it still remains very attractive to be able to reduce a  57 complex set of data to just one figure, especially when the results are eventually intended to be assessed by a non-specialist” (Platt et al. 1984). My intention is to give the best assessment of what the computed indices mean about the small mammal communities being compared. Evenness was included for comparative purposes, but for reasons giveh, it is ignored in the discussion of results section. The indices used for analyses, log series alpha, Shannon-Wiener, the reciprocal of Simpson’s index (referred to as Simpson), and Shannon-Wiener and Simpson evenness are all relatively highly discriminatory, and independent of sample size (number of animals caught. Other indices which are sensitive to sample size should not be used here because of the larger sample sizes (more animals caught per trap period) in food supplemented than control areas. FOCUSING THE DIVERSITY CONCEPT  Adding a prefix or modifier to diversity gives the concept boundaries, such as the word ‘biodiversity’. It has a prefix ‘bio’ which means life, so the concept is . From an anthropomorphic viewpoint, this is a 2 limited to the diversity of living things very broad limit. By adding other prefixes the notion becomes more focused. The following sequence of prefixes (modifiers) focus the concept of diversity to the narrower limits of the taxonomic order of Rodentia diversity: diversity, biodiversity, animal diversity, mammal diversity, Rodentia diversity (referred to simply as rodent diversity in Chapter 1). Although rodent diversity has global connotations, the notion may be further modified to impose spatial limits. Spatial limits could be strictly political, as in  practice some people consider non-living things (dcad wood lr example) part of biodiversity. There are living things in the dead wood. Also, things such as structures and Ilinctions are supervenient (come along with), and not necessary in the description (Wood 1993).  58 British Columbia, or geographical as in Vancouver Island. Furthermore, vegetation limits could be drawn, such as rodent diversity in an even-aged lodgepole pine stand. Thus, for the purposes of my study, using both geographical and vegetation limits, data are used to determine the diversity of rodents in 21 to 28-year-old lodgepole pine stands at 1300 m elevation in the upper Ewer Creek watershed of central B.C. This is a defined community of mammals interacting in a specific environment. These points show why when speaking of biodiversity, the context must be clearly defined. The very fact that there is conflict over how to measure diversity shows the strong dedication and concern by people for the well-being of ecosystems. There is  still a basic weakness in the concept of diversity, because its measures are all valueladen. This is acceptable if it is very clearly indicated what type of diversity is being shown, and what the basic data consist of so other indices can be used for a different perspective. Species abundance models seem good only for theoretical discussion, but not for classifying empirical data. There are many reasons for using Simpson and log series alpha to calculate diversity statistics for my data. Most importantly, they are both highly discriminatory, they are not unduly affected by sample size, and both are widely used, thus my results can be compared to other studies.  59  LITERATURE CITED Abramsky, Z. 1978. Small mammal community ecology changes in species diversity in response to manipulated productivity. Oecologia 34: 113-124. Adams, I.E. and McCune, E.D. 1979. Application of the generalized jack-knife to Grassle, Shannon’s measure of information used as an index of diversity. in diversity J.F., Patil, G.P., Smith, W. and C. Taille. eds. 1979. Ecological theory and practice. International Co-operative PubI. House, Fairland, MD, Pp. 117-13 1. Alatalo, R. and R. Alatalo. 1977. Components of diversity: multivariate analysis with interaction, in Magurran, A.E. 1988. Ecological diversity and its measurement. Princeton University Press. Princeton, New Jersey. 179 pp. Bouti ii, S. 1990. Food supplementation experiments with terrestrial vertebrates: patterns, problems, and the future. Can. J. Zool. 68: 203-220. Proceedings, Wildlife Bunnell, F.L. 1990. Biodiversity: what, where, why and how. Forestry Symposium: a workshop on resource integration for wildlife and forest managers. Prince George, B.C., March 1990. Pp. 29-45. Burton, P.J., Balisky, A.C., Coward, L.P., Cumming, S.G. and D.D. Kneeshaw. 1992. The value of managing for biodiversity. For. Chron. 68(2): 225-237. Dennis, B. and P. Patil. 1989. Species abundance, diversity, and environmental Grassle, J.F., Patil, G.P., Smith W., and C. Taillie. eds. predictability. 1979. Ecological diversity in theory and practice. International Co-operative Pub]. House. Fairland, MD. U.S.A. Pp. 93-114. Goodman, D. 1975. The theory of diversity-stability relations in ecology. jn Magurran, A.E. 1988. Ecological diversity and its measurement. Princeton University Press. Princeton, New Jersey. 179 pp. Hill, M.O. 1973. Diversity and evenness: a unifying notation and its consequences. Ecology 54(2): 427-432. Hunter, M.L. 1990. Wildlife, forests and forestry principles of managing forests for biological diversity. Prentice Hall Inc. Englewood Cliffs, New Jersey. 370 pp. -  Huribert, S.H. 1971. The non-concept of species diversity: a critique and alternative parameters. Ecology 52: 577-586.  60 Huribert, S.H. 1984. Pseudoreplication and experimental design of ecological field experiments. Ecol. Monogr. 54(2): 187-211. Hutcheson, K. 1970. A test for comparing diversities based on the Shannon-Wiener formula. J. Theor. Biol. 29: 151-154. Kimmins, H. 1992. Balancing act: environmental issues in forestry. UBC Press. 224 pp. Krebs, C.J. 1966. Demographic changes in fluctuating populations of Microtus caflfornicus. Ecol. Monogr. 36: 239-273. Krebs, C.J. 1989. Ecological methodology. Harper and Row Pubi., New York, New York. 654 pp. Lambshead, J. and H.M. Platt. 1985. Structural patterns of marine benthic assemblages and their relationships with empirical statistical models. Gibbs, P.E. ed. 1984. Proceedings of the 19th European marine biology symposium, Plymouth. Cambridge University Press, Cambridge. Pp. 371-380. Lloyd, M. and R.J. Ghelardi. 1964. A table for calculating the ‘equitability component of species diversity. J. Anim. Ecol. 33: 2 17-255. MacArthur, R.H. 1972. Geographical ecology: Patterns in the distribution of species. Harper and Row. Publ., New York. 269 pp. Magurran, A.E. 1988. Ecological diversity and its measurement. Princeton University Press. Princeton, New Jersey. 179 pp. May, R.M. 1975. Patterns of species abundance and diversity. n Cody, M.L., and J.M. Diamond. eds. 1975. Ecology and evolution of communities. Harvard University Press, Cambridge, MA. Pp. 81-120. Meidinger, D. and J. Pojar. eds. 1991. Ecosystems of B.C. Special report series, ISSN 0843-6452; no. 6. Crown Publ. Inc. Victoria, B.C. 330 pp. Norrie, M.B. and J.S. Millar. 1989. Food resources and reproduction in four microtine rodents. Can. J. Zool. 68: 641-650. Patrick, R., Hohn, M. and J. Wallace. 1954. A new method of determining the pattern of the diatom plora. in Abramsky, Z. 1978. Small mammal community ecology changes in species diversity in response to manipulated productivity. Oecologia 34: 113-124.  61 Peet, R.K. 1974. The measurement of species diversity. Ann. Rev. Ecol. System. 5: 285-307. Pielou, E.C. 1966. Species-diversity and pattern-diversity in the study of ecological succession. J. Theoret. Biol. 10: 370-383. Pielou, E.C. 1975. Ecological diversity. Wiley, New York. 165 pp. Platt, H.M., Shaw, K.M., and P.J.D. Larnbshead. 1984. Nematode species abundance patterns and their use in the detection of environiliental perturbations. Hydrobiolugia. 118: 59-66. Routledge, R.D. 1979. Diversity indicies: which ones are admissible. In Magurran, A.E. 1988. Ecological diversity and its measurement. Princeton University Press. Princeton, New Jersey. 179 pp. Seber, G.A.F. 1982. The estimation of animal abundance. 2nd ed. Charles Griffin and Co., London. 654 pp. Simpson, E.H. 1949. Measurement of diversity. Nature 163: 688. Southwood, T.R.E. 1978. Ecological methods: With particular reference to insect populations. 2nd ed. Chapman and Hall, London. 524 pp. Sullivan, T.P. 1979. The use of alternative foods to reduce conifer seed predation by the deer mouse, (Peroinyscus mcinicularus) J. Anim. Ecol. 16: 475—495. Sullivan, T.P. 1990. Responses of red squirrel (Tainiasciurus hudsonicus) populations to supplemental food. J. Mammal. 71(4): 579-590. Sullivan, T.P. and W. Klenner. 1993. Influence of diversionary food on red squirrel populations and damage to crop trees in young lodgepole pine forests. Ecol. Applic. (in press). Sullivan, T.P. and D.S. Sullivan. 1982. The use of alternative foods to reduce lodgepole pine seed predation by small mammals. J. Appi. Ecol. 19: 33-45. Sullivan, T.P., Sullivan, D.S. and C.J. Krebs. 1983. Demographic responses of a chipmunk (Eurainias townsend/i) population with supplemental food. J. Ani m. Ecol. 52: 743-755. Szaro, R. and H. Salwasser. 1990. Conserving the heritage: The USDA forest service on biological diversity. In The Society. 1989. Forestry on the frontier. Proc. Soc. Am. For. NatI. Cony., Sept. 24-27. Spokane, Washington. Bethesda, Md. Pp. 12-15  62 Tamarin, R.H. 1985. Biology of new world Microtus. Special Pubi. No. 8. The Amer. Soc. of Mammal. 893 pp. Taylor, R.H. 1978. Bates, Williams, Hutchinson a variety of diversities. in Mound, L.A. and N. Warloff. eds. 1978. Diversity of insect faunas: 9th symposium of the Royal Entomological Society. Blackwell, Oxford. Pp. 1-18. -  Taylor, L.R. 1984. Synoptic dynamics, migration and the Rothamsted insect survey. J. Anim. Ecol. 55: 1-38. Taylor, L.R., Kempton, R.A. and I.P. Woiwood. 1976. Diversity statistics and the log-series model. J. Anim. Ecol. 45: 255-272.  Tilman, D. 1982. Resource competition and community strucwre. Princeton University Press, Princeton. 296 pp. Washington, H.G. 1984. Diversity, biotic and similarity indicies: a review with special relevance to aquatic ecosystems. Water Res. 18(6): 653-694. Westman, W.F. 1990. Managing for biodiversity: Unresolved science and policy questions. Bioscience 40(1): 26-33. Wood, P.M. 1993. The priority of biological diversity conservation on forest land use decision making. PhD. Dissert. U.B.C. (in progress). Chapter 4.  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items