Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Community- and species-level consequences of competition in an unproductive environment: an experimental… Treberg, Michael Anthony 2007

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

Item Metadata

Download

Media
24-ubc_2008_spring_treberg_michael_anthony.pdf [ 8.85MB ]
Metadata
JSON: 24-1.0066297.json
JSON-LD: 24-1.0066297-ld.json
RDF/XML (Pretty): 24-1.0066297-rdf.xml
RDF/JSON: 24-1.0066297-rdf.json
Turtle: 24-1.0066297-turtle.txt
N-Triples: 24-1.0066297-rdf-ntriples.txt
Original Record: 24-1.0066297-source.json
Full Text
24-1.0066297-fulltext.txt
Citation
24-1.0066297.ris

Full Text

COMMUNITY- AND SPECIES-LEVEL CONSEQUENCES OF COMPETITION IN AN UNPRODUCTIVE ENVIRONMENT: AN EXPERIMENTAL APPROACH USING BOREAL FOREST UNDERSTORY VEGETATION by MICHAEL ANTHONY TREBERG B.Sc., University of Guelph, 1997 M.Sc., University of British Columbia 2000  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Botany)  THE UNIVERSITY OF BRITISH COLUMBIA NOVEMBER 2007  © Michael Anthony Treberg, 2007  ABSTRACT In this thesis, I describe three experimental studies that investigate the hotly debated role of competition in structuring communities in unproductive habitats. The studies were done in a boreal forest understory plant com munity in the southwestern Yukon. The first study was a traditional neighbo ur removal experiment. Ten of the most common species were transplanted as seedlings into transects with and without neighbours in a factorial design with two levels of water addition and two levels of fertilizer addition. The presence of neighbours increased survival and biomass of 6 species indicating a facilitative effect of neighbouring plants. The second study used the Community Density Series (CDS) methodology. The first of these was a 10-species experimental community established from seed and grown in sandboxes at 6 densities with 2 watering levels and 2 fertilizer levels in a factorial design. At the corn munity level, density dependence was observed at all life stages, but was not consistently competitive or facilitative - both emergence and final per plant shoot mass were density dependent, while survival to the end of the season was inversely density dependent. The effect of water was positive at seed emergence whereas fertilizer negatively affected survival. Species specific responses were also dependent on life stage. The final study was a 4-year CDS in the field using 9 com mon understory species at 6 densities and 3 fertilizer levels. Density negatively affected the comm unity every year except for the first with competition being important at all densities above x1/8 th the average community density. Constant final yield was reached in plots above the natural x1 density for the last two years of the study. Responses to de nsity were speciesspecific and 7 species declined with increasing density. No facilitative effects were observed. These studies demonstrate that density dependence is important in structuring this unproductive boreal understory habitat. The CDS approach allows us to quantify both the intensity and importance of plant competition at the community and species levels and to determine whether the importance of these biotic interactions depend on abiotic factors. The results clearly show that species-specific responses to biotic interactions are not necessarily the same as community level responses and if we are to understand community structure, it is necessary to use appropriate methodologies.  ii  TABLE OF CONTENTS ABSTRACT ^  TABLE OF CONTENTS ^  ii  LIST OF TABLES ^  vi  LIST OF FIGURES ^  ix  ACKNOWLEDGEMENTS ^  CO-AUTHORSHIP STATEMENT ^  xii  xiii  CHAPTER 1 INTRODUCTION: COMPETITION IN UNPRODUCTIVE ENVIRONMENTS ^ 1 The Grime/Tilman debate ^ 1 Methods to investigate the role of competition in structuring communities ^ 4 The community density series (CDS) ^ 6 Evidence on the importance of biotic interactions in unproductive regions ^ 7 Thesis overview ^ 10 References ^ 12  CHAPTER 2 FACILITATION IN A BOREAL FOREST UNDERSTORY: RESULTS FROM A NEIGHBOUR REMOVAL, FERTILIZATION AND WATERING EXPERIMENT ^ Introduction ^ Methods ^ Study site ^ Study species ^ Experimental Design ^ Statistical Analysis ^ Results ^ Discussion ^ The importance of neigh bours ^ The unimportance of fertilization ^ And just how important is watering? ^ Conclusion ^ References ^  17 17 18 18 19 19 21 22 27 27 29 31 32 33  iii  CHAPTER 3 THE COMMUNITY DENSITY SERIES (CDS) ^  36  Equations describing the competition-density effect and the yield density effect i n monocultures ^ 37 Analyzing the CDS ^ 46 References ^ 51  CHAPTER 4 DENSITY DEPENDENCE IN AN EXPERIMENTAL BOREAL FOREST UNDERSTORY COMMUNITY ^ 53 Introduction ^ Methods ^ Community description ^ Experimental design ^ Treatments ^ Monitoring and harvesting ^ Analysis ^ Results ^ Density dependence in the community ^ Water and fertilizer effects on the community ^ Species-specific responses ^ Discussion ^ References ^  53 54 54 55 57 59 59 61 61 64 65 69 76  CHAPTER 5 COMMUNITY- AND SPECIES-LEVEL CONSEQUENCES OF COMPETITION IN AN UNPRODUCTIVE ENVIRONMENT ^ 79 Introduction ^ Methods ^ Study site ^ Experimental design ^ Analysis ^ Results ^ Community-level responses ^ Species-level responses ^ Discussion ^ References ^  CHAPTER 6 CONCLUSIONS ^ References ^  79 82 82 84 87 88 88 93 97 102  106 110  iv  APPENDIX 1 HOW TO GROW AND KILL THE NATIVE PLANTS OF KLUANE ^ 112 Growing from seed ^ Growing from cuttings ^ Transplanting ^ Killing with Glyphosate ^ The problem of clonal plants ^ References ^  APPENDIX 2 ARE VOLES ATTRACTED TO FERTILIZER? A CAUTIONARY TALE ^  112 114 115 116 117 119  120  Introduction ^ 120 Methods ^ 122 Site description ^ 122 Vole nests observed in fertilized CDS plots ^ 122 Voles captured in fertilized plots ^ 123 Voles captured in traps with fertilizer as bait ^ 123 Voles entering locked-open traps to remove fertilizer ^ 124 Results ^ 125 Vole nests observed in fertilized CDS plots ^ 125 Voles captured in fertilized plots ^ 127 Voles captured in traps with fertilizer as bait ^ 128 Voles entering locked-open traps to remove fertilizer ^ 129 Discussion ^ 131 References ^ 133  LIST OF TABLES Table 1.1 Summary data for studies on biotic interactions done in northern and alpine environments. The methods column includes the methods used in all aspects of the authors' studies and the symbols used are: FE = field experiment, FE (obs.) = field observations, LE = lab experiment, SA = spatial association, N = nutrient manipulation, A = additive, R = removal, S = substitutive (or replacement series), and m = evidence for the mechanism is determined. The results column indicates whether the interactions observed were facilitation (+), competition (-) or no interaction (0). ^ 7 Table 2.1. Summary of ANOVA for a) the summed percent survival of all transplants along a transect and for b) the summed biomass of all transplants along a transect. Main effects are Neighbours (N), Fertilization (F) and Watering (W). Values in bold are significant at P < 0.05, and those in italics are significant at P < 0.10. ^ 22 Table 2.2. Probabilities derived from ANOVAs for the biomass of each of the 10 species. Values in bold values are significant (P < 0.05) ^ 24 Table 2.3 Probabilities derived from ANOVAs for the environmental variables monitored in the transects. Neighbours are not included in the ANOVA for LAI because there were no live plants in the neighbour removed plots. Values in bold are significant (P < 0.05) 27 Table 3.1. Best fitting lines and r2 for the data presented in Figure 3.7. ^  50  Table 4.1. Regression coefficients for the community response variables and density relationships shown in Figures 4.2, 4.3 and 4.4. The model type refers to the data transformation that best linearized the data. The degrees of freedom (df) are for the model and error combined. A negative slope indicates negative density dependence (or competition) and a positive slope indicates positive density dependence (or facilitation). Significant values (P < 0.05) are in bold ^ 61 Table 4.2. Summary of ANCOVAs and ANOVA for the response variables to density manipulations in the experimental communities. Significant values (P < 0.05) are in bold. 64 Table 4.3. Regression coefficients for the relationship between each species' survival or mean plant mass biomass and the initial planting density in the experimental communities. The proportional survival is the number of individuals of that species per plot divided by the initial planting density and mean plant mass is the mass of that species per plot divided by the relative initial density. The model type refers to the transformation required to best linearize the data. The degrees of freedom (df) are for the model and error combined. Significant values (P < 0.05) are in bold ^ 66 Table 4.4. P values from ANOVAs and ANCOVAs for each species using the proportional survival or the total biomass of each species per plot divided by the relative initial density. Significant values (P < 0.05) are in bold. ^ 69 Table 5.1. The abundance of all species found in the 63 1 m 2 CDS plots during the initial survey in 1999. Frequency is the percent occurrence in the 63 plots. Percent cover was estimated using a point frame with 100 pin drops per m 2 . Density was assessed by counting all individuals in the 1 m 2 plot. The density of Arctostaphylos uva-ursi, Festuca and Linnaea were not estimated (n/a) due to the difficulty in identifying distinct individuals ^ 83  vi  Table 5.2 The equations used to estimate biomass for the years 1999 through 2001 for all species growing in the CDS and control plots. These equations were the best fitting curves between biomass and various surrogates for biomass and were based on destructive sampling done in 1999. In the equations: C = cover, H = height, L = length of the longest leaf, N = number of leaves, and W = width of longest leaf. All equation components are 86 measured in mm and all estimated masses are in grams. ^ Table 5.3. Regression coefficients for the mean plant size index and density relationships for the years 1999 through 2002 and the evenness and density relationship in 2002. These data are plotted in Figures 5.1 and 5.3. The mean plant size index is the total plot mass divided by the density. In the absence of interactions, there should be no relationship between the mean plant size index and density. The negative slopes indicate negative density dependence (or competition) for the plant size index. Model type refers to the data transformation that best linearizes the data. The degrees of freedom are for the model and 89 error combined. Significant values (P < 0.05) are in bold ^ Table 5.4. Summary of ANCOVAs for the mean plant size index in the CDS for 1999 to 2002 and ANOVA for total plot biomass in the control plots in 2002. The control treatment for the total plot biomass ANOVA compares the mean of the unmanipulated control plots to the x1 density in the CDS plots. Significant values (P < 0.05) are in bold. ^ 92 Table 5.5. Regression coefficients for the relationship between each species mean plant mass and density. The model type is the transformation that best linearized the data. The degrees of freedom (df) are for the model and error combined. Significant values (P < 0.05) are in bold and values where P< 0.10 are in italics. These data are plotted in Figure 5.4.. 94 Table 5.6. Summary of ANCOVAs and ANOVAs on each species' mean plant mass in the CDS in 2002, in response to manipulations of density and fertilizer. When the effect of density was not significant (P < 0.05, Table 5.5) on the mean plant mass, an ANOVA with just fertilizer as an effect was performed. Significant values (P < 0.05) are in bold ^ 97 Table A 1.1. The mean (± 1 SE) percent germination of the most common understory species at the research site in Kluane described in chapters 2-4. These data are for seeds collected from 1999 to 2002. Seeds were sown onto wet sand in Petri plates with 50 seeds per Petri plate (n = 3 plates). 112 Table A 1.2. The mean (± 1 SE) percent germination of some of the less common species collected near the research site in 1999. Seeds were sown onto wet sand in Petri plates with 50 seeds per Petri plate (n = 3 plates). ^ 113 Table A 1.3. The mean (± 1 SE) percentage rooting of fresh cuttings after 30 days. All cuttings were from new leaves. The cut end of the leaf was dipped into commercially available rooting hormone and cuttings were placed in planting trays filled with moist sand. The tray was covered with plastic and the sand kept moist. This experiment ran from July 8 to Aug 7, 2002. ^ 115 Table A 1.4 The number of applications of a 1:20 (Glyphosate to water) concentration necessary to see significant die back in the named plant species. Applications were approximately 1 week apart and were applied with a pump sprayer. The leaves were soaked with the solution until dripping wet. The higher the number beside the species, the higher the resistance to the Glyphosate. ^ 117  vii  Table A 2.1. Summary of ANOVA of the number of over-winter nests observed in the CDS experiment. Significant effects (P < 0.05) are shown in bold. ^  126  Table A 2.2. Summary of ANOVAs on the number of mice and voles caught in the fertilized plots at the Boutellier and Microwave Road sites for the 2 trapping sessions. Significant effects 127 (P < 0.05) are shown in bold ^ Table A 2.3. Summary of ANOVA on the number of voles captured in traps with and without fertilizer included in the bait. ^  128  viii  LIST OF FIGURES Figure 2.1. The percent survival (± S.E.) of all transplants summed per transect with or without neighbours at low (F1) and high (F2) fertilizer addition and at low (W1) and high (W2) watering. ^ 23 Figure 2.2. The total biomass (g ± S.E) of all transplants summed per transect with or without neighbours at low (F1) and high (F2) fertilizer addition and at low (W1) and high (W2) watering. ^ 23 Figure 2.3. Significant interactions from the ANOVA in Table 2.2 between a) neighbour and fertilization, b) neighbour and watering and c) fertilization and watering on the transformed biomass (g ± S.E.) of Festuca. Columns sharing the same letter are not significantly different (P > 0.05) as determined by Tukey's HSD ^ 25 Figure 2.4. Significant interactions from the ANOVA in Table 2.2 between a) fertilization X watering X neighbours and b) watering X fertilization on the transformed biomass (g ± S.E) of Solidago. Columns sharing the same letter are not significantly different (P > 0.05) as determined by Tukey's HSD ^ 26 Figure 3.1. The Community Density Series and the relationship between density and final yield. Dc is the density at which competition begins to reduce the linear increase in final yield and Dm is the density where the mass of the community remains constant even if density still increases. This is also called constant final yield. Figure redrawn from Goldberg et al. (1995). ^ 37 Figure 3.2. The relationship through time between density of soybean plants and mean dry weight per plant on (a) arithmetic scales and (b) double logarithmic scales and between density and yield on (c) arithmetic scales and (d) double logarithmic scales. Figures (b) and (d) are redrawn from Kira et al. (1953). ^ 39 Figure 3.3. A sample dataset showing the relationship between density and (a) mean dry weight and (b) yield for subterranean clover. The solid line in (a) and (b) are the equations suggested by Kira et al. (1953) for mean dry weight and yield, respectively. The dashed lines are the reciprocal equations of Shinozaki and Kira (1956). In this case, the reciprocal equations are a better fit for both weight and yield. A square link is an old fashioned measure of area equivalent to 0.939034 m 2 . Data from Table 1 from Willey and Heath (1969) ^ 41 Figure 3.4. The effect of varying a single parameter in equations (4) and (5) when the other is held constant on the mean plant weight and total yield. In (a) and (b) the value of B is held constant and A varies. In (c) and (d) the value of A is held constant. Changes in either A or B do not affect the shape of the curve, they only shift the position of the curve. ^ 42 Figure 3.5. The relationship between density and (a) mean root weight and (b) root yield of the globe red beet. The general reciprocal equations (3.7) and (3.8) of Watkinson (1980) are fitted to the mean root weight and root yield data. This is an example of the parabolic relationship between yield and density. Data redrawn from Willey and Heath (1969). ^ 45 Figure 3.6. The effect of varying a single parameter in equations (3.7) and (3.8) when the others are held constant on the mean plant weight and yield. In (a) and (b) the value of c and w, are held constant and b varies. In (c) and (d) the value of c varies. Increasing iv, merely shifts the y-intercept higher for all figures (not shown). ^ 46 Figure 3.7. Mean plant mass and yield in relation to density for 1:1 mixtures of Silybum marianum and Cirsium vulgare. The legend in (a) applies to all 4 panels and the solid circle is  ix  the combined mean or yield for the two species. Figures (a) and (c) show the same data, except (a) has arithmetic axes and (c) has double logarithmic axes. Likewise for figures (b) and (d). The lines shown are the best fitting geometric power relationships of Kira et al. (1953) which are presented in Table 3.1. Data redrawn from Figure 3 of Austin et al. (1988). ^ 50 Figure 4.1. Sandboxes and greenhouses used in the CDS experiment at Kluane Lake. The author is shown using a propane torch to sterilize the surface of the sand. ^ 57 Figure 4.2. The effect of density on the performance of seedlings in the experimental communities. In (a) the emergence index is the total seedlings that germinated per square meter divided by the relative seed density, (b) the proportion surviving is the number of emerged seedlings that survived to the end of the season, and (c) the mean shoot mass per plant is the total plot mass divided by the number of plants in the plot. The coefficients for the best fit curve for each graph are shown in Table 4.1. ^ 62 Figure 4.3. The effect of initial planting density on the final plant density. The coefficients for the best fit line are shown in Table 4.1. ^ 63 Figure 4.4. The effect of initial seed density on (a) species richness and (b) evenness using Smith and Wilson's (1996) Evar, in the experimental community; lower values of Eva,- indicate that the species are very unequally represented in the community. The best fit curve for species richness is shown as a solid line and the coefficients are given in Table 4.1. The dashed line is the expected value of the species richness as calculated from the null community. Values in both figures are means (+/-95 % confidence interval). ^ 63 Figure 4.5. The effect of initial planting density on the proportional survival for the species in the experimental community that showed significant responses to varying density. The best fitting curves are described in Table 4.3. Species whose survival was not significantly related to initial planting density are not shown ^ 67 Figure 4.6. The effect of initial planting density on the mean plant mass for the three species in the experimental community that showed significant responses to varying density. The plant mass is the total g/m 2 for that species divided by its relative initial density. The best fitting curves are described in Table 4.3. Species whose mass was not significantly related to initial planting density are not shown. ^ 68 Figure 5.1. The effect of density on the mean plant size index (the total plot mass divided by density) for the years 1999 through 2002. All curves shown are statistically significant (P < 0.001) and the coefficients for the best fit curve are given in Table 5.3. In the absence of interactions, there should be no relationship between the mean plant size index and density. A negative slope indicates negative density dependence (competition). The y-axis for 2002 has a different scale than the graphs for other years. The natural field density is x 1. The density that competition began to reduce mean plant mass was at x 1/8 (i.e. 0.125) for all graphs. The vertical dashed line represents the density that constant final yield is reached for 2001 and 2002. Constant final yield was not reached in 1999 or 2000. ^ 90 Figure 5.2. The effect of fertilizer level on (a) mean plant size index (total plot mass divided by density) in the CDS, (b) evenness in the CDS and (c) the total plot biomass for the controls (density not manipulated) and the x1 density plots in the CDS experiment. Error bars are ± 1 S.E. All data are for 2002. Columns that share the same letter are not statistically different (Tukey's HSD, P > 0.05) 91 Figure 5.3. The effect of plant density on Smith and Wilson's (1996) index of evenness (Evar). Lower values of Eva , indicate that the species are very unequally represented in the community. The natural field density is x 1. The best fit curve is statistically significant (P < 0.001) and the coefficients for the curve are given in Table 5.3. ^ 93  Figure 5.4. The effect of plant density on the mean plant mass or mean plant size index (total plot mass divided by density) for species in the CDS plots in 2002. All curves shown are statistically significant at P < 0.05, except for Mertensia and Solidago, which are significant at P < 0.10. The coefficients for the best fit curve are given in Table 5.5. In the absence of interactions, there should be no relationship between the mean plant mass or mean plant size index and density. A negative slope indicates negative density dependence (competition). The natural field density is x 1. Solid vertical lines indicate the density that competition begins to reduce the mean plant mass. Dashed vertical lines indicate the density where the final constant yield for that species was reached ^ 95 Figure 5.5. The effect of fertilizer level on the mean plant mass (±1 S.E.) or mean plant size index (±1 S.E.) for each species in the CDS in 2002. Columns sharing the same letter are not statistically different (Tukey's HSD, P > 0.05). ^ 97 Figure A 2.1. An over-winter vole's nest in one of the plots in the main experiment of Chapter 5. The white dots are fertilizer pellets from the previous summer that have not yet dissolved. The nests are surrounded by long pieces of cut grass (Festuca altaica) and often contain many short (-2.5 cm) pieces of grass formed in small haystacks. These stacks are created by voles (primarily Microtus spp.) that cut the stems of Festuca in order to reach the seed heads (Forsyth 1999). ^ 121 Figure A 2.2. Plastic plate covered with vegetable oil and dusted with talcum powder. These were placed in the entrance tunnels of the Longworth traps to determine if mice and voles had entered the traps to remove fertilizer ^ 125 Figure A 2.3. The mean number of nests per plot and 95% confidence interval as a function of (a) fertilizer treatment (averaged over all plant densities) and (b) plant density (averaged over the three fertilizer levels) in the CDS experiment. In (a), columns sharing the same letter are not significantly different (P<0.05, Student's t). The control shown in (b) are those plots that did not undergo density manipulation but did have the three fertilizer treatments ^ 126 Figure A 2.4. The mean number of voles captured per plot and 95% confidence interval during the June trapping session in (a) the fertilized and unfertilized plots and the (b) fenced and unfenced plots at the Boutellier Summit and Microwave Road sites. Columns sharing the same letter are not significantly different (Student's t, P<0.05) 127 Figure A 2.5. The mean number of voles captured per trap and the 95% confidence interval for traps with and without fertilizer included in the standard barley + apple bait. These are not statistically different (P<0.05, Table A 2.3) ^ 128 Figure A 2.6. Evidence inside a Longworth mouse trap that voles had entered the trap. This trap was locked open and initially contained 5.0 g of fertilizer. ^ 129 Figure A 2.7. The number of traps entered per day (solid line and solid circle) and the total fertilizer removed (g) per day (dotted line and open circle). There was no correlation between number of traps entered and total mass of fertilizer removed (r = 0.356, P = 0.212). ^ 130 Figure A 2.8. Mean mass of fertilizer removed (g trap -1 ) and 95% confidence interval depending on the number of tracks going into and out of the traps. There is no statistical difference between these classes (Wilcoxon Test, P = 0.471) ^ 130  xi  ACKNOWLEDGEMENTS I would like to thank my supervisor, Dr. Roy Turkington, for the opportunity to be a member of his lab and to work at Kluane Lake. I could not imagine a better supervisor. I appreciate the help and insight from my committee members, Dr. Gary Bradfield and Dr. Greg Henry and from past members of the lab, especially Andrew M acDougall and Chris Lortie. Many thanks go out to my field assistants, Maureen Bezanson, Saskia Arnesen, and Kate Edwards, for their hard work (and for putting up with me). Assistance in the field also came from Andrew Bachmann, Rebecca Best, Lesley Dampier, Dan Gillis, Pippa Hett, Sarah Lord, Marilyn Makortoff, Jennie McLaren, Erika Olson, and Alda Ngo. Funding for this work came from various sources including a Natural Sciences and Engineering Research Council (NSE RC) grant to Roy Turkington and a NSE RC Graduate Scholarship to myself. I also received a University Graduate Scholarship from the University of British Columbia. Support for transportation and living expenses was provided from the Department of Indian and Northern Affairs in the form of Northern Science Training grants to me and to many of the field assistances we had at Kluane. I thank the staff of The Arctic Institute of North America base at Kluane Lake, especially Andy Williams, Sian Williams, Lance Goodwin, Jessica Log her and Dan Gill is. It was always a pleasure to stay at camp with such wonderful people. Finally, I would like to thank my wife, Elyn, for her love, support and patience.  xii  CO-AUTHORSHIP STATEMENT Chapters 2, 3, 4, and 5 were co-authored with Dr. Roy T urkington. I identified, designed and conducted the research and am solely responsible for all data analyses. Dr. Turkington assisted in manuscript preparation and revision in these chapters.  Appendix 2 was co-authored with Dr. Roy Turkington and Kate Edwards. I identified, designed and conducted the research and am solely responsible for all data analyses. Kate Edwards assisted in the field research and Dr. Roy Turkington assi sted in manuscript preparation and revision.  Chapter 1 INTRODUCTION: COMPETITION IN UNPRODUCTIVE ENVIRONMENTS Competition is important in structuring many plant communities, but whether the importance of competition changes along productivity gradients is hotly debated. It is agreed that competition is important in productive habitats, but the role of competition at low productivity levels is especially debated; one school argues that it is unimportant and another argues that it is just as important as in productive habitats. The primary objective of this thesis is to investigate the effects of competition in a relatively unproductive boreal understory community. Specifically, I will test (0 whether competition has a significant impact on individual fitness and comm unity structure and whether this changes along a gradient of productivity, and (ii) how individual-level patterns of competition intensity relate to community-level patterns and whether this relationship is affected by productivity. To achieve this I will use a traditional style of competition experiment supplemented by experiments using a relatively new method of assessing plant competition at the community level. There are two major theories predicting how competition structures plant communities and how competition intensity changes along gradients of productivity. Grime's theory predicts that competition intensity is positively correlated with productivity (Grime 1977) while Tilman's theory predicts that competition intensity remains constant along the entire gradient of productivity (Tilman 1988). The debate surrounding some of the conflicting predictions of these two theories has been one of the most controversial in plant ecology (Goldberg and Novoplansky 1997) . Both of these theories predict that competition is important at high levels of productivity so the major controversy between them concerns the ro le of competition at low productivity levels (Abrams 1995). This question, the role of competition in a low productive ecosystem, is the primary focus of my thesis.  THE GRIM E/TILMAN DEBATE Grime's model argues that there are three major plant life hi story strategies that are characterized by their levels of environmental stress and disturbance (Grime 1979).  1  Stress in this context (and throughout the rest of this thesis) can be defined as "the external constraints which limit the rate of dry-matter production of all or part of the vegetation" (p.1179, Grime 1977). Therefore, The three strategies are (I) ruderals, characteristic of high disturbance and low stress environments, (ii) stress-tolerators, characteristic of low disturbance and high stress environments, and (iii) competitors, characteristic of low disturbance and low stress environments. The superior competitor in the C-S-R model is the one with the highest resource-capturing abilities (Grime 1979; Grace and Clark 1991). According to Grime's model, competitive ability is positively correlated with the maximum relative growth rate (Grace 1990). This assumption is based on the observation that the maximum relative growth rate is correlated with the ability to capture resources (Grime 1979; Chapin 1980). Because of this positive correlation, a good competitor for one resource will be a good competitor for all resources thus the good competitor will dominate in fertile, undisturbed environments (Grime 1977; Grime 1979). Therefore, the ability to compete belowground for nutrients and water will be positively correlated with the ability to corn pete aboveground for light because of the assumption that a good competitor is the one that can capture the most resources due to its maximum relative growth rate. Because these are evolved characteristics, tradeoffs exist in this model in the sense that a species can only have one strategy. Therefore, a good competitor cannot also be a good stress-tolerator (Grime 1977). From the positive character correlation between competitive ability and maximum relative growth rate, it follows that competition intensity should increase a s productivity increases (Grime 1977; Grime 1979). Grime's predictions are generally consistent with those from some other theories (Huston 1979; Keddy 1989; Keddy 1990). Tilman's model (1988) is more mechanistic and quantitative, as opposed to Grime's rather qualitative approach, and is based on a set of equations developed from his resource-based theory of competition (Tilman 1980; Tilman 1982). Since all material and energy must go into the production of roots, shoots and leaves, there are tradeoffs associated with allocating more energy into one corn ponent than another (Tilman 1988; Tilman 1990). While many combinations are theoretically possible, not all are viable in the presence of competing species (Tilman 1990). The competitively superior morphology is totally dependent on the environmental constraints of the system. The superior competitor in Tilman's model is the one that can dominate a habitat by driving its competitors to extinction (Grace 1990). This means that competition can be defined  2  as the ability to reduce the levels of an available resource (Grace 1993) and the superior competitor is the species that lowers resources to levels that neighbouring species can not tolerate (Tilman 1988). Therefore, the species with the lowest R* (the resource concentration that a species requires to persist in a habitat) is the species that will be the superior competitor, and will dominate, in a certain habitat (Tilman 1990). The plant species with a competitively superior morphology in a given habitat wi II not be a good competitor in other habitats because of the associated tradeoffs in morphology (Tilman 1988; Tilman 1990). The model predicts that competition intensity along a productivity gradient will remain constant although the relative importance of shoot and root competition will vary with productivity. Therefore, shoot competition will be important in productive habitats where belowground resources are abundant and light is limiting, whereas in unproductive environme nts, most competition will occur belowground because light is not limiting (Tilman 1988). Tilman's predictions of total competition intensity being uniform across a gradient of productivity are also consistent with other models (Newman 1973; Grubb 1985 ; Taylor et al. 1990). There are a plethora of studies that support either Grime or Tilman. A few generalizations can be made about the experimental evidence supporting each ch theory, namely that results tend to support one model over another depending on the index of competition calculated and the type of productivity gradient used. However, first it is important to clarify some details about the definitions of competition used in each model. Grime's model primarily focuses on the capture of resources while Tilman's model is concerned with plant dominance and the plant traits that lead to it, and especially the ability to tolerate low resource levels. T his distinction is partially resolved when it is realized that Grime is concerned with the effect component of competition and Tilman is concerned with the response component of competition as defined by (Goldberg 1990). A plant can be a good effect competitor through the acquisition of resources or it can be a good response competitor by being able to withstand low resource levels. Since Grime stresses the importance of the traits that confer competitive success by having maximum relative growth rates, he is focusing on good effect competitors. In contrast, Tilman's species with the lowest R* are the best response competitors. Based on experimental evidence Grace (1993, 1995) suggested that results of experiments may differ depending o n how competition intensity is calculated. If competition intensity is calculated as an absolute measure, the results ten d to support Grime. Absolute competition intensity is calculated as the performance of a plant  3  species in monoculture minus the performance of the target species in mixture. If competition intensity is assessed as a relative measure such that the absolute value is divided by the performance of the target species in monoculture, then experimental results tend to support Tilman. This is a consistent pattern (Campbell and Grime 1992; Turkington et al. 1993; Reader et al. 1994). Goldberg (1994) and Grace (1995) have argued that the relative competition intensity index is the preferred method. There is also disagreement on whether it is best to stand ardize the relative measure by dividing by the monoculture performance or by the performance of individuals, since monocultures still may experience intraspecific competition (Miller 1996). Two kinds of productivity gradients are used in studies of competition intensity natural gradients of productivity or experimentally generated gradients of productivity. These two gradients tend to support different hypotheses (Goldberg and Barton 1992). Studies using natural gradients tend to support Grime (Gurevitch 1986; Wilson and Keddy 1986a; Wilson and Keddy 1986b; Reader and Best 1989), whereas studies using manipulated productivity gradients, usually through the addition of nitroge n, tend to support Tilman (Gurevitch et al. 1990; Wilson and Shay 1990; Wil son and Tilman 1991; Wilson and Tilman 1993; Wilson and Tilman 1995); there are a few exceptions to this generality (Belcher et al. 1995; Kadm on 1995).  METHODS TO INVESTIGATE THE ROLE OF COMPETITION IN STRUCTURING COMMUNITIES Plant community structure is a function of environment and biotic interactions. Numerous studies have demonstrated competitive interactions between individuals of different plant species (Aarssen and Epp 1990; Goldberg and Barton 1992) and most research has focused on the measurem ent of competition effects on individual fitness (Goldberg and Barton 1992). Theoretical considerations suggest that individual-level data will not necessarily, or perhaps even usually, predict community patterns, even on a local spatial scale (Goldberg 1994; Kareiva 1994) and re cent empirical evidence has shown that individual-level effects of competition could not predict community-level effects (Rajaniemi and Goldberg 2000). Three common experimental methods have been used to examine the role of competition in structuring plant communities in the field, all of which have major limitations (Rajaniemi and Goldberg 2000; Zam fir and Goldberg 2000).  4  The most commonly used method to study competition is to examine community structure along environmental gradients in the field (Zam fir and Goldberg 2000). These gradient studies may be either mensurative or manipulative, but in either case it is usually assumed that any changes in community structure along the natural gradient or altered abiotic environment are due to competitive interactions (Campbell and Grime 1992; Weihe r and Keddy 1995). This type of experiment cannot separate the effects of the environment on plant performance from the effects of plant interactions (Austin and Austin 1980; Austin et al. 1985; Campbell and Grime 1992). To demonstrate that competition is important in structuring a community, manipulation of the competition intensity and the abiotic environment is necessary (Rajaniem i and Goldberg 2000). A second method of examining competition is to directly manipulate the density of vegetation in the field to examine the effect on community structure. This is most commonly done by removing the dominant species and examining the effect on the remaining species (Gurevitch and Unnasch 1989; Aarssen and Epp 1990). This method only examines a small subset of the potential interactions in a community and therefore gives limited information on the overall effect of competition on the community. Another problem with this approach is that it is most usually done at two densities (usually the presence and absence of the dominant species). If the effects of neighbours on the targets change with density, the results cannot be interpolated to intermediate densities. Evidence suggests that competitive hierarchies may change with density and therefore the results at a single density (or in this case, two densities) may not be useful for predicting the influence of competition on community structure (Taylor and Aarssen 1989). A third method involves comparisons of competitive hierarchies, based on pairwise competition experiments of a subset, or all, of the species in a comm unity. It is widely accepted that competitive hierarchies exist i n plant communities and that they may be related to species abundance (Keddy 1990). However, the experimental methods used to determine the competitive hierarchies, namely the replacement series and the phytometer approach, have received a great deal of criticism (Bailey 1989; Firbank and Watkinson 1989; Silvertown 1989; Silvertown and Dale 1991; Snaydon 1991; Grace et al. 1992; Gibson et al. 1999). The replacement series is especially problematic because there may be severe size bias in the outcome not directly due to competitive interactions, and the experimental results are dependent on the densities  5  used (Connolly 1986; Connolly 1997). A few more recent papers have argued why replacement series have some good qualities (Cousens 2000; Jolliffe 2000). There are two newly developed methods of directly examining the effects of competition on community structure that use multi-species mixtures. The first is the combined monoculture method, which uses the species' performance in monocultures to standardize the performance in mixtures (Goldberg 1994). The second is the community density series (CDS) that uses the p erformance of species in very low-density mixtures (i.e. a null community without interactions) to standardize performance at high-density mixtures (Goldberg et al. 1995). These two methods have been compared recently and it was concluded that the results obtained by the CDS are more realistic in determining competitive hierarchies and the CDS method is a useful method for investigating the effect of competition on the comm unity (Zamfir and Goldberg 2000).  THE COMMUNITY DENSITY SERIES (CDS) This experimental method was developed by Goldberg et al. (1995) to examine the specific role of competition at the whole com munity level because density-depend ent regulation in plants may frequently occur at the level of the entire community, rather than within particular species. The design is an extension of traditional agronomic studies that examine the yield response of a single species to a range of planting densities (Harper 1977). In the CDS the entire comm unity is manipulated to levels below and above the naturally occurring community density, while keeping relative pro portions of each species constant. Similar to a yield-density experim ent of a single species, the yield-density relationship of the manipulat ed community will reach a total community yield or carrying capacity. A null community (with no species interactions) is defined experimentally with these manipulations rather than with the observational/statistical methods as in other corn petition studies. This approach has been successfully used in the Negev desert to exami ne the role of competition in structuring annual plant communities (Goldberg et al. 2001; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 2005), in experimental communities of bryophytes (Zamfir and Goldberg 2000), and in old-field communities (Rajaniemi and Goldberg 2000). A more detailed description of the CDS c an be found in Chapter 3. I will use this approach in Chapters 4 and 5.  6  EVIDENCE ON THE IMPORTANCE OF BIOTIC INTERACTIONS IN UNPRODUCTIVE REGIONS  As previously described, there is debate on the role of competition in unproductive regions with some arguing that competition increases with productivity, and is therefore unimportant in unproductive areas (Grime 1977; Grime 1979; Huston 1979; Keddy 1989; Keddy 1990), while others theorize that competition occurs in both productive and unproductive habitats (Newman 1973; Tilman 1982; Grubb 1985; Tilman 1988; Taylor et al. 1990). Recently, researchers have begun to discuss the importance of positive interactions in plant communities, especially in unproductive habitats (Bertness and Callaway 1994; Belcher et al. 1995; Callaway 1995; Bertness 1998; Brooker and Callaghan 1998). In general, many of these hypotheses predict that in highly productive systems, competition will dominate as the force structuring communities, and as productivity decreases, positive interactions (facilitation) will become more important. Table 1.1 presents a summary of nearly 30 research studies examining biotic interactions conducted in northern and alpine environments, normally considered unproductive. There are many examples of both competitive interactions as well as many facilitative interactions, often occurring in the same community with the same plant species. Unfortunately, many of these studies do not actually demonstrate the claimed biotic interactions but rather cite it as a possible hypothe sis for the observed results. This anecdotal approach seems to be more common in the earlier literature.  Table 1.1 Summary data for studies on biotic interactions done in northern and alpine environments. The methods column includes the methods used in all aspects of the authors' studies and the symbols used are: FE = field experiment, FE (obs.) = field observations, LE = lab experiment, SA = spatial association, N = nutrient manipulation, A = additive, R = removal, S = substitutive (or replacement series), and m = evidence for the mechanism is determined. The results column indicates whether the interactions observed were facilitation (+), competition (-) or no interaction (0). AUTHORS LOCATION LAT.^METHOD SPECIES^RESULTS INTERACTIONS AND LONG.^STUDIED^ MECHANISMS ALT. Aerts et al.^heathland in^52° 05' N FE, S, N^Erica tetralix,^ The authors attempted to (1990)^Netherlands^05° 50' E^Molinia caerulea,^demonstrate that competitive N/A^ Callum vulgaris^ability changes with differing nutrient availability. However, because of the methods used (S), the results must be viewed with caution.  7  AUTHORS LOCATION LAT.^METHOD SPECIES^RESULTS INTERACTIONS AND LONG.^STUDIED^ MECHANISMS ALT. ^+,Aksenova^alpine^43° 27' N FE, R^Anenome Removing one of the 5 dominant species in a plot and^tundra in^41° 41' E^speciosa, Blinnikov^northwestern 2800 m ^Antennaria dioica, resulted in an increase in (1998)^Caucasus,^ other dominants Festuca ovine, Russia^ (competition) and a reduction Trifolium in subordinate species polyphyllum, Carex spp. (facilitation). ^ Removal of all neighbours Arii and^boreal^61° 02' N FE, R, N^Achillea 0,Turkington^understory in 138° 22' ^millefolium, only benefited one species, (2002)^Yukon^W^ Anenome Achillea. 1000m^parvitlora, Festuca altaica, Lupinus arcticus, Mertensia paniculata Blundon et^alpine^51° 18'N^FE (obs.), Hedysarum Both species demonstrate nucleation; however al. (1993)^moraine in^119° 6W SA ^boreale var. Rockies,^1650 m^mackenzii, recruitment outside the Canada^ patches indicates that it is not Dryas drummondii necessary for succession at this site. Bonde^alpine in^N/A^FE (obs.)^Trifolium nanum^0^Although this paper has been (1968)^Colorado^ cited as an example of positive interactions (in Callaway 1995) there is no evidence in it of such. Callaway et various^See their FE, R^Various species^+ , 0 , -^In general, biomass, growth al (2002)^alpine sites^Table 1^ and reproduction were higher worldwide^for the 11^ in alpine plants with including^locations.^ neighbours nearby. Kluane^ Competition was more Yukon^ common at lower elevations where conditions were more favourable. At higher elevations and colder locales, interactions tended to be positive. Carlsson^arctic^68° 21' N SA, FE, ^Carex bigellowii,^Carex grew better with and^mountain^18° 42' E m^Cassiope^ shrubs possibly due to Callaghan^fellfield in^1150 m^tetragona,^ shelter or patchy resources (1991)^Sweden^ Empetrum^ which also favoured the hermaphoditum^growth of shrubs. Chapin et^alder shrub^66° 39' N FE, SA,^Anus crispa^ Regular spacing of Alnus is al. (1989)^tundra in^150° 40'^R, m^ demonstrated and removal of Alaska^yy^ individuals leads to an increase in the shoot 300 m^ biomass of alders only. This suggests that intraspecific competition for resources belowground is responsible for the observed pattern. del Moral^sub-alpine^47° 57' N FE, R, A,^all plant species^Subdominant species (1983)^meadow in^123° 15'^N, m^mostly Carex and^increased in biomass after Washington^w^ Festuca^ the removal of dominants State^1600-^ and reciprocal transplants survive best in low 1700 m^ productivity sites. This indicates that productivity sites and competition increases with productivity. Fetcher^tussock^65° 26' N FE, R, m^Eriophorum^ Removal of shrubs increased (1985)^tundra in^145° 30'^vaginatum^ the growth of new tillers, central^yy^ likely by reduced shading as Alaska^730 m^ opposed to increased nutrient availability. Gerdol et^subalpine^44° 08' N FE, R, N^Vaccinium^+ , 0, -^Removal of shrubs had  8  AUTHORS LOCATION LAT. ^METHOD SPECIES^RESULTS INTERACTIONS AND LONG.^STUDIED^ MECHANISMS ALT. al. (2000)^heathland in^10° 35' E^uliginosum,^ negative, neutral or positive Italy^1900m^Vaccinium^ effects depending on species myrtillus,^ identity. Increased nutrient Empetrum^ availability did not increase hermaphroditum^plant mass for any species. Heilbronn^sub-antarctic 54° 17' S FE (obs.) ^Phleum alpinum,^+^Phleum colonizes sorted and Walton fellfield^37° 27'^Deschampsia^ stone stripes first and acts to (1984)^ antarctica,^ stabilize the soil which allows W^ 15-120 m^Festuca contracta,^others species to colonize. Acaena  tenera  Hobbie et^tussock^68° 38' N FE, R^Carex bigelowii,^0 , -^Most species showed no al. (1999)^tundra in^149° 34'^Eriophorum^ response to removal of Alaska^W^ vaginatum,^ another species. Ledum increased with the removal of 760 m^Betula nana,^ Ledum palustre,^other shrub species and Vaccinium vitis-^Sphagnum increased when idaea,^ Betula was removed. mosses Houle^subarctic^55° 17' N FE, R^Honchenya^0^No evidence for interactions observed. ( 1997 )^coastal dune 77° 46' ^peploides,^ in northern^W^ Elymus mollis, Quebec^18 m^Lathyrus japonicus Jonasson^tundra in^68° 21' N FE, R, N^all plants,^+ , 0^Removal of dominant (1992)^Sweden^18° 49' E^Vaccinium^ deciduous shrubs did not affect species diversity 400 m^myrtillus,^ Betula nana^ indicating no competitive suppression. Vaccinium and lichens declined after removal of Betula indicating a possible positive interaction. Maillette^forest to^47° 40' N FE (obs.)^Vaccinium^+^Close proximity to dominant (1988)^alpine^70° 37'^angustifolium,^ Vaccinium species increases tundra in^W^ V. myrtilloides,^ the nondominant one that are Quebec^965 m^V. uliginosum^ outside of their optimal range. McGraw^seeds^N/A^LE, S, m^Dryas octopetala^Two different ecotypes of (1985)^collected^ Dryas demonstrated differing from Alaska competitive abilities when grown in a high resource environment. McGraw^arctic^64° 51' N FE, LE,^Eriophorum^ Each species grew best and and Chapin floodplain^147° 43'^S, A, m^vaginatum,^ competed best in the (1989)^and muskeg W^ E. scheuchzeri^ environmental conditions that in Alaska^300 m^ they typically dominant. Therefore adaptation to low or high resource levels may improve the ability to compete at that resource level. Moen^high alpine^69° 56' N FE, A, m^Ranunculus^ Seedlings of Oxyria survive (1993)^blockfield in^22° 55' E^g/acialis,^ better when not grown close Norway 850 m Oxyria digyna to Ranunculus possibly due to soil temperature being greater in sites cleared of vegetation. Morris and^sub-alpine^N/A^FE, A^Lupinus lepidus,^+^For the first year after Wood^volcanic^ Anaphalis^ transplant into Lupinus (1989)^debris in^ magaritacea,^ patches, the other species Washington^ Epilobium^ survived at a lower rate. State^ angustifolium^ Afterwards, both species grew larger than the controls. Olofsson et alpine in^69° 56' N FE, R^Ranunculus^ Ranunculus negatively al. (1999)^Norway^22° 55' E^g/acialis,^ affected the germination, 850-^ Oxyria digyna^ growth and survival for Oxyria. 1100m^ Sammul et^boreal^58° 35' N FE, R^Anthoxanthum^ The removal of neighbours  9  AUTHORS LOCATION LAT.^METHOD SPECIES^RESULTS INTERACTIONS AND LONG.^STUDIED^ MECHANISMS ALT. al. (2000)^woods and^23° 34' E^ordoratum^ had a positive effect on the meadows in^< 25 m^ study species. As the Estonia and^69° 46' N^ community productivity tundra in^23° 58' E^ increased, so did both the Norway^380-^ intensity and importance of competition. 600m^ Sammul et^boreal^58° 35' N FE, R^Solidago virgaurea -^The removal of neighbours al. (2006)^woods and^23° 34' E^ had positive effect on the meadows in^< 25 m^ study species. Competition Estonia and^69° 46' N^ intensity increases nontundra in^23° 58' E^ linearly with productivity and Norway^380-^ at low productivity interactions may be 600m^ mutualistic (though not detected in this study). Shevtsova^sub-arctic^69° 45' N FE, R^Empetrum nigrum, + , -^Positive and negative et al.^shrub tundra 27° 01' E^Vaccinium vitis-^responses were observed (1995)^in Finland^85 m^idaea,^ after the removal of other V uliginosum,^ shrub species. These were V. myrtilloides,^species specific (see their Ledum palustre^ Fig. 8). Shevtsova^sub-arctic^69° 45' N FE, R^Empetrum nigrum, + , -^Positive and negative et al.^shrub tundra 27° 01' E^Vaccinium vitis-^responses were observed (1997)^in Finland^85 m^idaea^ after the removal of neighbours depending on the environmental manipulations also applied. Sohlberg^high arctic^77° 44' N FE, R, m^Ranunculus^ Both species demonstrated and Bliss^meadow in^101° 10'^sabinei, Papaver^better growth with moss (1987)^Canada^W^ radicatum, moss^removed likely due to alteration in microclimate. Theodose^alpine^N/A^FE, R, A,^Kobresia^ Removal of Kobresia in the and^tundra in^3150 m^N, m^myosuroides,^ more productive environment Bowman^Rockies of^ Deschampsia^ allowed the Deschampsia to (1997)^Colorado^ caespitosa^ attain higher biomass. The Deschampsia responded better to neighbour removal than to N addition. Virtanen^alpine heath 69° 01' N FE, A, R^Vaccinium^0^The absence of neighbours (1998)^in Finland^20° 50' E^myrtillus^ did not affect the performance of Vaccinium. 600 m^ Wilson^alpine heath 38° 28' S FE, A, m^Eucalyptus^ The strongest belowground (1993)^and^148° 18'^pauciflora,^ competitive ability had the grassland in ^E^ Poa costiniana,^highest root:shoot ratio and Australia^1830 m^Celmisia longifolia^were found at the highest altitudes. Results suggest that belowground competition may exclude inferior competitors from the high altitude grassland.  THESIS OVERVIEW In this thesis, I will use the CDS and test the role of biotic interactions, primarily competition, in structuring the understory vegetation of the bo real forest near Kluane Lake in the southwestern Yukon. First, I will present a traditional neighbour removal experiment and monitor the responses of seedlings to the presence and absence of  10  neighbouring vegetation. This will test the role of biotic interactions on 10 common understory species at varying levels of productivity created by adding fe rtilizer and water. The second chapter will further develop the CDS method. Next, I will present a greenhouse CDS experiment using 8 com mon species that will simulate the understory vegetation of the main study site to test the importance of biotic interactions and the role of productivity by adding fertilizer and water. The main and final chapter of the thesis will present data from the field CDS which ran for 4 years and involved the 9 most com mon species (all perennial) of the boreal understory in the Kluane region. This chapter will examine the role of density dependence on this community and the effect of varying the productivity with fertilizer addition.  11  REFERENCES Aarssen, L.W. and Epp, G.A., 1990. Neighbour manipulati ons in natural vegetation: a review. Journal of Vegetation Science, 1: 13-30. Abrams, P.A., 1995. Monotonic or unimodal diversity-productivity gradients: what does competition theory predict? Ecology, 76(7): 2019-2027. Aerts, R., Berendse, F., de Caluwe, H. and Schmitz, M., 1990. Competition in heathiand along an experimental gradient of nutrient availability. Oikos, 57: 310-318. Aksenova, A.A., Onipchenko, V.G. and Blinnikov, M.S., 1998. Plant interactions in alpine tundra: 13 years of experimental removal of dominant species. Ecoscience, 5: 258-270. Arii, K. and Turkington, R., 2002. Do nutrient availability and competition limit plant growth of herbaceous species in the boreal forest understory? Arctic, Antarctic, and Alpine Research, 34: 251-261. Austin, M.P. and Austin, B.O., 1980. Behavior of experimental plant communities along a nutrient gradient. Journal of Ecology, 68: 891-918. Austin, M.P., Groves, R.H., Fresco, L.M.F. and Kaye, P.E., 1985. Relative growth of six thistle species along a nutrient gradient with multispecies competition. Journal of Ecology, 73: 667-684. Bailey, R.C., 1989. Plant competition. Nature, 337: 122. Belcher, J.W., Keddy, P.A. and Twolan-Strutt, L., 1995. Root and shoot competition intensity along a soil depth gradient. Journal of Ecology, 83: 673-68 2. Bertness, M.D., 1998. Searching for the role of positive interactions in plant communities. Trends in Ecology and Evolution, 13: 133-134. Bertness, M.D. and Callaway, R., 1994. Positive interactions in communities. Trends in Ecology and Evolution, 9(5): 191-193. Blundon, D.J., Maclsaac, D.A. and Dale, M.R.T., 1993. Nucleation during primary succession in the Canadian Rockies. Canadian Journal of Botany, 71: 10931096. Bonde, E.K., 1968. Survival of seedlings of an alpine clover ( Trifolium nanum Torr.). Ecology, 49: 1193-1195. Brooker, R.W. and Call aghan, T.V., 1998. The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oi kos, 81(1): 196-207. Callaway, R.M., 1995. Positive interactions among plants. Botanical Review, 61(4): 306349. Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., Michalet, R., Paolini, L., Pugnaire, F.I., Newingham, B., Aschehoug, E.T., Armas, C., Kikodze, D. and Cook, B.J., 2002. Positive interactions among alpine plants increase with stress. Nature, 417: 844-848. Campbell, B., D. and Grime, J.P., 1992. An experimental test of plant strategy theory. Ecology, 73(1): 15-29. Carlsson, B.A. and Callaghan, T.V., 1991. Positive plant interactions in tundra vegetation and the importance of shelter. Journal of Ecology, 79: 973-983. Chapin, F.S., Ill, 1980. The mineral nutrition of wild plants. Annual Review of Ecology and Systematics, 11: 233-260. Chapin, F.S., Ill, McGraw, J.B. and Shaver, G.R., 1989. Com petition causes regular spacing of alder in Alaskan shrub tun dra. Oecologia, 79: 412-416. Connolly, J., 1986. On difficulties with replacement-series methodology in mixture experiments. Journal of Applied Ecology, 23: 125-137.  12  Connolly, J., 1997. Substitutive experiments and the evidence for competitive hierarchies in plant communities. Oikos, 80(1): 179-182. Cousens, R., 2000. Greenh ouse studies of interactions between plants: the flaws are in interpretation rather than design. Journal of Ecology, 88: 352-353. del Moral, R., 1983. Competition as a control mechanism in subalpine meadows. American Journal of Botany, 70(2): 232-245. Fetcher, N., 1985. Effects of removal of neighboring specie s on growth, nutrients, and microclimate of Eriophorum vaginatum. Arctic and Alpine Research, 17: 7-17. Firbank, L.G. and Watkinson, A.R., 1989. Plant competition. Nature, 337: 122. Gerdol, R., Brancaleoni, L., Menghini, M. and Marchesini, R., 2000. Response of dwarf shrubs to neighbour removal and nutrient addition and their influence on community structure in a subalpine heath. Journal of Ecology, 88: 256-266. Gibson, D.J., Connolly, J., Hartnett, D.C. and Weidenhamer, J.D., 1999. Designs for greenhouse studies of interactions between plants. Journal o f Ecology, 87(1): 116. Goldberg, D. and Novoplansky, A., 1997. On the relative importance of competition in unproductive environments. Journal of Ecology, 85: 409-418. Goldberg, D.E., 1990. Corn ponents of resource competition in plant communities. In: J. Grace and D. Tilman (Editors), Perspectives in plant competition. Academic Press, San Diego, pp. 27-4 9. Goldberg, D.E., 1994. Influence of competition at the community level: an experimental version of the null models approach. Ecology, 75(5): 1503-1506. Goldberg, D.E. and Barton, A.M., 1992. Patterns and consequences of inte rspecific competition in natural communities: a review of field experiments with plants. American Naturalist, 139(4 ): 771-801. Goldberg, D.E., Turkington, R. and Olsvig-Whittaker, L., 1995. Quantifying the community-level consequences of competition. Folia Geobotanica Phytotaxonomica, 30: 231-242. Goldberg, D.E., Turkington, R., Olsvig-Whittaker, L. and Dyer, A.R., 2001. Densitydependence in an annual plant community: variation among life history stages. Ecological Monographs, 71: 423- 446. Grace, J.B., 1990. On the relationship between plant traits and competitive ability. In: J.B. Grace and D. Tilman (Editors), Perspectives on.plant competition. Academic Press, New York, pp. 51-65. Grace, J.B., 1993. The effects of habitat productivity on competition intensity. Trends in Ecology and Evolution, 8(7): 229-230. Grace, J.B. and Clark, J.S., 1991. Com petition and life history in the Grime/Tilman debate. Vegetation Section New sletter, 10(1): 3-8. Grace, J.B., Janet, K. and Guntenspergen, G.R., 1992. Size bias in traditional analyses of substitutive competition experiments. Oecologia, 90: 429-434. Grime, J.P., 1977. Evidence for the existence of three primary strategies in plants an d its relevance to ecologica I and evolutionary theory. American Naturalist, 111: 11 691194. Grime, J.P., 1979. Plant strategies and vegetation processes. John Wiley, Chichester, UK. Grubb, P.J., 1985. Plant populations and vegetation in relation to habitat, disturbance and competition: problems of generalization. In: J. White (Editor), The population structure of vegetation. Dr W. Junk Publishers, Dordrecht, pp. 595-621. Gurevitch, J., 1986. Com petition and the local distribution of the grass Stipa neomexicana. Ecology, 67: 46-57.  13  Gurevitch, J. and Unnasch, R.S ., 1989. Experimental removal of a dominant species at two levels of soil fertility. Canadian Journal of Botany, 67: 3470-3 477. Gurevitch, J., Wilson, P., Stone, J.L., Teese, P. and Stoutenburg h, R.J., 1990. Competition among old-field pere nnials at different levels of soil fertility and available space. Journal of Ecology, 78: 727-744. Harper, J.L., 1977. Population biology of plants. Chapman & Hall, London, UK. Heilbronn, T.D. and Walton, D.W.H ., 1984. Plant colonization of actively sorted stone stripes in the Subantarctic. Arctic and Alpine Research, 16(2): 161-172. Hobbie, S.E., Shevtsova, A. and Chapin, F.S., III, 1999. Plant responses to species removal and experimental warming in Alaskan tussock tundra. Oikos, 84(3): 417434. Houle, G., 1997. No evidence for int erspecific interactions between plants in the first stage of succession on coastal dunes in subarctic Quebec, Canada. Canadian Journal of Botany, 75: 902-915. Huston, M., 1979. A general hypothesis of species diversity. American Naturalist, 113(1): 81-101. Jolliffe, P.A., 2000. The replacement series. Journal of Ecology, 88: 371-385. Jonasson, S., 1992. Plant responses to fertilization and species rem oval in tundra related to community structure and clonality. Oikos, 63: 420-429. Kadmon, R., 1995. Plant competition along soil moisture gradients: a field experiment with the desert annual Stipa capensis. Journal of Ecology, 83: 253-262. Kareiva, P., 1994. Higher order interactions as a foil to reductionist ecology. Ecology, 75(6): 1527-1528. Keddy, P.A., 1989. Competition. Chapman and Hall, New York. Keddy, P.A., 1990. Competitive hierarchies and centrifugal organization in plant communites. In: J. Grace and D. T ilman (Editors), Perspectives on plant competition. Academic Press, San Diego, pp. 265-290. Lortie, C.J. and Turkington, R., 2002. The effect of initial seed density on the structure of a desert annual plant community. Journal of Ecology, 90: 435-445. Maillette, L., 1988. Apparent commensialism among three Vaccinium species on a climatic gradient. Journal of Ecology, 76: 877-888. McGraw, J.B., 1985. Experimental ecology of Dryas octopetala ecotypes: relative response to cometitors. New Phytologist, 100: 233-241. McGraw, J.B. and Chapin, F .S., Ill, 1989. Competitive ability and adaptation to fertile and infertile soils in two Eriophorum species. Ecology, 70(3): 736-749. Miller, T.E., 1996. On quantifying the intensity of competition across gradients. Ecology, 77(3): 978-981. Moen, J., 1993. Positive versus negative interactions in a high alpine block field: germination of Oxyria digyna seeds in a Ranunculus glacialis community. Arctic and Alpine Research, 25(3): 201-206. Morris, W.F. and Wood, D.M., 1989. The role of lupine in succession on Mount St. Helens: facilitation or inhibition. Ecology, 70(3): 697-703. Newman, E.I., 1973. Competition and diversity in herbaceous vegetation. Nature, 244: 310. Olofsson, J., Moen, J. and Okansen, L., 1999. On the balance between positive and negative plant interactions in harsh environments. Oikos, 86: 539-543. Rajaniemi, T.K. and Goldberg, D.E., 2000. Quantifying individual- and co mmunity-level effects of competition using experimentally determined null species pools. Journal of Vegetation Science, 11: 433-442.  14  Reader, R.J. and Best, B.J., 1989. Variation in competition along an environmental gradient: Hieracium floribundum in an abandoned pasture. Journal of Ecology, 77: 673-684. Reader, R.J., Wilson, S.D., Belcher, J.W., Wisheu, I., Keddy, P.A., Tilman, D., Morris, E.C., Grace, J.B., McGraw, J.B., Olff, H., Turkington, R., Klein, E., Young, Y., Shipley, B., van Hulst, R., Johansson, M.E., Nilsson, C., Gurevitch, J., Grigulis, K. and Beisner, B.E., 1994. Plant competition in relation to neighbor biomass: an intercontinental study with Poa pratensis. Ecology, 75(6): 1753-1760. Sammul, M., Kull, K., Oksanen, L. and Veromann, P., 2000. Competition intensity and its importance: results of field experiments with Anthoxanthum odoratum. Oecologia, 125: 18-25. Sammul, M., Oksanen, L. and Magi, M., 2006. Regional effects on com petitionproductivity relationshi ps: a set of field experiments in two distant regions. Oikos, 112: 138-148. Shevtsova, A., Haukioja, E. and Ojala, A., 1997. Growth response of subarctic dwarf shrubs, Empetrum nigrum and Vaccinium vitis-idaea, to manipulated environmental conditions and species removals. Oikos, 78: 440-458. Shevtsova, A., Ojala, A., Neuvonen, S., Vieno, M. and Haukioja, E., 1995. Growth and reproduction of dwarf shrubs in a subarctic plant community: annual variation and above-ground interactions with neighb ours. Journal of Ecology, 83: 263-275. Shilo-Volin, H., Novoplansky, A., Goldberg, D.E. and Turkington, R., 2005. Density regulation in annual plant com munities under variable resource levels. Oikos, 108: 241-252. Silvertown, J., 1989. Plant competition. Nature, 337: 122-123. Silvertown, J. and Dale, P., 1991. Corn petitive hierarchies and the structure of herbaceous plant communities. Oikos, 61(3): 441-444. Snaydon, R.W., 1991. Replacement of additive designs for competition studies. Journal of Applied Ecology, 28: 930-946. Sohlberg, E.H. and Bliss, L.C., 1987. Responses of Ranunculus sabinei and Papaver radicatum to removal of the moss layer in a high-arctic meadow. Canadian Journal of Botany, 65: 1224-1228. Taylor, D.R. and Aarssen, L.W., 1989. On the density dependence of replacementseries competition experiments. Journal of Ecology, 77: 975-988. Taylor, D.R., Aarssen, L.W. and Loehle, C., 1990. On the relationship between r/K selection and environmental carrying capacity: a new habitat template for plant life history strategies. Oikos, 58: 239-250. Theodose, T.A. and Bowman, W.D., 1997. The influence of interspecific competition on the distribution of an alpine grami noid: evidence for the importance of plan t competition in an extreme environment. Oikos, 79: 101-114. Tilman, D., 1980. Resources: a gra phical-mechanistic approach to competition and predation. American Naturalist, 116: 362-393. Tilman, D., 1982. Resource competition and community structure. Princeton University Press, Princeton. Tilman, D., 1988. Plant strateg ies and the dynamics and structure of plant communities. Princeton University Press, Princeton, NJ. Tilman, D., 1990. Constraints and tradeoffs: toward a predictive theory of competition and succession. Oikos, 58: 3-15. Turkington, R., Goldberg, D.E., Olsvig-Whittaker, L. and Dyer, A.E., 2005. Effects of density on timing of emergence and its consequences for survival and growth in two communities of annual plants. Journal of Arid Environments, 61: 377-396.  15  Turkington, R., Klein, E. and Chanway, C.P., 1993. Interactive effects of nutrients and disturbance: an experimental test of plant strategy theory. Ecology, 74(3): 863878. Virtanen, R., 1998. Im pact of grazing and neighbour removal on a heath plant community transplanted onto a snowbed site, NW F innish Lapland. Oikos, 81: 359-367. Weiher, E. and Keddy, P.A., 1995. The assembly of experimental wetland plant communities. Oikos, 73: 323-335. Wilson, S.D., 1993. Competition and resource availability in heath and grassland in the Snowy Mountains of Australia. Journal of Ecology, 81: 445-451. Wilson, S.D. and Keddy, P.A., 1986a. Measuring diffuse competition along an enironmental gradient: results from a shoreline plant community. American Naturalist, 127(6): 862-869. Wilson, S.D. and Keddy, P.A., 1986b. Species competitive ability and position along a natural stress/disturbance gradient. Ecology, 67: 1236-1242. Wilson, S.D. and Shay, J.M., 1990. Competition, fire, and nutrients in a mixed-grass prairie. Ecology, 71(5): 1959-1967. Wilson, S.D. and Tilman, D., 1991. Components of plant competition along an experimental gradient of nitrogen availability. Ecology, 72(3): 1050-1063. Wilson, S.D. and Tilman, D., 1993. Plant competition and resource availability in response to disturbance and fertilization. Ecology, 74(2): 59 9-611. Wilson, S.D. and Tilman, D., 1995. Competitive responses of eight old-field plant species in four environments. Ecology, 76(4): 1169-1180. Zamfir, M. and Goldberg, D.E., 2000. The effect of initial density on interactions betwe en bryophytes at individual and community levels. Journal of Ecology, 88: 243-255.  16  Chapter 2  FACILITATION IN A BOREAL FOREST UNDERSTORY: RESULTS FROM A NEIGHBOUR REMOVAL, FERTILIZATION AND WATERING EXPERIMENT' INTRODUCTION  Traditionally, studies on the factors structuring plant communities have focused on negative interactions such as competition (Connell 1983; Schoener 1983; Keddy 1989; Grace and Tilman 1990) and have tended to neglect positive interactions such as facilitation (Brooker and Callaghan 1998). Only recently have researchers begun to thoroughly investigate the potential role of facilitation in community structure, especially in unproductive environments (Bertness and Callaway 1994; Callaway 1995; Bertness 1998; Brooker and Callaghan 1998). In productive environments, plants may compete for resources because conditions are favourable, but if environmental conditions are harsh, the ability to capture resources may be restricted and any amelioration of these conditions by neighbouring plants may outweigh potential negative interactions (Brooker and Callaghan 1998). Others have argued that competition is also important in these unproductive environments, though the particular plant strategies that would be optimal under those conditions are different than in more productive environments (Newman 1973; Tilman 1982; Tilman 1988). The boreal forest of is generally considered to be limited by soil nutrients and can vary from extremely dry to extremely wet and from cold and harsh to relatively temperate (Larsen 1980; Bonan and Shugart 1989). The boreal region of the southwestern Yukon is dry, cool, and nutrient limited (Turkington et al. 1998; Turkington et al. 2002), and is thought to be water limited, although there is little evidence to suggest this (Carrier and Krebs 2002). Therefore, this region is an appropriate location to test the role of biotic interactions in low productivity habitats. Much of the focus on plant interactions in 1 A version of this chapter will be submitted for publication. Treberg, M. A. and Turkington, R. (2008). Facilitation In A Boreal Forest Understory: Results From A Neighbour Removal, Fertilization And Watering Experiment.  17  stressed low productivity systems comes from arid ecosystems (Fowler 1986; Goldberg et al. 2001); however, for a review of some studies focusing on biotic interactions in cold, harsh environments see Arii and Turkington (2002). Specifically, I tested the hypothesis that if competition were important in structuring this boreal forest understory community, the removal of neighbours, the addition of fertilizer and addition of water would all benefit the transplanted seedlings. Alternatively, if facilitation were more important, the removal of neighbours would be a detriment to the transplants but fertilization and watering would still be beneficial.  METHODS  STUDY SITE The study site is located within the boreal forest close to Kluane Lake in the southwestern Yukon Territory (138 ° 16' W; 61 ° 00'N) at an approximate altitude of 1000 m above sea level. The closest climate station is Burwash Airport, YT, approximately 52 km to the north. At Burwash the yearly total precipitation is 279.7 mm, of which 192.1 mm is rainfall (Environment Canada 1971-2000 Climate Normals). The 30-year normal precipitation (including some snow) for the months during which the study took place (June, July, August) is 158.2 mm. In 2001, the first year of the study, this value was 149.2 mm. In 2002, 104.2 mm of precipitation fell in July and August and no data are available for June. For these months the 30-year normal daily average temperature was 11.1 °C and the daily maximum and minimum were 18.0 °C and 4.7 °C respectively. White spruce (Picea glauca (Moench) Voss s.l.) is the dominant tree species with a density of 583 stems ha -1 (95% CI of 486 to 697 ha -1 ). Many willows (predominantly Salix glauca L. s. I.) and some dwarf birch (Betula glandulosa Michx.) make up the shrub understory. The ground layer or understory species had a mean aboveground biomass of 196 g rri 2 (n = 9 1 m 2 quadrats, 95 % CI of 141 to 250 g m-2 ) in 2002. The cover of moss and lichens is < 5 %. Although there are herbivores such as snowshoe hares, red squirrels and microtine rodents at this site, the abundance of the ground layer is more affected by the limited soil nutrients than by herbivores (John and Turkington 1995; John and Turkington 1997; Turkington et al. 2002).  18  STUDY SPECIES Ten of the most common understory species were chosen as representative species. Eight of the species are herbaceous perennials: Achillea millefolium L. ssp. borealis (Bong.) Breitung (yarrow), Anenome parvitiora Michx. (northern anemone), Epilobium angustifolium L. s.l. (fireweed), Festuca altaica Trin. (northern rough fescue), Lupinus arcticus Wats. (arctic lupine), Mertensia paniculata (Ait.) G. Don var. paniculata  (bluebells), Senecio lugens Richards. (black-tipped groundsel), Solidago multiradiata Ait. (goldenrod). The remaining two species are woody perennials: Arctostaphylos uva-ursi (L.) Spreng. s.l. (bearberry) and Linnaea borealis L. ssp. americana (twinflower). Hereafter I will refer to species using their generic name. All nomenclature follows Cody (1996). EXPERIMENTAL DESIGN The role of biotic interactions in structuring the boreal forest understory was tested using a 2x2x2 fully factorial design using 2 levels of each treatment (neighbour removal, fertilization, and watering), 10 understory species and 3 replicates per treatment. In mid-June of 2001 (after snowmelt), I laid out 24 transects, each 3.3 m long and 0.3 m wide with at least 1 m separating each transect. Half of the transects were randomly assigned to the neighbour removal treatment and were sprayed with a nonselective systemic herbicide (RoundUpTM, active ingredient Glyphosate) which is known to quickly bind to soil particles, become inactive and is ultimately broken down by soil microbes (see Appendix 1 for more details) . Weekly, a 1:20 ratio of Glyphosate to water was sprayed on the neighbour removal transects until vegetation was completely covered with solution. This was repeated until all plants within the 3.3 m x 0.3 m area were dead. Some species were resistant to Glyphosate and complete death did not occur until mid-July; litter was not removed. Throughout the remainder of the study to August 2002, no regrowth occurred in the sprayed areas and all natural seedling recruits were removed manually. Transplants were prepared in late May or early June using seeds collected the previous year. Seeds of all species except Arctostaphylos and Linnaea were germinated in peat plugs that were 2.5 cm wide by 5 cm deep. I was unable to germinate the seeds of these two species and instead used cuttings grown in peat plugs to make the transplants. The cuttings were taken from healthy plants growing close to  19  the study site. Each was 5 cm long with half of the length left as leaves and half stripped of leaves and treated with a growth hormone (Wilson Roots® Liquid Root Stimulator) to promote root growth. An individual transplant of each of the 10 study species was planted into each transect in mid-July 2001 after all plants in the removal transects had died. The transplants were planted down the center line of the 3.3 m transect approximately 0.3 m apart. Transplants that died within the first 10 days were replaced. At the beginning of the 2002 season (mid-June) all transplants that had died during the previous season or overwinter were replaced with seedlings germinated in May 2002, or in the case of Arctostaphylos and Linnaea, with fresh cuttings. Water-soluble fertilizer (Plant Prod® 20-20-20) was applied as a solution every two weeks at 4 mL fertilizer per 1 L H 2 O for the high fertilization treatment. The total application of nitrogen, phosphorus and potassium was 12 g m 2 for the 2002 season. In -  2001, approximately half of this amount was added. This fertilizer addition rate corresponds with other experiments conducted in this system, which demonstrated significant vegetation responses to nutrient addition (Turkington et al. 1998; Dlott and Turkington 2000; Arii and Turkington 2002; Turkington et al. 2002). The low fertilization treatment had no fertilizer added to serve as a control. More naturally occurring seedlings were observed in the field in years of higher than average rainfall such as in 2000 when there was about 54% more rainfall than usual. Yet, I required additional water for successful transplant establishment. Therefore, for the watering treatments, water was added at a rate of 3 L m 2 (which -  corresponds with a rainfall of 3 mm) for the low level and 6 L m  -2  (6 mm of rain) for the  high level. Water was added every other day through the season and the total application was approximately 120 and 240 mm for the low and high treatments respectively for 2002. The previous season was shorter due to site preparation and as a result only about half of this total amount was added. Environmental variables were measured at three times throughout both growing seasons. Volumetric water content (VWC) was measured using a CS620 HydroSense® Water Content Sensor (Campbell Scientific Inc., Logan, UT). Soil temperature at 5 cm was measured using a digital thermometer (model TPD 32, OMEGA, Laval, QC). Two measures of the abundance of neighbouring plants along the transect were estimated. The leaf area index (LAI) was calculated by lowering a pin through the vegetation every 10 cm along the 3.3 m long transect and recording the number of times the pin intercepted leaves. I also estimated the percent of light transmitted through the  20  vegetation by measuring the incoming photosynthetically active radiation (PAR) above the vegetation and at the soil surface at 3 points along the transect. PAR was measured using a line quantum sensor (Apogee Instruments Inc., Logan, UT). The three sets of measurements for each environmental variable for 2002 were averaged for analysis. STATISTICAL ANALYSIS The effects of neighbours, fertilization and watering were analyzed for survival and total biomass of all species in each transect combined using the standard 2x2x2 fully factorial design analysis of variance (ANOVA) with SYSTAT (SYSTAT Software Inc 2002). ANOVAs were also done on the biomass of the 10 species individually. During preliminary data analysis, the planting date of each transplant was included as a covariate to remove size effects due to differences in age. But, for all of the species, planting date was not a statistically significant covariate and those analyses are not presented here. For each ANOVA, the usual assumptions of independence, homogeneity of variance and normality were checked. All data conform to the assumption of independence. The survival of species is a proportion and are usually transformed prior to ANOVA; I did not do this because most of the data fell within the middle range of 0.3 to 0.7 and it is therefore not necessary to transform the data if assumptions about variance and normality are met (Zar 1984). Most species had individuals die in some of the transects. This led to zero values and skewed distributions when the treatment effects on the biomass of the individuals of each species were individually tested in ANOVA. To remove some of the skewness and to meet the assumption of homogeneity of variance, all of the biomass data for each individual species were square root transformed with X' = (X) % + (X+1)'h . This transformation is useful when data values are small (X 2) and contain some zero values (Zar 1984). Although the data have variances that are similar (and this is the more important assumption for ANOVA), the normality assumption has not been met for any of the 10 study species. However, this is not a concern because ANOVA is robust to the violation of the normality assumption especially when the experiment is large (having either many samples or many treatments) and the samples are balanced, such as in this experiment (Zar 1984; Underwood 1997). Differences between treatment means, or multiple comparisons, were tested using Tukey's HSD (honestly significant difference) test (Zar 1984).  21  RESULTS  The summed survival of all transplants (Table 2.1a, Fig. 2.1) and their total biomass (Table 2.1b, Fig. 2.2) was significantly higher in the presence of neighbours (P < 0.05). Both the combined survival and the combined biomass of all species showed an increase with fertilization (P = 0.058 and P = 0.056 respectively), and survival increased with watering (P = 0.096, Table 2.1).  Table 2.1. Summary of ANOVA for a) the summed percent survival of all transplants along a transect and for b) the summed biomass of all transplants along a transect. Main effects are Neighbours (N), Fertilization (F) and Watering (W). Values in bold are significant at P < 0.05, and those in italics are significant at P < 0.10.  Source  df  MS  F-ratio  P  a) Survival N F W NxF NxW FxW NxF xW Error  1 1 1 1 1 1 1 16  70.042 9.375 7.042 0.042 0.375 1.042 1.042 2.250  31.130 4.167 3.130 0.019 0.167 0.463 0.463  <.001 0.058 0.096 0.894 0.689 0.506 0.506  b) Biomass N F W NxF NxW FxW NxF xW Error  1 1 1 1 1 1 1 16  0.407 0.085 0.014 0.022 0.000 0.005 0.008 0.020  20.364 4.240 0.716 1.093 0.001 0.230 0.393  <.001 0.056 0.410 0.311 0.973 0.638 0.540  22  o no neighbours  100  Eg neighbours  90 80 70 — 60 > 50 40 cr) 30 20 10 0 Fl W1  ^  F1 W2^F2 W1^F2 W2  Figure 2.1. The percent survival (± S.E.) of all transplants summed per transect with or without neighbours at low (F1) and high (F2) fertilizer addition and at low (W1) and high (W2) watering. 0.7 0.6 0.5 -  o no neighbours E2 neighbours  en cn 0.4 co  cg 0.3 0  0.2 0.1 0 F1 W1^F1 W2  F2 W1  F2 W2  Figure 2.2. The total biomass (g t S.E) of all transplants summed per transect with or without neighbours at low (F1) and high (F2) fertilizer addition and at low (W1) and high (W2) watering.  Transplants of 6 of the 10 species had significantly higher biomass with  neighbours than without (Table 2.2). However, only 2 species had significant responses to fertilization with Anenome decreasing and Mertensia increasing in biomass (Table 2.2). Watering increased the biomass of 3 species, Achillea, Festuca and Solidago (Table 2.2). Arctostaphylos, Festuca and Solidago have interaction effects so some caution is required in interpretation. For example, Festuca had all first-order interactions significant and both neighbours and watering significant (Table 2.2). The interactions of  23  neighbours and fertilization, and of neighbours and watering, both show higher biomass in the higher fertilization or watering treatments but only with neighbours present (Fig. 2.3 a,b). The interaction between water and nutrients indicates higher biomass for Festuca but only when nutrients and water were high (Fig. 2.3c). Therefore, the main  treatment effects that were significant, neighbours and watering, were due largely to the interaction effects. A 2nd order interaction (F x W) was significant for Solidago (Table 2.2). The high fertilizer and high water with neighbours treatment had significantly higher biomass for Solidago than for all other treatments with the exception of the low fertilizer and low water with neighbours treatment, which was the same as all others (Fig. 2.4a). The high and low water treatments were different within the high fertilizer effect, but there was no difference between any of the other nutrient and water treatments (Fig. 2.4b). As for Festuca, the significant effects of the neighbour and water treatments for Solidago are due to the significant interaction terms.  Table 2.2. Probabilities derived from ANOVAs for the biomass of each of the 10 species. Values in bold values are significant (P < 0.05). Species Achillea millefolium Anenome parviflora Arctostaphylos uva-ursi Epilobium angustifolium Festuca altaica Linnaea borealis Lupinus arcticus Mertensia paniculata Senecio lugens Solidago multiradiata  N 0.010 0.904 <.001 0.712 0.002 0.012 0.023 0.172 0.580 <.001  F 0.251 0.008 0.827 0.309 0.110 0.803 0.823 0.037 0.505 0.833  W 0.024 0.337 0.892 0.890 0.012 0.430 0.462 0.427 0.457 0.023  NxF 0.801 0.117 0.562 0.520 0.011 0.696 0.823 0.105 0.948 0.118  NxW 0.388 0.208 0.618 0.393 0.003 0.163 0.462 0.278 0.752 0.098  FxW 0.088 0.593 0.453 0.591 0.048 0.163 0.064 0.806 0.908 0.033  NxFxW 0.594 0.829 0.020 0.217 0.277 0.430 0.064 0.414 0.616 0.007  24  a)  1.25 iii^ co u)^1.2 -;  E  0 low fertilizer rdhigh fertilizer  1.15  no^yes Neighbour  b)  1.25 -  o low water F2 high water  no^yes Neighbour  c) 0, 2  --  E  0  0  E  1 25  o low water 1.2  E2 high water  1.15 1 1.1  ab  1.05 -  low^high Fertilizer  Figure 2.3. Significant interactions from the ANOVA in Table 2.2 between a) neighbour and fertilization, b) neighbour and watering and c) fertilization and watering on the transformed biomass (g ± S.E.) of Festuca. Columns sharing the same letter are not significantly different (P > 0.05) as determined by Tukey's HSD.  25  a^  13  )  1 25  o no neighbours E3 neighbours  ab  1.1^j 0  c 1.05 Ea  0.95 1Fl W1  F1 W2^F2 W1  F2 W2  Treatment  b  1.3 -  o low water 1.25  el high water  1.2  b  1.15 1.1  ab  ab  1.05  0.95 high  low Fertilizer  Figure 2.4. Significant interactions from the ANOVA in Table 2.2 between a) fertilization X watering X neighbours and b) watering X fertilization on the transformed biomass (g ± S.E) of Solidago. Columns sharing the same letter are not significantly different (P > 0.05) as determined by Tukey's HSD.  The treatments had little statistically significant effects on the environmental variables measured in the transects (Table 2.3). The VWC was significantly higher in the high water treatment (mean = 16.4 %, 95% CI 14.4 to 18.4%) than the low water treatment (mean = 12.6%, 95% CI 11.0 to 14.2%). There was significantly more light  26  transmitted to the surface in the neighbour removal transects, though this is not surprising given that there was no live vegetation. The fertilization and watering had no effect on either the % transmittance or LAI.  Table 2.3 Probabilities derived from ANOVAs for the environmental variables monitored in the transects. Neighbours are not included in the ANOVA for LAI because there were no live plants in the neighbour removed plots. Values in bold are significant (P < 0.05). Variable  N  F  W  NxF  NxW  FxW  Soil temperature VWC % transmittance LAI  0.746 0.815  0.888 0.397 0.622 0.800  0.556  1.000 0.519 0.523  0.388 0.605 0.697 0.983  0.231 0.795 0.764  <0.001  0.008 0.637 0.723  NxFxW 0.352 0.471 0.917  DISCUSSION THE IMPORTANCE OF NEIGHBOURS The presence of neighbours did not show a suppressing or competitive effect on the survival or growth of transplants. Rather, their presence significantly increased the survival and the biomass of most of the transplants while fertilization and watering slightly increased the survival and biomass. Given the results reported from previous research conducted near this site, I would not have predicted the importance of neighbours to the success of seedling establishment and growth. All of these study species have been included in previous experiments examining the role of nutrient availability and herbivory (John and Turkington 1997; Turkington et al. 1998; Dlott and Turkington 2000; Turkington et al. 2002) and four of the species have been included in a study specifically examining the role of competition (Arii and Turkington 2002). Competitive interactions have been inferred from many of the patterns observed in these experiments and has been demonstrated for Achillea (Arii and Turkington 2002). Also, Lupinus increased in percent cover and decreased in leaf mortality when neighbours were removed (Graham and Turkington 2000). Here, I assumed that if I expanded the number of species to include more than the 4 species examined in Arii and Turkington (2002), that I would at least see additional evidence that the presence of neighbours had a negative impact on at least some of the species. Surprisingly, I found that both survival and biomass of most of the transplanted species are higher when neighbours are present indicating a positive interaction rather than a competitive one. The positive  27  effect of neighbours has been previously reported from an experiment conducted in alpine tundra less than 10 km from where this study took place (Callaway et al. 2002). Callaway et al. (2002) used elevation as a stress gradient and hypothesized that with increasing elevation, the importance of facilitation would also increase. They reported that the target species at both low and high elevation had a significantly positive neighbour effect and they argued that both elevations were highly stressed (Callaway et al. 2002). The idea that positive interactions become more important as stress increases in not new (Bertness and Callaway 1994; Callaway 1995; Callaway 1997; Brooker and Callaghan 1998). The positive effect of neighbours has been observed at many other low productivity high stressed sites in subarctic forests (Shevtsova et al. 1995; Shevtsova et al. 1997) in alpine tundra (Carlsson and Callaghan 1991; Aksenova et al. 1998; Gerdol et al. 2000) and in arctic tundra (Jonasson 1992). It has been argued that the relative importance of facilitation should increase with increasing environmental stress (Bertness and Callaway 1994; Callaway and Walker 1997; Brooker and Callaghan 1998). However, this has recently been criticized by Maestre et al. (2005) as not being consistent with field results from arid and semi-arid systems and that both facilitation and competition occur in stressful environments. A reexamination of these results suggests that there is insufficient evidence to reject the stress-gradient hypothesis (Lortie and Callaway 2006). A meta-analysis examining the role of positive and negative interactions along productivity gradients (Goldberg et al. 1999) reported that facilitative interactions were common at low standing crop when biomass and growth rates were examined. Given the abundance of literature that demonstrates the role of positive interactions in highly stressed systems (Bertness and Callaway 1994; Callaway 1995; Callaway and Walker 1997; Brooker and Callaghan 1998), it is clear that facilitation is important and our results add to the growing list of studies that support this conclusion. There have been a number of proposals to explain the positive effects of neighbours on survival and biomass. These include moderating temperature and water extremes by providing shade (Callaway 1995; Holmgren et al. 1997) although I did not detect any significant differences between neighbour or no neighbour transects in either the 5 cm soil temperatUre or soil volumetric water content (Table 2.3). Unfortunately, I did not determine the soil or plant surface temperature, measurements that would be more important to the individual transplants. The cover provided by even a very low canopy of herbaceous vegetation and low shrubs would absorb and reflect a great deal  28  of the incoming solar radiation and would highly modify the surface microclimate. For example, in a clearcut in a British Columbia Southern Interior forest, soil surface temperatures were significantly higher during the daytime when neighbours were removed using Glyphosate compared to untreated areas with intact vegetation; this was due to the interception of incoming solar radiation by the grass canopy layer (Fleming et al. 1998). It would not be unreasonable to assume that surface temperatures of our transplants were also much higher in the absence of a protective canopy of neighbours. The benefits of the presence of neighbours do not negate the possibility or even probability that negative interactions also occur between the transplants and the canopy plants (Holmgren et al. 1997). In this particular case it is clear that the net outcome for transplants is positive simply because of the increased chance of survival. Previous demonstrations of competitive interactions in this study location were with existing individuals (Graham and Turkington 2000; Arii and Turkington 2002) rather than with transplanted seedlings and cuttings. Differences in life stage may account for these differences. Seedlings may only be able to survive with a facilitative neighbour, with the interaction between species becoming negative once the seedling develops into adulthood (Callaway and Walker 1997). Relatively few studies have explicitly attempted to examine changes in biotic interactions as a function of life history stages and most of those studies focus on changes in competition intensity or ability (Howard and Goldberg 2001; Lamb and Cahill 2006). Goldberg et al. (2001) observed changes in the importance of facilitation and competition among life stages in desert annual plant species. They observed strong interference competition at the emergence stage, followed by neutral or mildly facilitative interactions at the survival stage and the growth stage was dominated by exploitive competition. Our study does not include the emergence stage; however, the survival and growth of seedlings was increased in the presence of neighbours indicating a primarily facilitative role. It would have been interesting to extend this experiment to determine if the role of neighbours switched from facilitative to more competitive as the seedlings became larger. THE UNIMPORTANCE OF FERTILIZATION In general, transplants tended to have higher survival and biomass when fertilized, yet only two species had a significant response. The overall lack of speciesspecific responses to fertilization is surprising given the previous positive responses to nutrient addition (Turkington et al. 1998; Turkington et al. 2002) and the known lack of  29  nutrients in the soil, particularly nitrogen which was previously measured as 0.005 g total N per kg of soil (Arii and Turkington 2002). Also, the leaf area index (LAI) of the fertilized vegetation did not increase in transects in which neighbours were not removed, a response normally expected in a nutrient limited system. The only transplanted species that responded positively, Mertensia, has been shown to respond favorably to the addition of fertilizer in other studies (John and Turkington 1997; Arii and Turkington 2002; Turkington et al. 2002). A possible explanation for the lack of response could be that seeds have their own initial stored resources; however, at least some of the seedlings survived and grew over the two seasons and would have depleted any stored reserves and would have benefited from fertilization if under nutrient stress. Another explanation is that the seedlings and transplants were too small to adequately capture the additional resources added to the system or in the case where neighbours were present, were "out-competed" for them. It has been speculated that plants in this community have such slow growth that they are unable to adequately capture short-term increases in nutrients (Graham and Turkington 2000). This idea is consistent with Grime's (1977) strategy of stress tolerators where it is assumed that species evolved to persist in harsh ecosytems with low productivity also have slow potential growth rates. The negative response of Anemone to fertilization is not surprising and has been observed in this system previously (John and Turkington 1997; Turkington et al. 1998; Arii and Turkington 2002). This is probably a direct toxic reaction to the fertilizer (Arii and Turkington 2002) and the explanation seems reasonable given the rapid response observed in this experiment and in Arii and Turkington (2002). In other experiments the decline in Anenome took a few years and it was thought that the reduction may occur due to increased competition imposed by other species utilizing the added fertilizer (John and Turkington 1997). Most of the previous studies exhibiting significant responses to fertilization done in this locale have tended to use existing individuals of unknown age as in the cases where competitive interactions were observed. However, Dlott and Turkington (2000) used seed-derived transplants. Contrary to our weak positive response to fertilization, they observed a general decrease in survival and growth of their transplants. Unfortunately, their results are not separated by species, and of their 8 species only 2, Lupinus arcticus and Festuca altaica, are in common with this study. Regardless, their  results, and ours, highlight that in experiments using seedlings, the responses to  30  fertilization may be quite different than the generally positive response observed from already established plants. AND JUST HOW IMPORTANT IS WATERING? Water limitation has not been studied to the same degree that nutrient limitation has for the boreal ecosystem. With one exception (Carrier and Krebs 2002) I know of no other experiments that have attempted to manipulate the water availability to boreal forest understory plants. In my study, three out of 10 species showed a significant increase in biomass due to watering and in general the overall biomass and survival increased for the transect. In the low water treatment I supplied approximately 75% more precipitation than the 30-yr annual average in an attempt to prevent death of our seedlings. Initially, I intended that the low watering treatment would only be slightly more water than the normal precipitation. However, it quickly became apparent that I would have had considerable death of transplanted seedlings. By increasing the amount of water applied I weakened the strength of our test because I reduced the ability to detect whether the system is water limited. It is possible that the low watering treatment provided sufficient water such that the seedlings were no longer water limited even though the naturally occurring vegetation would normally be water limited. Therefore, there may not be a biologically significant difference between the low and high watering even though there was a statistical one. Still, the biomass increased for three of the transplanted species with increased watering and the overall survival increased and was weakly significant (P = 0.096) indicating some water limitation. Although I have not observed direct evidence for competition, given that three species at the seedling stage responded positively to increased water, it seems likely that competition for water maybe of some importance in shaping the structure of this community. It is important to note that although I may have observed water limitations it does not necessarily mean that there will also be competition for water (Casper and Jackson 1997). Many of the examples of competition for water come from arid systems (Fowler 1986; Goldberg et al. 2001) though there are examples from other stressed ecosystems such as alpine meadows (Theodose and Bowman 1997) and oldfields (Stevens et al. 2006). It would be interesting to conduct a more detailed experiment examining the negative and positive interactions of plant communities in this dry boreal ecosystem using a water gradient.  31  CONCLUSION Previous research on the factors structuring this boreal forest understory community has focused on the role of nutrient limitation and competitive interactions with little emphasis on the role of water limitation and facilitative interactions. However, in this study, the presence of neighbours related to increased biomass and survival of seedlings while additional water had minor beneficial effects and nutrient addition had negligible effects. This study highlights the important role facilitation has in structuring this boreal understory community. The occurrence of these positive interactions may in part be due to the life history stage examined. These results provide further evidence that facilitative interactions are key in structuring plant communities in stressed environments.  32  REFERENCES Aksenova, A.A., Onipchenko, V.G. and Blinnikov, M.S., 1998. Plant interactions in alpine tundra: 13 years of experimental removal of dominant species. Ecoscience, 5: 258-270. Arii, K. and Turkington, R., 2002. Do nutrient availability and competition limit plant growth of herbaceous species in the boreal forest understory? Arctic, Antarctic, and Alpine Research, 34: 251-261. Bertness, M.D., 1998. Searching for the role of positive interactions in plant communities. Trends in Ecology and Evolution, 13: 133-134. Bertness, M.D. and Callaway, R., 1994. Positive interactions in communities. Trends in Ecology and Evolution, 9(5): 191-193. Bonan, G.B. and Shugart, H.H., 1989. Environmental factors and ecological processes in boreal forests. Annual Review of Ecology and Systematics, 20: 1-28. Brooker, R.W. and Callaghan, T.V., 1998. The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oikos, 81(1): 196-207. Callaway, R.M., 1995. Positive interactions among plants. Botanical Review, 61(4): 306349. Callaway, R.M., 1997. Positive interactions in plant communities and the individualisticcontinuum concept. Oecologia, 112: 143-149. Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., Michalet, R., Paolini, L., Pugnaire, F.I., Newingham, B., Aschehoug, E.T., Armas, C., Kikodze, D. and Cook, B.J., 2002. Positive interactions among alpine plants increase with stress. Nature, 417: 844-848. Callaway, R.M. and Walker, L.R., 1997. Competition and facilitation: a synthetic approach to interactions in plant communities. Ecology, 78(7): 1958-1965. Carlsson, B.A. and Callaghan, T.V., 1991. Positive plant interactions in tundra vegetation and the importance of shelter. Journal of Ecology, 79: 973-983. Carrier, P. and Krebs, C.J., 2002. Trophic effects of rainfall on Clethrionomys rutilus voles: an experimental test in a xeric boreal forest in the Yukon Territory. Canadian Journal of Zoology, 80: 821-829. Casper, B.B. and Jackson, R.B., 1997. Plant competition underground. Annual Review of Ecology and Systematics, 28: 545-570. Cody, W.J., 1996. Flora of the Yukon. NRC Research Press, Ottawa, Ontario, Canada. Connell, J.H., 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. American Naturalist, 122(5): 661696. Dlott, F. and Turkington, R., 2000. Regulation of boreal forest understory vegetation: the roles of resources and herbivores. Plant Ecology, 151: 239-251. Fleming, R.L., Black, T.A., Adams, R.S. and Stathers, R.J., 1998. Silvicultural treatments, microclimatic conditions and seedling response in Southern Interior clearcuts. Canadian Journal of Soil Science, 78: 115-126. Fowler, N., 1986. The role of competition in plant communities in arid and semiarid regions. Annual Review of Ecology and Systematics, 17: 89-110. Gerdol, R., Brancaleoni, L., Menghini, M. and Marchesini, R., 2000. Response of dwarf shrubs to neighbour removal and nutrient addition and their influence on community structure in a subalpine heath. Journal of Ecology, 88: 256-266. Goldberg, D.E., Rajaniemi, T., Gurevitch, J. and Stewart-Oaten, A., 1999. Empirical approaches to quantifying interaction intensity: competition and facilitation along productivity gradients. Ecology, 80(4): 1118-1131.  33  Goldberg, D.E., Turkington, R., Olsvig-Whittaker, L. and Dyer, A.R., 2001. Densitydependence in an annual plant community: variation among life history stages. Ecological Monographs, 71: 423-446. Grace, J.B. and Tilman, D. (Editors), 1990. Perspectives on plant competition. Academic Press, New York. Graham, S.A. and Turkington, R., 2000. Population dynamics response of Lupinus arcticus to fertilization, clipping, and neighbour removal in the understory of the boreal forest. Canadian Journal of Botany, 78: 753-758. Grime, J.P., 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist, 111: 11691194. Holmgren, M., Scheffer, M. and Huston, M.A., 1997. The interplay of facilitation and competition in plant communities. Ecology, 78(7): 1966-1975. Howard, T.G. and Goldberg, D.E., 2001. Competitive response hierarchies for germination, growth, and survival and their influence on abundance. Ecology, 82: 979-990. John, E. and Turkington, R., 1995. Herbaceous vegetation in the understorey of the boreal forest: does nutrient supply or snowshoe hare herbivory regulate species composition and abundance? Journal of Ecology, 83: 581-590. John, E. and Turkington, R., 1997. A 5-year study of the effects of nutrient availability and herbivory on two boreal herbs. Journal of Ecology, 85: 419-430. Jonasson, S., 1992. Plant responses to fertilization and species removal in tundra related to community structure and clonality. Oikos, 63: 420-429. Keddy, P.A., 1989. Competition. Chapman and Hall, New York. Lamb, E.G. and Cahill, J.F., 2006. Consequences of differing competitive abilities between juvenile and adult plants. Oikos, 112: 502-512. Larsen, J.A., 1980. The boreal ecosystem. Academic Press, New York. Lortie, C.J. and Callaway, R.M., 2006. Re-analysis of meta-analysis: support for the stress-gradient hypothesis. Journal of Ecology, 94: 7-16. Maestre, F.T., Valladares, F. and Reynolds, J.T., 2005. Is the change of plant-plant interactions with abiotic stress predictable? A meta-analysis of field results in arid environments. Journal of Ecology, 93: 748-757. Newman, E.I., 1973. Competition and diversity in herbaceous vegetation. Nature, 244: 310. Schoener, T.W., 1983. Field experiments on interspecific competition. American Naturalist, 122(2): 240-285. Shevtsova, A., Haukioja, E. and Ojala, A., 1997. Growth response of subarctic dwarf shrubs, Empetrum nigrum and Vaccinium vitis-idaea, to manipulated environmental conditions and species removals. Oikos, 78: 440-458. Shevtsova, A., Ojala, A., Neuvonen, S., Vieno, M. and Haukioja, E., 1995. Growth and reproduction of dwarf shrubs in a subarctic plant community: annual variation and above-ground interactions with neighbours. Journal of Ecology, 83: 263-275. Stevens, M.H.H., Shirk, R. and Steiner, C.E., 2006. Water and fertilizer have opposite effects on plant species richness in a mesic early successional habitat. Plant Ecology, 183: 27-34. SYSTAT Software Inc, 2002. SYSTAT Version 10, Richmond, CA. Theodose, T.A. and Bowman, W.D., 1997. Nutrient availability, plant abundance, and species diversity in two alpine tundra communities. Ecology, 78(6): 1861-1872. Tilman, D., 1982. Resource competition and community structure. Princeton University Press, Princeton.  34  Tilman, D., 1988. Plant strategies and the dynamics and structure of plant communities. Princeton University Press, Princeton, NJ. Turkington, R., John, E., Krebs, C.J., Dale, M.R.T., Nams, V.O., Boonstra, R., Boutin, S., Sinclair, A.R.E. and Smith, J.N.M., 1998. The effects of NPK fertilization for nine years on boreal forest vegetation in northwestern Canada. Journal of Vegetation Science, 9(3): 333-346. Turkington, R., John, E., Watson, S. and Seccombe-Hett, P., 2002. The effects of fertilization and herbivory on the herbaceous vegetation of the boreal forest in north-western Canada: a 10-year study. Journal of Ecology, 90: 325-337. Underwood, A.J., 1997. Experiments in ecology: their logical design and interpretation using analysis of variance. Cambridge University Press, Cambridge, UK. Zar, J.H., 1984. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, New Jersey.  35  Chapter 3 THE COMMUNITY DENSITY SERIES (CDS) 1 This experimental method was developed by Goldberg et al. (1995) to examine the specific role of competition at the whole corn munity level because density-depend ent regulation in plants may frequently occur at the level of the entire community, rather than within particular species. The design is an extension of traditional agronomic studies that examine the yield response of a single species to a range of planting densities (Harper 1977). In the CDS the entire community is manipulated to levels below and above the naturally occurring community density, while keeping relative pro portions of each species constant. Similar to a yield-density experim ent of a single species, the yield-density relationship of the manipulat ed community will reach a total community yield or carrying capacity (Fig. 3.1). A null community (with no species interactions) is defined experimentally with these manipulations rather than with the observational/statistical methods as in other competition studies. The null community can be visualized as occurring below the competition threshold density, Dc (Fig. 3.1). Therefore, a major assumption of this method is that the potential for biotic interactions increases with density of the entire corn munity (Goldberg et al. 1995). Responses of community parameters such as diversity or dominance indices can also be analyzed along the gradient of initial community density. Community parameters below Dc characterize the null community and reductions (or increases) in the parameters after that point are due to competitive interactions (Zobel 1992). This approach has been successfully used in the Negev desert to examine the role of competition in structuring annual plant communities (Goldberg et al. 2001 ; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 2005), in experimental communities of bryophytes (Zamfir and Goldberg 2000), and in old-field communities (Rajaniem i and Goldberg 2000).  ' A version of this chapter will be submitted for publication. Treberg, M. A. and Turkington, R. (2008). The Community Density Series: Theoretical Considerations.  36  Constant final yield  Density  Figure 3.1. The Community Density Series and the relationship between density and final yield. Dc is the density at which competition begins to reduce the linear increase in final yield and Dm is the density where the mass of the community remains constant even if density still increases. This is also called constant final yield. Figure redrawn from Goldberg et al. (1995).  EQUATIONS DESCRIBING THE COMPETITION-DENSITY EFFECT AND THE YIELD DENSITY EFFECT IN MONOCULTURES Most of the early research describing the mathematical relationships of the competition-density and the yield-density effects was done by Japane se researchers. Their first paper (Kira et al. 1953) described the relationship between density and mean plant weight as:  w = Kd'^  (3.1)  where w is the mean plant weight, d is density and K and a are constants that are dependent on the stage of growth of the plants. This can be rewritten as:  log w = log K — a log d  (3.2)  Since the total yield is simply the density multiplied by the mean plant mass, yield Y can be determined by:  37  Y = Kc1 1- a^  (3.3)  The constant a was described by Kira et al. (1953) as a measure of space utilization by the plant and it changes with plant age, increasing from 0, when no competition is occurring, and approaching unity as the plants grow to maturity (Figs. 3.2b, 3.2d). When a = 0 at the start of an experiment, w is the weight of the seed and w  = K. As the plants grow, if a reaches 1, Y becomes independent of density. This principle is now referred to as the "law of constant final yield" (Firbank and Watkinson 1990). The slope of the geometric power relationship in equation (3.1), constant a, can also be used as an index of the intensity of corn petition and it was concluded that the intensity of competition increases (Fig. 3.2b) in monocultures as plants age (Kira et al. 1953). The relationships between density and either mean plant weight or yield are most clearly visualized with double logarithm ic axes (Figs. 3.2b, 3.2d) rather than on arithmetic axes (Figs. 3.2a, 3.2c).  38  ▪  ^  (a) 70 _  (b) 100 -  3;^•  "E '^ 60 - I^ l adaYs "C'' co^^ ^ ns -a_ a 50 I^ ,15^ 1,0^ if) 2 40 -c^HI cr) 0)^( 84 days .6 O 30 1 3^ 3  113 days  10  84 days  -  45 days  a" 20 -I ‘,„,, • ‘;. 45days^ -0 "1:3 c ca^ 6t3 10 -' 42, , 3ldays^ a) 2ldays i^41__, __  2  (c)  0  •  a)  0 days  0.1 10^100^1000  Densityof surviving plants (no. m -2 )  Density of survi\4ng plants (no. m -2 )  1000  (d)  -  800  cr,  3ldays  ii=s= 12 d ays^2ldays  0^100^200^300^400^500  •  c., 'E  1  VP days  •  IND  .— •E  600^84 days 400^.7*--  ----  _---•  45 days 31days 2ldays '2 days  >-  31days  ili______.--  100 -  84 days  rn  00^0____------- -^45 days  200^  1/3__ days  2ldays  0 days  _------ —012 days • 0 days  0^  200^400^600  ^  Density of surviving plants (no. m -2 )  ^  10^100^1000  Densityof surviving plants (no. m -2)  Figure 3.2. The relationship through time between density of soybean plants and mean dry weight per plant on (a) arithmetic scales and (b) double logarithmic scales and between density and yield on (c) arithmetic scales and (d) double logarithmic scales. Figures (b) and (d) are redrawn from Kira et al. (1953).  At very low plant densities, the geometric power relationship (equation 3.1) does  not adequately describe the relationship between density and mean plant mass. For example, in Figure 3.2b. at the 12 day and 21 day censuses, the yield does not chan ge at the lowest densities. A more suitable reciprocal equation was suggested (Shinozaki and Kira 1956) such that:  (Ad +B) 1^(3.4)  39  where A and B are constants. Equation (3.4) can be rewritten to describe yield as:  Y =d(Ad + B) I^(3.5)  These reciprocal equations are a much better fitting model when no effect of competition is observed in low densities. For example, in Figs 3.3a & 3.3b, the geometric model of Kira et al. (1953) provides a statistically significant fit for the relationships of density and dry plant weight and yield of clover with r2 values of 0.845 and 0.924 respectively; however, the reciprocal model provides fits r2 values of 0.999 and 0.999 respectively. The constants A and B in equations (3.4) and (3.5) do not affect the shape of the competition-density curve which is smooth through the whole range of densities (Fig. 3.4); the curve is only translocated (Shinozaki and Kira 1956). An increase in A moves the curve horizontally to the left while an increase in B pushes the curve down (Fig. 3.4). The competition intensity index, a, can be calculated using A and B by:  (B  1  a= — x- +1 A d  ) 1  =wAp=1—wB  (3.6)  but a differs when calculated from the reciprocal equations (versus the power equation) in that it is dependent on either d or w (Shinozaki and Kira 1956).  40  (a)  -a 'm oo  (b) 100 -  •  'E•  10  eL  o, 100  -0  cc  N 2^10 ^  ^0^1^10^1000 Density (plants /square link)  of'  0.1 0.1  10  1000  Dens ity(plants / square link)  Figure 3.3. A sample dataset showing the relationship between density and (a) mean dry weight and (b) yield for subterranean clover. The solid line in (a) and (b) are the equations suggested by Kira et al. (1953) for mean dry weight and yield, respectively. The dashed lines are the reciprocal equations of Shinozaki and Kira (1956). In this case, the reciprocal equations are a better fit for both weight and yield. A square link is an old fashioned measure of area equivalent to 0.939034 2 m . Data from Table 1 from Willey and Heath (1969).  41  (b)  (a)^  A = 0.001  .  A = 0.001  0) 0  A = 0.1 Log density  Log density  (d)  (c)  Log Density  Log Density  Figure 3.4. The effect of varying a single parameter in equations (4) and (5) when the other is held constant on the mean plant weight and total yield. In (a) and (b) the value of B is held constant and A varies. In (c) and (d) the value of A is held constant. Changes in either A or B do not affect the shape of the curve, they only shift the position of the curve.  While adequately fitting many data sets, the reciprocal models (Shinozaki and Kira 1956) had two short-comings; the models did not include the power term that was demonstrated to be important earlier (Kira et al. 1953; Firbank and Watkinson 1990) and they can only describe a n asymptotic yield-density relationship that do es not fit some data sets that do not follow the law of constant fi nal yield (for example, Fig. 3.5). The latter concern was addressed by earlier attempts that used par abolic curves or other reciprocal equations to describe density and yield (Willey and Heath 1969). However,  42  these equations lack the generality required to make them useful for describing all data. Therefore, a more general reciproca I equation was proposed (Watkinson 1980):  w = wm (1+ Cd) b^(3.7)  where wm , b and c are constants. Yield can be described as:  Y = wm c/(1+ cd) -b^(3.8)  All of the parameters in equations (3.7) and (3.8) are time-dependent (Watkinson 1984) and biological interpretations have been given for these constants ( Watkinson 1980; Firbank and Watkinson 1990). The term W m estimates the mean weight of isolated plants not subject to com petition and c estimates the area needed for a plant to grow to wm and has the dimensions of area, meaning 1/c is the density at which plants begin to have effects of com petition (Watkinson 1980). While b has been interpreted as a measure of resource utilization by individual plants (Watkinson 1980; Firbank and Watkinson 1990), it has also been interpreted as the rate at which the effects of competition change with density (Vandermeer 1984) . Unlike A and B in equations (3.4) and (3.5), b has dramatic effects on the shape of the competition-density and yield-density curves ( Figs. 3.6a, 3.6b). Except when b = 0 or b = 1, the competition-density curve is non -linear. If b = 0, then w = w, and y increases linearly with density. As b increases but while b < 1, y increases constantly and never reaches an asymptote. When b = 1, w is linearly related to the reciprocal of density and y is asymptotic and follows the law of constant fi nal yield. If b > 1, then y follows a parabolic relationship and decreases after reaching a maximum. The c term has a similar effect as b, although does not have the same effect if c > 1 because y does not follow a parabolic curve (Figs. 3.6c, 3.6d). Equation (3.1) can be considered a simplification of equation (3.7) when a = b and:  K= w,„(1+^  (3.9)  43  Similarly, equation (3.7) will provide the same best fitting line as the simpler reciprocal equation (3.4) if b = 1. Though equation (3.7) is the most general form of the competition - density relationship, its inappropriate use can lead to so me problems if we try to interpret the parameters. For example, we can fit equation (3.7) to data that does not demonstrate competition at low densities. Normally we would use equation (3.1) on this type of data. If we were to use equation (3.7) on this type of data, both w, and c will tend to have very high values, regardless of a or b, making the previously mentioned biological explanations for Wm and c seem unreasonable. As previously noted, in equation (3.1 ), the slope a can describe the intensity of competition (Kira et al. 1953). Given that equation (3.1) is a simplification of equation (3.7), we can also interpret parameter b as a measure of the intensity of competition. This model has some other useful characteristics under certain situations. If competition does not occur at lower de nsities, w m can be used to determine the mean plant mass in the absence of competition (since this is its definition) and the density at which competition begins to affect the mass can be determined as the point at which mass significantly decreases below W m . Also, when b equals or is close to unity, the yield will approach an asymptote. This is the point at which constant final yield is achieved. The parameters in equations (3.7) and (3.8) can be estimated using iteration and least-squares or maximum likelihood functions (Firbank and Watkinson 1990). However, there are no well defined r 2 for these methods and the standard errors of the estimates are only approximations (SAS 1995).  44  (b)  30  1  25 '2 • 20 cn  •  0  15 -1 10  0  5-  0 10^100^1000^0^50^100^150 ^ Dens ity (x 10 4 plants acre) Dens ity( x 10 4 plants acre"')  Figure 3.5. The relationship between density and (a) mean root weight and (b) root yield of the globe red beet. The general reciprocal equations (3.7) and (3.8) of Watkinson (1980) are fitted to the mean root weight and root yield data. This is an example of the parabolic relationship between yield and density. Data redrawn from Willey and Heath (1969).  45  b = 0^  b =0  (b)  b = 0.5  b = 0.5  =1  b=1 V  a)  a)  0  b=2 b=2  Log Density  Log Density c=0  (c)  c=0  (d)  c = 0.1 c = 0.1 c = 1.0 c = 1 .0  T.) '5,  c = 10  0  c = 10  Log Density  Log Density  Figure 3.6. The effect of varying a single parameter in equations (3.7) and (3.8) when the others are held constant on the mean plant weight and yield. In (a) and (b) the value of c and vv,,, are held constant and b varies. In (c) and (d) the value of c varies. Increasing W m merely shifts the yintercept higher for all figures (not shown).  ANALYZING THE CDS Experiments using the CDS with many species can be analyzed like the traditional agronomic studies involving a monoculture. The lowest density plots, where density is low enough to preclude interactions, characterize the "null" comm unity and this can be compared to higher density plots where biotic interactions are affecting the plant community as a whole (Zobel 1992; Goldberg et al. 1995; Goldberg et al. 2001). However, when analyzing the entire community, instead of a single species, som e information is lost. The response of the community is the additive effects of all species. Some species will increase, while others will decrease and the subtleties of species-  46  specific responses to density may be overwhelmed by the response of the dominant species. However, the CDS has some very important advantages over traditional competition studies. Firstly, although the species specific responses are combined when analyzing the CDS as a community, the independent responses of each species to changing density in the CDS can still be evaluated. Therefore both species- and community-level patterns can be examined. The second major advantage of the CDS is that both negative and positive density dependent processes can be detected when analyzing the relationship betwe en plant mass and density. While all the data presented above are examples of competitive interactions, positive interactions could be observed and plotted in a manner similar. A third advantage of the CDS is that the slope (either a or b) and the R2 from the regression of the yield-density relationship corre spond to the intensity and importance of competition respectively. Kira et al. (1953) were aware that the slope of the yield-density regression was an index of the inten sity of competition and this has been restated more recently (Welden and Slauson 1986; Aarssen and Epp 1990). To compare the intensity of competition between different communities or species, it is advisable to use the same model describing the C-D or Y-D relationships for each community or species in the comparison. Unfortunately, sometimes the best fitting model is not always the same for each species (or community). In traditional yield-density studies, the coefficient of determination, or ► 2 , from a simple linear regression (Welden and Slauson 1986; Aarssen and Epp 1990; Weigelt and Jolliffe 2003) or it's multivariate equivalent (M cLellan et al. 1997; Sammul et al. 2000), R2 , is the importance of competition because it is the proportion of variation in yield that is directly due to density (Welden and Slauson 1986; Weigelt and Jolliffe 2003) compared to other possible factors affecting yield. The ability to quantify both the intensity and importance of competition can help untangle some of the debates surrounding the role of competition in structuring plant communities (Welden and Slauson 1986; Brooker et a I. 2005). Unfortunately, for the general reciprocal model, equations (3.7) and (3.8) of Watkinson (1980), there is no commonly accepted method of calculating r 2 ; however, the approximations of the residua I sum of squares error and the total sum of squares (which is a character istic of the data and not the model) can be used to estimate an r2 . The CDS is an extension of trad itional single species yield-density studies applied to multi-species communities. It can also be considered an extension of other  47  common experimental methods for investigating interspecific corn petition, such as the simple pair-wise design, or as any design with equal proportions and increasing density (for example Austin et al. 1988). It could also be identical to a simple additive design where plants are grown with and without neighbours (Freckleton and Watkinson 2000). Usually a pair-wise experiment is done with mixtures of two species maintaining a 1:1 ratio (Gibson et al. 1999) although there is no ecological reason why this ratio must be used. For example, in an extremely simple plant community with only two species, there may be al :3 ratio between species A and B. A traditional pair-wise experiment to determine if the two species are competing with each other could use the naturally occurring 1:3 planting ratio. Similarly, we could determine if competition is important at the community level by using the CDS and manipulating the community density below and above natural conditions while maintaining the 1:3 ratio; these two experiments are identical. Obviously, if many densities are used while maintaining the species proportions, a pair-wise experiment is the simplest form of the CDS. However, there is one significant difference in methodology between these experim ents. The goal of the CDS is to manipulate the density of the whole comm unity below the level where competition begins to structure the community, thus characterizing the null community (Goldberg et al. 1995). I n a typical pair-wise experiment where plants are grown with and without neighbours, the absence of competitive interactions is characterized by the growth of a single plant species, usually individually. However, if the pair-wise experiment is completed with a suitable range in planting densities for the component where neighbours are present, the lowest densities are directly analogous to the null community determined in a CDS. Although not a natural community, the two-species pair-wise experiment of Austin et al. (1988) can be reanalyzed as the sim plest form of a CDS (apart from a monoculture community). The specific goals of the experiment were to test the suitability and effectiveness of the additive and substitutive designs for t esting the effects of interspecific competition. Austin et al. (1988) grew two species, Silybum marianum and Cirsium vulgare, at constant 1:1 proportions at varying densities (Fig. 3.7). Using the CDS approach, we can see that as a whole the combination (or community for this example) of the two species was strongly negatively influenced by density (Fig. 3.7, Table 3.1). In this case, the geometric power relationship between mean plant mass and density provided a good fit with the data (Table 3.1). The majority of the mean plant mass and yield responses are due to the changes in mass and yield of S. marianum.  48  This highlights the tremendous effect that a dominant species may have in analyzing a CDS. However, it also highlights the benefit of being able to examine the response of each species individually. For example, while S. marianum may have been the dominant species in the "community", it responded to increases in density in the sam e manner as C. vulgare, i.e. they had very similar slopes (Fig. 3.7, Table 3.1). It is important to note that the increase in density is an increase in both inter- and intraspecific competition, which we may also call diffuse competition, especially if we are dealing with many species. The importance of competition for C. vulgare was lower than for S. marianum. This rather simple example used only two species, tho ugh the experiment was not initially designed as a CDS. T here are now exam pies of the CDS using multi-species communities (Rajaniem i and Goldberg 2000; Zamfir and Goldberg 2000; Goldberg et al. 2001; Lortie and Turkington 2002).  49  (a)  (b) 3000  1400 1200 1 °  E 1000  ^ S. marianum  (e)  A C. vulgare  800  •  2500  • Combined 715.  •  2000 1500  600 as a)  2  400  I  8  500 -  200  a  0 0  A  .  .^ ....  0^  10^20^30^40^0^20^40 Density(plants pof 1 )  (c)  Density(plants por e )  (d)  10000 -  1000 U)  10000  1  1000  -  100  -  0  E  100  A^A  g.--S A  a a)  10 -  10  2  1 1  1 ^ ^ 100 1^10^100 ^ Density(plants pot 1 ) Density(plants pot 1 )  ^ ^ 10  Figure 3.7. Mean plant mass and yield in relation to density for 1:1 mixtures of Silybum marianum and Cirsium vulgare. The legend in (a) applies to all 4 panels and the solid circle is the combined mean or yield for the two species. Figures (a) and (c) show the same data, except (a) has arithmetic axes and (c) has double logarithmic axes. Likewise for figures (b) and (d). The lines shown are the best fitting geometric power relationships of Kira et al. (1953) which are presented in Table 3.1. Data redrawn from Figure 3 of Austin et al. (1988).  Table 3.1. Best fitting lines and r2 for the data presented in Figure 3.7. Best fit lines Silybum marianum^log w = log 1811 — 0.794 log d^0.924 log Y = log 57.39 + 0.426 log d^0.288 Cirsium vulgare^log w = log 989.1 — 0.779 log d^0.423 log Y = log 906.0 + 0.206 log d^0.449 Combined^ log w = log 114.8 — 0.574 log d^0.920 log Y = log 986.1 + 0.221 log d^0.478  50  REFERENCES Aarssen, L.W. and Epp, G.A., 1990. Neighbour manipulati ons in natural vegetation: a review. Journal of Vegetation Science, 1: 13-30. Austin, M.P., Fresco, L.F.M., Nicholls, A.O., Groves, R.H. and Kaye, P.E., 1988. Competition and relative yield: estimation and interpretation at different densities and under various nutrient concentrations using Silybum marianum and Cirsium vulgare. Journal of Ecology, 76: 157-171. Brooker, R., Kikvidze, Z., Pugnaire, F.I., Callaway, R.M., Choler, P., Lortie, C.J. and Michalet, R., 2005. The im portance of importance. Oikos, 109: 63-70. Firbank, L.G. and Watkinson, A.R., 1990. On the effects of competition: from monocultures to mixtures. In: J. Grace and D. Tilman (Editors), Perspectives on plant competition. Academic Press, San Diego, pp. 165-192. Freckleton,,B.P. and Watkinson, A.R., 2000. Designs for greenhouse studies of interactions between plants: an analyti cal perspective. Journal of Ecology, 88: 386-391. Gibson, D.J., Connolly, J., Hartnett, D.C. and Weidenhamer, J.D., 1999. Designs for greenhouse studies of interactions between plants. Journal of Ecology, 87(1): 116. Goldberg, D.E., Turkington, R. and Olsvig-Whittaker, L., 1995. Quantifying the community-level consequences of competition. Folia Geobotanica Phytotaxonomica, 30: 231-242. Goldberg, D.E., Turkington, R., Olsvig-Whittaker, L. and Dyer, A.R., 2001. Densitydependence in an annual plant community: variation among life history stages. Ecological Monographs, 71: 423-446. Harper, J.L., 1977. Population biology of plants. Chapman & Hall, London, UK. Kira, T., Ogawa, H. and Sakazaki, N., 1953. Intraspecific com petition among higher plants. I. Mean plant weight-den sity interrelationship in reg ularly dispersed populations. Journal of the Institute of Polytechnics, Osaka City University, Series D, 4: 1-16. Lortie, C.J. and Turkington, R., 2002. The effect of initial seed density on the structure of a desert annual plant community. Journal of Ecology, 90: 435-445. McLellan, A.J., Law, R. and Fitter, A.H., 1997. Response of calcare ous grassland plant species to diffuse competition: results from a removal experiment. Rajaniemi, T.K. and Goldberg, D.E., 2000. Quantifying individual- and co mmunity-level effects of competition using experimentally determined null species pools. Journal of Vegetation Science, 11: 433-442. Sammul, M., Kull, K., Oksanen, L. and Veromann, P., 2000. Competition intensity and its importance: results of field experiments with Anthoxanthum odoratum. Oecologia, 125: 1 8-25. SAS, 1995. JM P. SAS Institute Inc., Cary, North Carolina. Shilo-Volin, H., Novoplansky, A., Goldberg, D.E. and Turkington, R., 2005. Density regulation in annual plant communities under variable resource levels. Oikos, 108: 241-252. Shinozaki, K. and Kira, T., 1956. Intraspecific competition among higher plants. VII . Logistic theory of the C-D effect. Journal of the Institute of Polytechnics, Osaka City University, Series D, 7: 35-72. Turkington, R., Goldberg, D.E., Olsvig-Whittaker, L. and Dyer, A.E., 2005. Effects of density on timing of emergence and its consequences for survival and growth in two communities of annual plants. Journal of Arid Environments, 61: 377-396.  51  Vandermeer, J., 1984. Plant competition and the yield-density relationship. Journal of Theoretical Biology, 109: 393-399. Watkinson, A.R., 1980. Density-dependence in single-species populations of plants. Journal of Theoretical Biology, 83: 345-357. Watkinson, A.R., 1984. Yield-density relationships: the influence of resource availability on growth and self-thinning in populations of Vulpia fasciculata. Annals of Botany, 1984: 469-482. Weigelt, A. and Jolliffe, P., 2003. Indices of plant competition. Journal of Ecology, 91: 707-720. Welden, C.W. and Slauson, W.L., 1986. The intensity of competition versus its importance: an overlooked distinction and some implications. Quarterly Review of Biology, 61(1): 23-44. Willey, R.W. and Heath, S.B., 1969. The quantitative relationships between plant population and crop yield. Advances in Agronomy, 21: 281-321. Zamfir, M. and Goldberg, D.E., 2000. The effect of initial density on interactions betwe en bryophytes at individual and community levels. Journal of Ecology, 88: 243-255. Zobel, M., 1992. Plant species coexistence: the role of historical, evolutionary and ecological factors. Oikos, 65: 314-320.  52  Chapter 4 DENSITY DEPENDENCE IN AN EXPERIMENTAL BOREAL FOREST UNDERSTORY COMMUNITY' INTRODUCTION There is abundant evidence showing that competit ion occurs within plant communities (Connell 1983; Schoener 1983; Keddy 1989; Goldberg and Barton 1992; Goldberg et al. 1999) and there is a growing body of evidence docum enting the prevalence of facilitation in plant communities (Hunter and Aarssen 1988; Callaway 1995; Bertness 1998; Brooker and Callaghan 1998). Most of these studies only examine a small subset of species present in the community leaving the corn munitylevel consequences of biotic interactions unexplored and this is, at least in part, due to a lack of appropriate experimental methods (Goldberg et al. 1995). To evaluate the community consequences of biotic interactions, both the effects of competition and facilitation on the entire community, and not just a few select species, an d their effects in varying abiotic conditions need to be quantified (Goldberg and Barton 19 92; Goldberg and Scheiner 1 993; Goldberg et al. 1 995). Here, I use a comm unity density series (CDS) to determine the role of biotic interactions and abiotic influences on an experimental community. The CDS uses the performance of the comm unity at reduced levels of abundan ce to quantify the null community, the community without interactions, to determine the effect of competition or facilitation at higher, natural, levels of abundance (Goldberg et al. 1995). This approach is a community-level modification of traditional density-yield experiments. Density dependence is usually studied with individual plants or with individual species and a major problem with this approach is that plants almost always live in mixtures. Goldberg et al. (2001) noted three key problems with current approaches to studying density dependence. The first is th at most studies that actually manipulate density only focus on a single species. The second is that most studies are descriptive or mensurative rather than experimental. Obviously natural gradients are useful for  ' A version of this chapter will be submitted for publication. Treberg, M. A. and Turkington, R. (2008). Density Dependence in an Experimental Boreal Forest Understory Community. 53  examining natural patterns; however, it is often difficult to untangle the effects due to density and those due to underlying environmental factors. Unfortunately, the experiments that have been done may not tell us much about the patterns of density dependence since they are usually done to determine the effect of the presence of neighbours (such as Chapter 2) or focus on only one life history stage. The third problem is that most studies, even those that a re experimental, only study one or a few species in the field making it difficult to extrapolate to other plant species or to other environments. This does not mean that what happen s to a single species is of no interest for it maybe the most important issue in cases of conservation. However, it is unlikely that single-species responses are useful in predicting the community-level consequences of com petition or facilitation. Here I follow the terminology advocated by Crawl ey (1997) and used by others who have done com munity density experiments (Goldberg et al. 2001; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 2005). Negative density dependence, or competition, indicates negative effects of increasing density and positive density dependence, or facilitation, indicates positive effects of increasing density. In this study, I recreated an understory plant community to investigate the role of density dependence at the community- and species-levels and to determine whether changing the productivity of the system alters these relationships. Specifically, I ask the following questions: i) do density dependent biotic interactions affect the structure of this community? ii) do these interactions affect the different life history stages of emergence, survival and growth differently? iii)do these interactions affect the species in the community differently? iv) does varying water and fertilizer modify the productivity and how does this affect community structure?  METHODS COMMUNITY DESCRIPTION I attempted to recreate the understory plant community of the boreal forest near Kluane Lake in the southwestern Yukon. This ecosystem has been, and continues to be, extensively studied for both its animal and plant components and was used for th e  54  Kluane Boreal Ecosystem Project (Krebs et al. 2001). Previous research has show n this ecosystem is nutrient limited (Turkington et al. 1998; Turking ton et al. 2002), may be water limited (Carrier and Krebs 2002; Carrier 2003) and biotic interactions affect some plant species (Arii and Turkington 20 02). The overstory is com posed of white spruce (Picea glauca (Moench) Voss Si.) with a shrub canopy of willows (pred ominantly Salix glauca L. s. I.) and some dwarf birch (Betula glandulosa Michx.). The 10 most common understory species were chosen as representative species for the community density series experiment. Eight of the species are herbaceous perennials: Achillea millefolium L. ssp. borealis (Bong.) Breitung (yarrow), Anenome parviflora Michx. (northern anemone), Epilobium angustifolium L. s.l. (fireweed), Festuca altaica Trin. (northern rough fescue), Lupinus arcticus Wats. (arctic lupine), Mertensia paniculata (Ait.) G. Don var. paniculata (bluebells), Senecio lugens Richards. (black-tipped groundsel), Solidago multiradiata Ait. (goldenrod). The remaining two species are woody perennials: Arctostaphylos uva-ursi (L.) Spreng. s.l. (bearberry) and Linnaea borealis L. ssp. americana (twinflower). All nomenclature follows Cody (1996). These 10 species account for 73% of the individual understory plants and make up 95% of the understory cover.  EXPERIMENTAL DESIGN I tested for density dependence in an experimental community density series (CDS) of the most common understory plant species. The CDS is a multi-species refinement of the tradition al single-species density-yield experiments (Harper 1977) and was first described by Goldberg et al. (1995). Traditionally, a single species would be planted, or manipulated, such that its density varied from very low to very high. At lower densities the species' yield would increase linearly with density since the space between individual plants precludes competitive interactions. At some point as density continues to increase, competition would begin to affect yield, and while yield w ill still increase, it will do so at a lesser rate. Eventually, the carrying capacity would be reached and increasing planting density would no longer affect final yield. Similarly, an entire community can be manipulated to densities above and below the "normal" or initial community density (ICD) while keeping the relative proportions of each species constant (Goldberg et al. 1995). It can be assumed that if we lower the density of the com munity such that the density precludes interactions then we have characterized the null community (Zobel 1992; Goldberg et al. 1995). The CDS was successfully applied using  55  plants from an annual community in the Negev desert (Goldberg et al. 2001; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 200 5), and has been applied once in a perennial community (Rajaniemi and Goldberg 2000); however, the latter study only lasted one growing season and did no t use densities greater than the natural plant density of the community. This experiment was conducted during the summer of 2002 at the Arctic Institute of North America Kluane Lake Research Station (61 02' N, 138 25' W, altitude of 785 m). Six 120 cm wide x 240 cm long x 15 cm high boxes were built out of wood and each box was subdivided into 8 equal "plots" of approximately 60 cm x 60 cm with partitions of 2.5 cm thick wood. Some of the 60 cm x 60 cm plots were further subdivided into smaller 30 cm x 30 cm plots using 6 mm thick plywood partitions. The boxes were filled with beach sand and the surface was sterilized with heat for 30 seconds using a pr opane torch (Figure 4.1). The sterilization was sufficient because no plant species growing on the beach germinated in the sandboxes during the experiment. Greenhouse frames were built and covered with a clear 6 mil (0.15 mm) plastic vapour barrier. This plastic transmitted 73% of the incoming photosynthetically active radiation to the sand surface (measured using a qua ntum sensor, Apogee Instruments Inc., Logan, UT). The greenhouse covers allowed us: (i) to vary the volume of water added to each plot individually, (ii) to increase the temperature and humidity to enhance germination and survival of seedlings, (iii) to reduce damage from herbivores and (iv) to reduce the immigration of weed seeds into the plots. To ensure that the surface temperature of the plots did not increase to the point where it may injure seedlings, the greenhouse frames were propped open to allow some air flow. On particularly hot days, the frames were removed. To reduce the impact of herbivores, especially the granivorou s mice and voles, snap-traps were set out around the perimeter of the sandboxes.  56  Figure 4.1. Sandboxes and greenhouses used in the CDS experiment at Kluane Lake. The author is shown using a propane torch to sterilize the surface of the sand. The sandboxes were not treated as blocks. In an experiment the previous summer the sandboxes were treated as blocks and there were no statistical differences between blocks for soil moisture, soil temperature, light at the soil surface, seedling emergence, seedling survival or seedling grow th. Therefore, this experiment is treated as a completely randomized design.  TREATMENTS A geometric series of six Initial Community Densities (ICDs) was constructed: 1/8, 1/4, 1/2, 1, 2, and x4 the natural field density. The x1 density closely approximated the density of the vegetation estimated at a natural field site where another CDS experiment was established in 1999 (Chapter 5); this corresponded with a density of 132 plants m -2 (95% CI of 113 to 151 plants m- 2 ). The CDS was generated from mature seeds of the 10 most common plant species that were collected the previous sum mer and stored over winter. The average germination rate for each species was used to  57  calculate how much seed was necessary to give the density of indiv iduals desired to approximate the numbers observed in the field. Germination rates were estimated in the spring prior to this experiment using the standard method of Petri plates and moist sand (Baskin and Baskin 1998). For each plot, the exact numbers of seeds for each species were counted and all seeds were sown onto their respective plots and li ghtly raked into the surface of the sand. For example, for a plot with the treatment at the x1 density, or natural field density, I planted 1374 seeds m -2 for the expected 132 plants m -2 . In total, 12876 seeds were sown. In order to conserve seeds, the highest density plots (x1, x2, and x4 densities) were sown in 30 cm x 30 cm plots rather than 60 cm x 60 cm plots used for the lower densities. In the previous experiment done in these sandboxes, no plot size effects were observed for seedling emergen ce, survival or growth. I was unable to germi nate seeds of the two woody shrubs, Arctostaphylos and Linnaea. Instead I used cuttings rather than seed in the appropriate numbers (in total, 252 of Arctostaphylos and 504 of Linnaea) to approximate the natural density. The stem cuttings were collected from close to the study area and were 5 cm long with half of the length left as leaves and half stripped of leaves and treated with a growth hormone (Wilson Roots® Liquid Root Stimulator) to promote root growth. Anenome seeds collected from the previous season exhibited particularly low germination rates. Therefore, I did not have sufficient seed numbers and did not include it in the CDS. All seeds were sown on June 13, 2002. This coincided with the emergence of seedlings at the nearby field site. There were two watering levels. Water was added as necessary to keep the low water level moist enough to prevent obvious water stress and seedling death. The high water level was double the low water treatment and it was applied at the same time. Usually water was applied early morning, and again in the afternoon or evening if needed during hot or windy days. Volum etric water content (VWC) was monitored using a CS620 HydroSense® Water Content Sensor (Cam pbell Scientific Inc., Logan, UT). The low water treatment had an average VWC of 9.9 % (95 % CI of 9.7 to 10.1) and the high water treatment had an average VWC of 11.1 % (95 % CI of 10.9 to 11.3). These were statistically different (t - Test, df=70, P<0.0001). Water soluble fertilizer (Plant Prod® 20-20-20) was added as a solut ion every two weeks at the manufacturer's recommended rate; this was the low level treatment (4 mL fertilizer per 1 L H 2 0). This amount was doubled for the high level treatment.  58  Throughout the sum mer there were 5 fertilizer applications with a total addition of 11.4 g N rn -2 and 22.7 g N m -2 added to the low and high treatment levels respectively. These values correspond to other fertilizer applications in this region that have produced significant plant responses (Turkington et al. 1998). There was no "control" treatment without fertilizer added because this experiment was done using sand which is inherently nutrient poor and also has poor nutrient retention. All treatments were replicated three times in a factorial 6 x 2 x 2 design (density x water x fertilizer).  MONITORING AND HARVESTING All newly germinated seedlings were marked with coloured toothpicks every week to monitor emergence. Unfortunately, the toothpicks were removed mid-season by magpies (Pica pica) that crawled into the greenhouses at night while the houses were propped off of the ground. No plants were damaged, but the r emoval of the toothpicks made it such that I was unable to identify which weekly cohort each plant belonged to. Therefore, I am unable to comment on the species-specific changes in abund ance on a weekly basis and can only discuss the total emerged seedlings per week. At the end of the season, on Aug 29-31, 2002, all individuals in the plots were identified to species and counted, and all aboveground biomass removed by clipping. The plant material was sorted to species and dried to prevent spoilage before being sent to the University of British Columbia, Vancouver, BC, for final drying and weighing.  ANALYSIS The effect of density, water and fertilizer on the experimental community was examined using analysis of covariance (ANCOVA) with density as the covariate (because it is a continuous variable) and water an d fertilizer as categorical variables, each with 2 levels. All analyses were done using JMP 4 (SAS 1995). The effect of density on the community was analyzed using an individual performance approach (Goldberg et al. 2001). I n this method, the variable of interest is divided by the appropriate measure of density and the variable remains constant with density in the absence of interactions. An increase in the variable with increasing density would be indicative of facilitative interactions and a decrease would indicate competitive interactions. In order to use ANCOVA, the density and response variable relationship was transformed to best linearize the relationship. Four transformati ons were attempted: linear, power, sem ilog and reciprocal. The best fitting model, with the  59  highest R2 , is reported. These transformations also assisted in making the AN OVAs and ANCOVAs better meet the usual statistical assumptions. The effect of density was examined for three life-stages of the experimental community: emergence, survival and final plant size. The appropriate covariate, or measure of density to use, varied in each of the analyses. For the emergence stage I used the initial planting density. I calculated an index of emergence from the total number of seedlings emerged divided by the planting density. The measure of density used for the survival stage was the cum ulative density of all emerged seedlings. Survival was the proportion of those seedlings that survived until the end of the season. For the final plant mass, the appropriate covariate was the final plant de nsity. Plant mass was the total aboveground biomass of the plot divided by the final plant density. All of these analyses were also completed using the initial planting density as a covariate to determine if the choice of density measure affected the outcome of the statistical tests. There were no notable change s in test results. The effect of density on species diversity in plots was also examined. For both species richness and evenness, the initial planting density was used as a covariate in the ANCOVA. The evenness index used was Eva, (Smith and Wilson 1996). Because species richness is expected to increase with increased sampling, the expected species richness in the absence of interactions was also calculated from the null community following the procedure described by Goldberg and Estabrook (1998). Species-specific effects of density were exam fined on the proportional survival and the final plant m ass. The proportional survival was the final plant de nsity divided by the total seed planting density. Because survival was a proportion and many numbers were close to, or at zero, all survival data were arcsine transformed such that X' = arcsine (p) 112 (where X' = transformed value and p = the proportion). Final plant mass was the mass of all individuals of that species in the plot divided by the relative planting density. Those species with significant relationships were a nalyzed using AN COVA with initial planting density as the covariate and fertilizer and water treatments as categorical variables. If the species' survival or fi nal plant mass demonstrated no relationship with density, an analysis of variance (ANOV A) of the water and fertilizer treatments was done.  60  RESULTS  DENSITY DEPENDENCE IN THE COMMUNITY Density dependence was evi dent at all three life stages in the experimental communities, although it was not con sistently competitive or facilitative (Tables 4.1 & 4.2; Fig. 4.2). For all three stages, emergence, survival and shoot mass, the sem ilog model provided the best fit and all relationships were statistically significant (Table 4.1). Emergence was negatively related to density with much higher germination in the lowest density plots and lower germination at the highest density (Fig. 4.2a); this means that seeds in high density plots were somehow "sensing" their high density. Survival to the end of the season was positively associated with density indicating fa cilitative interactions at higher densities (Fig. 4.2b). The final stage, final per plant shoot mass, was negatively related to density, with extrem ely high shoot biomass only possible in the lowest density plots (Fig. 4.2c). The final observed plant density was best related to the initial planting density with the power model, w here both density terms are logged (Table 4.1). The final density was strongly and positively re lated to the initial planting density (Table 4.2; Fig. 4.3). Species richness was also highly positively and linearly density dependent (Tables 4.1 & 4.2; Fig. 4.4). Because species richness increases as the number of individuals in a sample increases, I also calculated the expected species richness ba sed on data from the null community. The expected species richness indicates how m any species to expect due to the increased sam piing. At all densities, the observed species richness is lower than the expected species r ichness (Fig. 4.4). There was no significant relationship between species evenness,  Evan  and density (Table 4.1; Fig. 4.4).  Table 4.1. Regression coefficients for the community response variables and density  relationships shown in Figures 4.2, 4.3 and 4.4. The model type refers to the data transformation that best linearized the data. The degrees of freedom (df) are for the model and error combined. A negative slope indicates negative density dependence (or competition) and a positive slope indicates positive density dependence (or facilitation). Significant values (P < 0.05) are in bold. Variable Emergence index Survival Shoot mass Final plant density Species richness Evenness (Evan)  Model semilog semilog semilog power linear linear  Df 71 71 71 70 70 70  Intercept 269.188 0.177 0.779 -2.276 4.760 0.343  Slope -37.046 0.083 -0.175 1.045 0.493 -0.020  142 0.097 0.252 0.224 0.805 0.185 0.017  P 0.008 <.001 <.001 <.001 <.001 0.277  61  •  ( a ) 800 700 600  •  •g 500 E 400 •  ,  9  •  c, 5 300 E 200  100 0 0  ^  1000^2000  ^  3000  ^  4000  ^  5000  Initial planting density (no.seeds / m 2 )  •  ••  •  • •  0  0^250^500^750^1000^1250^1500 Cumulative density (no. emergents / m 2 ) (C)  rn  2  a) 1.5 it) o_ (0  1  -5 0.5 0 _c  •  •  S1•  • •  I  I  0^250^500^750^1000^1250^1500 Final plant density (no. plants / m 2 )  Figure 4.2. The effect of density on the performance of seedlings in the experimental communities. In (a) the emergence index is the total seedlings that germinated per square meter divided by the relative seed density, (b) the proportion surviving is the number of emerged seedlings that survived to the end of the season, and (c) the mean shoot mass per plant is the total plot mass divided by the number of plants in the plot. The coefficients for the best fit curve for each graph are shown in Table 4.1.  62  •  "i 1500 Fs 1250  • •  g 1000 750 C  a) -o 500  •  co 250 6 0 1000^2000  •• ^  •  3000^4000  5000  Initial planting density (no. seeds / m 2 )  Figure 4.3. The effect of initial planting density on the final plant density. The coefficients for the best fit line are shown in Table 4.1. (a) 8 --^  .................................  7 :a)2 6 5 U  4  ci 3  .(  a)  0^1000^2000^3000  4000^5000  Initial planting density (no.seeds / m 2 )  (6) 0.6 0.5 0.4 0.3 0.2 0.1 0 1000^2000^3000^4000^5000 Initial planting density (no.seeds / m 2 )  Figure 4.4. The effect of initial seed density on (a) species richness and (b) evenness using Smith and Wilson's (1996) E„,., in the experimental community; lower values of Eva,. indicate that the species are very unequally represented in the community. The best fit curve for species richness is shown as a solid line and the coefficients are given in Table 4.1. The dashed line is the expected value of the species richness as calculated from the null community. Values in both figures are means (+/-95 % confidence interval).  63  WATER AND FERTILIZER EFFECTS ON THE COMMUNITY The effects of water and fertilizer addition were not consistent among life stages. Watering significantly increased the emergence of seedlings and species richness (Table 4.2). Fertilizer had an opposite effect and reduced seedling survival and species richness (Table 4.2). Some interaction terms were significant. The density and fertilizer interaction, and the water and fertilizer interactions, were significant for the emergence index (Table 4.2). The low fertilizer treatment had higher e mergence at lower densities than the high fertilizer treatment. At high densities there was no difference in emergence between the fertilizer treatments. The low water and high fertilizer treatment had significantly lower emergence than the low water and low fertilizer, high water and low fertilizer, or the high water high fertilizer treatments. For the final biomass, the density and water interaction was significant (Table 4.2) with the higher shoot mass in the low water treatment at low density than the high water treatment. At high densities there was no difference between the water treatments. For the final plant density, the density and water interaction was also significant (Table 4.2) and exhibited the same pattern as for the emergence index, where only the low water and high fertilizer treatment was significantly lower than the other treatments.  Table 4.2. Summary of ANCOVAs and ANOVA for the response variables to density manipulations in the experimental communities. Significant values (P < 0.05) are in bold. Response variable Emergence index  Survival  Shoot mass  Effect Initial planting density Water Fertilizer Density x Water Density x Fertilizer Water X Fertilizer Density x Water x Fertilizer Error Cumulative density Water Fertilizer Density x Water Density x Fertilizer Water x Fertilizer Density x Water x Fertilizer Error Final plant density Water Fertilizer Density x Water Density x Fertilizer  df 1 1 1 1 1 1 1 64 1 1 1 1 1 1 1 64 1 1 1 1 1  SS 138471 333268 26565 2738 64733 81574 623 784284 0.451 0.066 0.276 0.032 0.000 0.008 0.010 1.695 0.956 0.234 0.005 0.362 0.176  F Ratio 11.300 27.196 2.168 0.223 5.282 6.657 0.051  0.001 <.001 0.146 0.638 0.025 0.012 0.822  17.032 2.492 10.437 1.226 0.013 0.320 0.391  <.001 0.119 0.002 0.272 0.909 0.574 0.534  13.276 3.249 0.067 5.032 2.450  <.001 0.076 0.796 0.028 0.123  64  Response variable  Final plant density  Species richness  Evenness (E ar)  Effect Water x Fertilizer Density x Water x Fertilizer Error Initial planting density Water Fertilizer Density x Water Density x Fertilizer Water x Fertilizer Density x Water x Fertilizer Error Initial planting density Water Fertilizer Density x Water Density x Fertilizer Water x Fertilizer Density x Water x Fertilizer Error Water Fertilizer Water x Fertilizer Error  df 1 1 64 1 1 1 1 1 1 1 64 1 1 1 1 1 1 1 63 1 1 1 67  SS 0.041 0.089 4.537 96.499 11.576 5.235 0.004 0.256 2.206 0.185 32.406 31.649 12.401 0.164 0.007 4.491 0.342 93.186 0.000 0.065 0.031 2.938  F Ratio 0.564 1.232  190.127 22.914 10.362 0.007 0.508 4.367 0.367  P  0.455 0.271 <.001 <.001 0.002  0.935 0.479 0.041  0.547  21.909 21.397 8.384 0.111 0.005 3.036 0.231  0.740 0.945 0.086 0.632  0.001 1.479 0.709  0.975 0.228 0.403  <.001 <.001 0.005  SPECIES-SPECIFIC RESPONSES Four species' survival was significantly related to increased d ensity with two, Achillea and Arctostaphylos, responding negatively and two, Festuca and Solidago,  responding positively (Table 4.3; Fig. 4.5). Only three species demonstrated significant relationships between initial pl anting density and final plant biomass (Table 4.3). Achillea and Lupinus decreased in individual mass with increasing planting density, while Festuca was positively density dependent (Table 4.3; Fig. 4.6). Mertensia was the only species to show any positive response to fertilizer addition with increased survival (Table 4.4). All others, except Lupinus and Solidago, showed significant decreases in survival with increased fertilizer (Table 4.4). Only two species' biomass responded negatively to increased fertilizer, Arctostaphylos and Senecio (Table 4.4). No species  responded with increased growth to fertilizer addition. Increased water aided survival for Festuca, Linnaea, Lupinus and Mertensia, while Festuca, Lupinus and Mertensia all  increased in mass (Table 4.4). No species responded negatively to increased water. No interaction terms in the ANOVAs or ANCOVAs were significant. Although Epilobium seeds did germinate in the sandbox plots, none survived to the end of the season.  65  Table 4.3. Regression coefficients for the relationship between each species' survival or mean plant mass biomass and the initial planting density in the experimental communities. The proportional survival is the number of individuals of that species per plot divided by the initial planting density and mean plant mass is the mass of that species per plot divided by the relative initial density. The model type refers to the transformation required to best linearize the data. The degrees of freedom (df) are for the model and error combined. Significant values (P < 0.05) are in bold. Species Survival  Achillea Arctostaphylos Festuca  Linnaea  Lupinus Mertensia Senecio Solidago  Model  df  Intercept  Slope  R2  P  linear semilog semilog linear semilog semilog semilog semilog  71 71 71 71 71 71 71 71  0.984 1.887 0.218 0.064 0.625 -0.014 0.153 -0.172  -7.6 x 10- 5 -0.110 0.054 5.4 x 10 -6 -0.048 0.018 0.016 0.106  0.061 0.063 0.064 0.009 0.024 0.027 0.002 0.086  0.037 0.033 0.033 0.437 0.197 0.166 0.693 0.013  semilog linear linear linear linear linear linear linear  71 71 71 71 71 71 71 71  178.767 1.872 1.245 0.197 1.204 1.734 0.416 0.867  -21.253 -1.5 x 10 -5 4.3 x 10 4 -2.7 x 10 -5 -2.5 x 10 4 -1.6 x 10 4 -5.7 x 10 -5 1.2 x 10 -5  0.151 0.001 0.107 0.013 0.060 0.005 0.011 0.020  <.001 0.826 0.005 0.342 0.038 0.540 0.385 0.236  Mean Plant Mass  Achillea Arctostaphylos Festuca  Linnaea  Lupinus Mertensia Senecio Solidago  66  Achillea borealis ^1  -  •  •  •^•^•  0, 0.8 -I^•^•  • • • •  Z^•  H 0.6 -^• c^•  ^o  0.4 1^•^• • • •  ci. 2 0.2 ,^•^ a.  • •  •  0  0^1000^2000^3000^4000^5000 Initial planting density (no. seeds / m 2 )  Arctostaphylos uva-ursi 1 • • • E2 0.8  •  g 0.6  ••  C  0 0.  0.4 -  •  •  •  •  • • •  •  •  2 0.2  • •  •^  0  1000^2000^3000^4000^5000  0  Initial planting density (no. seeds / m 2 ) Festuca altaica  0.8  •  •  •  g 0.6  C  •  II  •• •  •  ••  0 0.4  2 0.2 0  I • 1 000  $ 2000  3000  4000^5000  Initial planting density (no. seeds / m 2 )  Solidago multiradiata 1 0, 0.8  •  •  •  •• •  •  • •  -  g 0.6 -  :20 0.4 , 2 0.2 a_  •  •  • • • •  0^•^•  0^1000^2000^3000^4000^5000 Initial planting density (no. seeds / m 2 )  Figure 4.5. The effect of initial planting density on the proportional survival for the species in the experimental community that showed significant responses to varying density. The best fitting curves are described in Table 4.3. Species whose survival was not significantly related to initial planting density are not shown.  67  ^  (a) 200 180 -1 160 - • 140 2 120 - •  m  Achillea millefolium ssp. borealis  H  E 100 -^•  e' 80^• i5 60 • • 40 - I • 20 - • 0 0  •  1000^2000^3000^4000  5000  Initial planting density (no. seeds / m 2 )  (b) 10  Festuca altaica  -  9- • 8'En 7^ • en 6 u)  E 5- • .C. 4 2 • co^ a.- 3 •• *^•  •  0 lit 1---1-1-0^1000^2000^3000^4000^5000 Initial planting density (no. seeds / m 2 ) (C) 8 7  Lupinus arcticus  •  • • 6 cn 5^• 4 • 'E n; 3^• • °- 2 • 1 0^ • 0  •  •  •  • 1000  2000  ^  3000  ^  4000  ^  5000  Initial planting density (no. seeds / m 2 )  Figure 4.6. The effect of initial planting density on the mean plant mass for the three species in the experimental community that showed significant responses to varying density. The plant mass is the total g/m 2 for that species divided by its relative initial density. The best fitting curves are described in Table 4.3. Species whose mass was not significantly related to initial planting density are not shown.  68  Table 4.4. P values from ANOVAs and ANCOVAs for each species using the proportional survival or the total biomass of each species per plot divided by the relative initial density. Significant values (P < 0.05) are in bold.  Species Survival  Achillea Arctostaphylos Festuca Linnaea Lupinus Mertensia Senecio Solidago  Fertilizer  Water  Density  0.009 0.008 <.001 0.007 0.603 <.001 0.002 0.195  0.561 0.159 <.001 0.010 <.001 0.021 0.164 0.812  0.031 0.027 0.003  0.148 0.003 0.209 0.115 0.095 0.154 0.003 0.083  0.455 0.235 0.002 0.072 0.005 0.009 0.693 0.862  0.014  Fertilizer x Water  Fertilizer x Water  Fertilizer x Density  Water x Density  0.316 0.865 0.114 0.332 0.478 0.317 0.844 0.209  0.141 0.475 0.134  0.567 0.510 0.464  0.798 0.236 0.317  0.445  0.817  0.821  0.170  0.283  0.346  0.306  0.237  0.162  0.084  0.273  0.153  Density  Mean Plant Mass  Achillea Arctostaphylos Festuca Linnaea Lupinus Mertensia Senecio Solidago  <.001 0.003 0.025  0.190 0.793 0.173 0.910 0.509 0.814 0.638 0.650  DISCUSSION Plant communities are structured by density dependent processes (Silvertown and Lovett Doust 1993; Crawley 1997). However, Goldberg et al. (2001) identified many problems with our past and current methods of investigating density dependence, the principal one being the investigation of only a single or few species while apparently ignoring that regulation may occur at the comm unity level. This traditional approach tells us little about how the focal species interact s with other species and what the overall consequences of community density are for either the population of the species studied or the community as a whole. The experimental approach utilized in the CDS avoids this problem by manipulating the density of the entire community and examining the relationship of density with individual species and the community (Goldberg et al. 1995; Goldberg et al. 2001). In this study, all of the life history stages examined demonstrated density dependence at the comm unity level; however, whether the density dependence was positive or negative depended on the life history stage examined. While seed germination and the final mean plant mass were negatively related to de nsity, indicating  69  competitive interactions, survival was positively related, indicating facilitative interactions. These patterns of density dependence par allel those of Goldberg et al. (2001) and are similar to those of Lortie and Turkington (2002) although Lortie and Turkington (2002) did not demonstrate density dependence for survival. The major distinction between these studies is that Goldberg et al. (2001) and Lortie and Turkington (2002) used desert annual plant communities, whereas I have applied the CDS to boreal understory perennial species grown from seed (albeit for only one season). Only one other study (Rajaniem i and Goldberg 2000) has used perennial species, with old field species grown from seed at natural densities and in thinned plots (approximately 1/20 the natural density). All of these studies, regardless of habitat or community studied, demonstrate that lower than natur ally occurring densities of plants have higher mean mass than the normal (x1) density. In this study, the germination rate was significantly reduced at the higher densities indicating negative density depend ent emergence. Other studies have also observed negative density dependent germination (McMurray et al. 1997; Murray 1998; Goldberg et al. 2001; Lortie and Turkington 2002) while others have observed that increased seed density can be facilitative (M cMurray et al. 1997; Dyer et al. 2000). Although Murray (1998) reported that 4 species out of 12 demonstrated negative density dependence, he detected no relationship between density and germination for any of the other species in his study. A possible mechanism for the negative density dependent response could be the release of lea chates from the seed coat that directly inhibits germination (M urray 1998). This may lead to earlier germination and emergence, ( Dyer et al. 2000), which may be an adaptive response to get a head start in competitive environments, or to decreased germination (M urray 1998; Goldberg et a I. 2001; Lortie and Turkington 2002) or delayed germination (Turkington et al. 2005 ). Any of these responses may have subsequent effects on later life history stages such as survival and growth. Turkington et al. (2005) observed that delayed germination did not significantly effect survival though seedlings that germinated earlier, tended to be larger at the end of the growing season. In this study there was no relationship between number of seeds emerging and date (data not shown) suggesting t hat there was no change in the timing of emergence. Adding water significantly increased emergence indepe ndently of density; however, there was also a significant intera ction of fertilizer and water. The low water treatment coupled with high fertilizer had significantly lower emergence than any other  70  combination with water. Although it has been demonstrated that fertilizer addition can reduce survival of one species common to this system, Anenome (Arii and Turkington 2002), there is no other evidence to suggest that increased fertilization reduces germination. The overall effect of water is to increase germination rates and counter any negative effect of fertilizer addition. Given that the water and density interaction term is not significant, it is not likely that exploitive competition for water was important at least at the emergence stage. It is possible that allelochem icals or leachates would be washed away in the high water treatment; howe ver, there was still a significant relationship between density and emergence and this was not related to water (i.e. the interaction term was not significant). Seeds are somehow inhi bited from germinating due to their high density and increased water may simply increase germination rates as has been demonstrated for some arctic species (Oberbauer and Miller 1982). Although it seems unlikely that exploitive competition occurs at the emergence stage, I can only speculate that direct interference competition is important. Survival at the community level (i.e. counting all seedlings regardless of species identity) was positively related to density indicating a facilitative relationship. T he only plots that had low survival were the lowest density where the potential for interactions between plants are presumed to be the lowest. Although there is a well established negative relationship between density and fitness , there has not been a general realization that increased population density can also increase survival (Bruno et al. 2003). Positive interactions between pl ants has only recently become a common research topic (Callaway 1995; Callaway 1997; Brooker and Callaghan 1998). The specific mechanism for the increased survival detected in high density plots is not clear. Often seedlings benefit by having adult neighbours (Chapter 2). In studies showing traditional nurse plant effects the larger adult plant shelters the smaller plants reducing evapotranspiration losses and moderating temperature extremes (Callaway 1995; Holmgren et al. 1997). In this study I did not have the usual disparity between the sizes of neighbours. If low water was i mportant, I would expect that water add ition would alleviate some of this effect. Although, water did not have a significant effect on seedling survival at the corn munity level, when examined at the species level, half of the plant species demonstrated a positive response to watering with no species showing a negative reaction. Surprisingly, of the 4 species that demonstrated a significant relationship with density, half of those responded positively and half negatively to increased density. So although the community as a whole benefited from increasing  71  density, not all species did. Fertilizer was more consistent in its effect by reducing survival at the comm unity level for 5 out of the 8 species that survived to the end of the study. Only one species, Mertensia, had increased survival with fertilization. As already mentioned, in some previous studies Anenome responded negatively to fertilization showing both decreased survival and growth (Chapter 2, Arii and T urkington 2002). It was speculated that this may be a toxic reaction to the fertilizer (Arii and Turkington 2002). Overall, it is possible that close proximity to neighbours reduces the surface temperature of both the soil surface and the leaf surface to reduce evaporation and transpiration as in the nurse plant effect. It is not clear whether the plants in our study, which were quite small, were actually large enough to protect each other. Commonly, increased densities of neighbours leads to decreased growth (Goldberg et al. 1999; Goldberg et al. 200 1). The final mean plant mass at the community level was negatively related to den sity; however, at the species level only 2 species were negatively related to density and one specie s was positively related. The species-specific observations are a very poor predictor of community-level patterns. Also, fertilizing decreased the mass of 2 species and watering increased the mass of 3 species. Neither fertilizer nor water alone had a significant effect on the community-level mean plant biomass. Water and density interacted such that the highest plot biomass was achieved in plots with the lowest planting density with the low water treatment. If exploitive interspecific competition is occurring, we would expect the observed decrease in growth with an increa se in density; however, it is not clear what specifically the plants are competing for. My results are parallel to Goldberg et al. (2001) who demonstrated a switch in biotic interactions from competitive to facilitative and back to competitive as life history stages progressed. In general, studies that have measured the effect of competition through different life history stages tend to show that neighb ours are less likely to negatively affect survival even though they may have severe negative effects on subsequent growth (Goldberg et al. 1999; Howard and Goldberg 2001). On the whole, the effect of neighbours on developing seedlings m ay be neutral if corn petitive interactions and facilitative interactions are occurring simultaneously (Callaway and Walker 1997; Holmgren et al. 1997; Brooker and Callaghan 1998; Howard and Goldberg 2001). The positive influence of neighbours early in the development of seedlings, especially during emergence and early survival, is usually associated with the beneficial shading of larger neighbours that reduce temperature and moisture extremes (Callaway  72  1995; Callaway and Walker 1997). There are also other examples showing a switch from facilitation being important early in a seedling's life due to the neighbours sheltering them from predators, followed by a switch to neighbours competing with them as growing plants (Zanini et al. 2006). Fertilizer had a significant negative effect on plant survival at the community- and species-level in this study and this resulted in red uced final plant density and reduced species richness. Normally, increased fertilizer increases community biomass and decreases diversity (DiTomamaso and Aarssen 1989; Gough et al. 2000; Rajaniemi 2003). While the specific shape of the relationship between productivity and diversity is often debated and may be either linear, curvilinear or unimodal, nevertheless, it is generally agreed that competition plays an important part in the observed reduction of diversity (Grime 1973; Grime 1979; Huston 1979; Tilman 1982; DiTomamaso and Aarssen 1989; Gough et al. 2000; Rajaniemi 2003). In this case, although species richness declined with fertilization, there was no increase in final biomass and the largest effects of fertilizer were to increase the death rate at the survival stage. Both water and density increased survival at the species- and community-levels resulting in an incre ase in species richness. Overall, watering consistently increased survival at the community level and increased em ergence, survival and shoot mass for some of the species. These opposing effects of fertilizer and water have recently been reported by Stevens et al. (2006). Our results differ in that I did not observe any increase in comm unity-wide mean plant biomass with fertilizing or watering, although two individual species did increase in mean mass with watering. The increase in species richness was predominately due to increased emergence and survival which increased the chance of at least one individual from each spec ies surviving until the end of the season. Although species richness is expected to increase with density simply because more indivi duals are sampled (Goldberg and Estabrook 1998), I ob served less species richness than would be predicted from the null c ommunity. Results similar to these have been purported to show that competition, specifically competitive exclusion, leads to reduced species richness (Goldberg and Estabrook 1998 ). In our study, species richness may be lower than expected due to a number of factors such as reduced germination with increased density or negative fertilizer effects acting on either germination or survival. These factors have been shown to affect species d ifferently and may be the cause for lower than expected species richness.  73  These species-specific responses highlight a pro blem with traditional approa ches to density dependence and to competition studies in general. Here, species within a community mixture responded very differently to the treatments. Little could be said about general trends if only one or two species were chose n in a study unless all species performed similarly. If species are affected differently by competition, there will be the potential for changes in diversity (Rajaniemi and Goldberg 2000). In a similar study in an old field community, Rajaniemi and Goldberg (2000) reported that species' individual growth responded to increased competit ion, yet there was no corresponding change in diversity. Both that study and this demonstrate that individual-level patterns are not necessarily useful for predicting community-level changes. It is also important to measure the species' and community's responses over more than one life stage since one stage may not respond the same as another, even within the same species (Goldberg et al. 2001; Lortie and Turkington 2002). The CDS technique offers a novel approach to investigate questions about how both regulation and limitation at the species- and community-level affect community structure. Other approaches have been used to generate a null community, which is one without interactions, such as the combined monocultures (CM) approach (Campbell and Grime 1992; Goldberg 1994). The CM approach compares the performance of a species in monocultures to standardize the performance of a species in mixtures. It has been shown that the CDS gi ves more consistent results which are less likely to depend on abundance than the CM approach (Zamfir and Goldberg 2000). A potentially interesting question arising from these results and those in Chapter 2 is the role that community density has in recruitment of new seedlings. Given that higher densities facilitate the survival of seedlings and the presence of neighbours increases survival and biomass of transplants (Chapter 2), we might predict that a moderate to high density community would be the ideal place for new seeds to germinate and grow in this community. Similarly, areas with lower density might have lower seedling survival. T his may make areas of low vegetation cover especially slow to colonize. One possible criticism of this study is that it may not adequately represent environmental conditions or be a realistic scenario which might normally occur in the boreal understory. A much more plausible scenario would be to manipulate the density of intact plants in the field to differing levels. To our knowledge this has only been attempted once before in a perennial oldfield system, al though the experiment was only  74  conducted over one growing season ( Rajaniemi and Goldberg 2000). A preliminary experiment under greenhouse conditions would be a logical first step in exploring community-level density dependence in the field. If significant plant interactions are not observed in controlled conditions, it its unlikely that they would be detected or be of any importance in the field (Gibson et al. 1999). Given that this experimental community demonstrated density dependence, I now ask what role plant interactions may have in regulating the boreal understory plant community under natural conditions. In the next chapter I use the CDS with intact plants in a natural community over multiple years.  75  REFERENCES Arii, K. and Turkington, R., 2002. Do nutrient availability and competition limit plant growth of herbaceous species in the boreal forest understory? Arctic, Antarctic, and Alpine Research, 34: 251-261. Baskin, C.C. and Baskin, J.M., 1998. Seeds: ecology, biogeography and evolution of dormancy and germination. Academic Press, San Diego, CA. Bertness, M.D., 1998. Searching for the role of positive interactions in plant communities. Trends in Ecology and Evolution, 13: 133-134. Brooker, R.W. and Call aghan, T.V., 1998. The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oi kos, 81(1): 196-207. Bruno, J.F., Stachowicz, J.J. and Bertness, M.D., 2003. Inclusion of facilitation into ecological theory. Trends in Ecology and Evolution, 18: 119-125. Callaway, R.M., 1995. Positive interactions among plants. Botanical Review, 61(4): 306349. Callaway, R.M., 1997. Positive interactions in plant communities and the individualisticcontinuum concept. Oecologia, 112: 143-149. Callaway, R.M. and Walker, L.R., 1997. Competition and facilitation: a synthetic approach to interactions in plant communities. Ecology, 78(7): 1958-1965. Campbell, B., D. and Grime, J.P., 1992. An experimental test of plant strategy theory. Ecology, 73(1): 15-29. Carrier, P., 2003. Effects of water addition on biotic and abiotic con, ponents of a dry boreal forest in the Yukon. Ph.D. Thesis, University of British Columbia, Vancouver, BC. Carrier, P. and Krebs, C.J., 2002. T rophic effects of rainfall on Clethrionomys rutilus voles: an experimental test in a xeric boreal for est in the Yukon Territory. Canadian Journal of Zoology, 80: 821-829. Cody, W.J., 1996. Flora of the Yukon. NRC Research Press, Ottawa, Ontario, Canada. Connell, J.H., 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. American Naturalist, 122(5): 661696. Crawley, M.J., 1997. The structure of plant communities. In: M.J. Crawley (Editor), Plant Ecology. Blackwell Scientific, Oxford, pp. 475-531. DiTomamaso, A. and Aarssen, L.W., 1989. Resource manipulations in natural vegetation: a review. Vegetatio, 84: 9-29. Dyer, A.E., Fenech, A. and Rice, K.J., 2000. Accelerated seedling emergence in interspecific competitive neighbourhoods. Ecology Letters, 3: 523-529. Gibson, D.J., Connolly, J., Hartnett, D.C. and Weidenhamer, J.D., 1999. Designs for greenhouse studies of interactions between plants. Journal of Ecology, 87(1): 116. Goldberg, D.E., 1994. Influence of competition at the community level: an experimental version of the null models approach. Ecology, 75(5): 1503-1506. Goldberg, D.E. and Barton, A.M., 1992. Patterns and consequences of interspecific competition in natural communities: a review of field experiments with plants. American Naturalist, 139(4 ): 771-801. Goldberg, D.E. and E stabrook, G.F., 1998. Separating the effects of number of individuals sampled and competition on species diversity: an experimental and analytic approach. Journal of Ecology, 86: 983-988.  76  Goldberg, D.E., Rajaniemi, T., Gurevitch, J. and Stewart-Oaten, A., 1999. Em pirical approaches to quantifying interaction intensity: competition and facilitation along productivity gradients. Ecology, 80(4): 1118-1131. Goldberg, D.E. and Scheiner, S .M., 1993. ANOVA and ANCOVA: field competition experiments. In: S.M. Scheiner and J. Gurevitch (Editors), Design and analysis of ecological experiments. Chapman and Hall, New York, pp. 69-93. Goldberg, D.E., Turkington, R. and Olsvig-Whittaker, L., 1995. Quantifying the community-level consequences of competition. Folia Geobotanica Phytotaxonomica, 30: 231-242. Goldberg, D.E., Turkington, R., Olsvig-Whittaker, L. and Dyer, A.R., 2001. Densitydependence in an annual plant community: variation among life history stages. Ecological Monographs, 71: 423- 446. Gough, L., Osenberg, C.W., Gross, K.L. and Coll ins, S.L., 2000. Fertilization effects on species density and primary productivity in herbaceous plant communities. Oikos, 89: 428-439. Grime, J.P., 1973. Competitive exclusion in herbaceous vegetation. Nature, 242: 344347. Grime, J.P., 1979. Plant strategies and vegetation processes. John Wiley, Chichester, UK. Harper, J.L., 1977. Population biology of plants. Chapman & Hall, London, UK. Holmgren, M., Scheffer, M. and Huston, M.A., 1997. The interplay of facilitation and competition in plant communities. Ecology, 78(7): 1966-1975. Howard, T.G. and Goldberg, D.E., 2001. Competitive response hierarchies for germination, growth, and survival and their influence on abundance. Ecology, 82: 979-990. Hunter, A.F. and Aarssen, L.W., 1988. Plants helping plants. Bioscience, 38(1): 34-40. Huston, M., 1979. A general hypothesis of species diversity. American Naturalist, 113(1): 81-101. Keddy, P.A., 1989. Competition. Chapman and Hall, New York. Krebs, C.J., Boutin, S. and Boonstra, R., 2001. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York. Lortie, C.J. and Turkington, R., 2002. The effect of initial seed density on the structure of a desert annual plant community. Journal of Ecology, 90: 435-445. McMurray, M.H., Jenkins, S.H. and Longland, W.S., 1997. Effects of seed density on germination and establishment of a native and introduced species dispersed by granivorous rodents. American Midland Naturalist, 138: 322-330. Murray, B.R., 1998. Density-dependent germination and the role of seed leachate. Australian Journal of Ecology, 23: 411-4 18. Oberbauer, S. and Miller, P.C., 1982. Effect of water potential on seed germination. Holarctic Ecology, 5: 218-220. Rajaniemi, T.K., 2003. Explaining productivity-diversity relationships in plants. Oikos, 101: 449-457. Rajaniemi, T.K. and Goldberg, D.E., 2000. Quantifying individual- and co mmunity-level effects of competition using experimentally determined null species pools. Journal of Vegetation Science, 11: 433-442. SAS, 1995. JM P. SAS Institute Inc., Cary, North Carolina. Schoener, T.W., 1983. Field experiments on interspecific competition. American Naturalist, 122(2): 240-285. Shilo-Volin, H., Novoplansky, A., Goldberg, D.E. and Turkington, R., 2005. Density regulation in annual plant corn munities under variable resource levels. Oikos, 108: 241-252.  77  Silvertown, J.W. and Lovett Doust, J., 1993. Introduction to plant population biology. Blackwell Scientific, Cambridge. Smith, B. and Wilson, J.B., 1996. A consumer's guide to evenness indices. Oikos, 76: 70-82. Stevens, M.H.H., Shirk, R. and Steiner, C.E., 2006. Water and fertilizer have opposite effects on plant species richness in a mesic early successional habitat. Plant Ecology, 183: 27-34. Tilman, D., 1982. Resource competition and community structure. Princeton University Press, Princeton. Turkington, R., Goldberg, D.E., Olsvig-Whittaker, L. and Dyer, A.E., 2005. Effects of density on timing of emergence and its consequences for survival and growth in two communities of annual plants. Journal of Arid Environments, 61: 377-396. Turkington, R., John, E., Krebs, C.J., Dale, M.R.T., Nams, V.O., Boonstra, R., Boutin, S., Sinclair, A.R.E. and Smith, J.N.M., 1998. The effects of NPK fertilization for nine years on boreal forest vegetation in northwestern Canada. Jour nal of Vegetation Science, 9(3): 333-346. Turkington, R., John, E., Watson, S. and Seccombe-Hett, P., 2002. The effects of fertilization and herbivory on the herbaceous vegetation o f the boreal forest in north-western Canada: a 10- year study. Journal of Ecology, 90: 325-337. Zamfir, M. and Goldberg, D.E., 2000. The effect of initial density on interactions between bryophytes at individual and community levels. Journal of Ecology, 88: 243-255. Zanini, L., Ganade, G. and Hiibel, I., 2006. Facilitation and competition influence succession in a subtropical old field. Plant Ecology, 185: 179-190. Zobel, M., 1992. Plant species coexistence: the role of historical, evolutionary and ecological factors. Oikos, 65: 314-320.  78  Chapter 5 COMMUNITY- AND SPECIES-LEVEL CONSEQUENCES OF COMPETITION IN AN UNPRODUCTIVE ENVIRONMENT' INTRODUCTION  Competition is important in structuring plant communities (Connell 1983; Schoener 1983; Keddy 1989; Goldberg an d Barton 1992; Goldberg et al. 1999), but the intensity at which it occurs is dependent upon local conditions. The way in which the intensity of corn petition changes along gradients of productivity has been one of the most controversial debates in plant ecology (Goldberg and Novoplansky 1997) and there are two main theories with sometimes conflicting predictions. Grim e's theory predicts that competition intensity increases as the productivity of a habitat increases (Grime 1977; Grime 1979) while Tilman's model predicts that competition intensity remains constant along a gradient of productivity (Tilman 1982; Tilman 1988). Both models have been independently tested and Grim e's predictions have been corroborated (Huston 1979; Keddy 1989; Keddy 1990) while Tilman's predictions are have also been corroborated (Newman 1973; Grubb 1985; Taylor et al. 1990). Both theories predict that the intensity of competition is high at high levels of productivity so the major disagreement between them concerns the role of competition in unproductive environments (Abrams 1995). The idea that competition is not important in stressful environments was not new at the time of Grime's work and goes back at least as far as Darwin (1859). For example, Darwin stated, "When we travel northward...the num ber of species of all kinds, and therefore competitors, decreases" and that "when we reach the Arctic regions, or snow-capped sum mits, or absolute deserts, the struggle for life is exclusively with the elements" (Darwin 1859). While the important role of abiotic stress in environments such as northern tundra, alpine tundra or arid and semiarid regions is unquestionable, the re is evidence to ' A version of this chapter will be submitted for publication. Treberg, M. A. and Turkington, R. (2008). Community- and species-level consequences of competition in an unproductive environment.  79  suggest that biotic interactions play a role in stru cturing plant communities in stressful arid and semiarid environments (Fowler 1986; Gurevitch 1986; Kadm on 1995; Goldberg and Novoplansky 1997; Goldberg et al. 2001). Similarly, there are many examples of facilitation in and environments (Callaway 1995; Flores and Jurado 2003) leading some to suggest that facilitation may in fact be more common in severely stressed communities (Callaway and Walker 1997; Brooker and Callaghan 1998; Callaway et al. 2002; Lortie and Callaway 2006). There are n o thorough reviews of biotic interactions in northern stressed environments such as arctic or alpine tundra (for minor reviews see Arii and Turkington 2001, and Chapter 1), although there are many examples of both competition (Aerts et al. 1990; Shevtsova et al. 1995; Arii and Turkington 200 2) and facilitation (Maillette 1988; Morris and Wood 1989; Carlsson and Callaghan 1991; Shevtsova et al. 1995; Callaway et al. 20 02). These studies demonstrate that biotic interactions, both com petition and facilitation, are prevalent in natural communities in a wide range of en vironments, yet they tell us little about the overall effect of competition and facilitation on the community, only the effect or response of certain species. The majority of research has focused on the measurement of competition effects on individual fitness (Goldberg and Barton 1992; Goldberg 1994; Goldberg et al. 1995 ). Theoretical considerations suggest that individual-level data will not necessarily, or perhaps even usually, predict community patterns, even on a local spatial scale (Goldberg 1994; Kareiva 1994) and recent empirical evidence has shown that indiv idual-level effects of competition could not predict community-level effects (Rajaniemi and Goldberg 2000). Two newly developed methods for directly examining the effects of competition on community structure use multi-species mixtures. The first is the combined monoculture method, which uses the species' performance in monocultures to standardize the performance in mixtures (Goldberg 1994). The second is the community density series (CDS) that uses the p erformance of species in very low-density mixtures (i.e. a null community without interactions) to standardize performance at high-density mixtures (Goldberg et al. 1995; Goldberg et al. 2001). These two methods have been compared recently and it was concl uded that the results obtained by the CDS are more valid for determining competitive hierarchies and that the CDS method is a more useful method for investigating the effects of competition on the comm unity (Zamfir and Goldberg 2000).  80  The community density series (CDS) was first described by Goldberg et al. (1995). It is a multi-species version of the traditional single-species yield-density experiments (Harper 1977). Traditionally, a single species would be planted in den sities ranging from very low (where intraspecific corn petition would not occur) to very high densities (where intraspecific competition would be very high). As density increases, initially the species' yield would increase linearly and at some density, intraspecific competition would begin to reduce the rate of increase until finally the yield reached a plateau. This is often called the law of constant final yield (Kira et al. 1953). Goldberg et al. (1995) argued that the density of a community could be manipulated in the same way so as to obtain densities below and above the natural condition of the community and predicted that a similar pattern would result. The lowest density plots, where density is low enough to preclude interactions, characterize the "null" community and this can be compared to higher density plots w here biotic interactions are affecting the plant community as a whole (Zobel 1992; Goldberg et al. 1995). Also, each species in the CDS can be considered separately to determine if they respond similarly to changing density. Another advantage of the CDS is that both negative and positive density dependent processes are detected. The influence of productivity on these responses can also be investigated by examining changes in the yield-density relationship. For example, if productivity is increased, we may expect an increase in the constant final yield or an increase in the intensity of competition. One previously unmentioned advantage of the CDS is that the slope and the R 2 from the regression of the yield-density relationship tell us a bout both the intensity and importance of competition respectively (sensu Welden and Slauson, 1986). The earliest researchers using yield-density experiments realized that the slope of the regression of yield and density is an index of the intensity of competition (Kira et al. 1953). This has been rediscovered more recently (Welden and Slauson 1986; Aarssen and Epp 1990). In traditional yield-density studies, the coefficient of determination, or R2 , from simple linear regression (Welden and Slauson 1986; Aarssen and Epp 1990; Weigelt and Jolliffe 2003) or it's multivariate equivalent (McLellan et al. 1997; Sammul et al. 2000) can be interpreted as being the importance of competition. The R2 represents the importance of competition, over other possible factors affecting yield, because it is the proportion of variation in yield that is directly due to the density (Welden and Slauson 1986; Weigelt and Jolliffe 2003). The ability to quantify both the intensity and importance of competition can help untangle some of the debates surrounding the role of  81  competition in structuring plant communities (Welden and Slauson 1986; Brooker et al. 2005). This is described further in Chapter 3. The CDS has been successfully applied in both ex perimental and natural communities of annual plants in the Negev Desert (Goldberg et al. 2001; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 2005) in an experimental bryophyte community (Zamfir and Goldberg 2000), for a single season in an old-field community (Rajaniemi and Goldberg 2000) and in an experimental boreal understory community (Chapter 3). However, it has never been applied in a multi-year study in a perennial system. Using the CDS, I investigated how competition structures an unproductive boreal understory plant community. Specifically, I asked: 1. Is the boreal understory affected by competition at the community level? 2. Does the productivity of the system affect the role of competition at the community level? 3. At what community density does competition begin to have an effect? 4. At what community density does the maximum constant final yield begin? 5. Does each of the species in the system respond similarly to density? 6. Does the productivity of the system change the species-specific responses to increased density?  METHODS STUDY SITE The study site is located within the boreal forest close to Kluane Lake in the southwestern Yukon Territory (138 ° 16' W; 61 ° 00'N) at an approximate altitude of 1000 m above sea level. This ecosystem is extensively studied for both its animal and plant components and was used for the Kluane Boreal Ecosystem Project (Krebs et al. 2001). Previous research has shown this ecosystem is nutrient limited (Turkington et al. 1998; Turkington et al. 2002), may be water limited (Carrier and Krebs 2002; Carrier 2003) and biotic interactions affect some plant species (Arii and Turkington 2002). Although there are herbivores such as snowshoe hares, red squ irrels and microtine rodents at this site, the abundance of the ground layer is more affected by the lim ited soil nutrients than by herbivores (John and Turkington 1995; John an d Turkington 1997; Turkington et al. 2002).  82  White spruce (Picea glauca (Moench) Voss s.I.) is the dominant tree species and in 2003 had a density of 583 stem s ha -1 (95% CI of 486 to 697 ha -1 ). Beginning in 1995, an outbreak of spruce bark beetle caused the death of many of the overstory trees, resulting in a rather open canopy. Many smaller willows (predominantly Salix glauca L. s. I.) and some dwarf birch (Betula glandulosa Michx.) make up the shrub understory. The ground layer or understory species (Table 1) h ad a mean biomass of 196 g m  -2  (n =  9 x 1 m 2 quadrats, 95 % CI of 141 to 250 g m -2 ) in 2002. The cover of moss and lichens is < 5 %.  Table 5.1. The abundance of all species found in the 63 1 m 2 CDS plots during the initial survey in 1999. Frequency is the percent occurrence in the 63 plots. Percent cover was estimated using a point frame with 100 pin drops per m 2 . Density was assessed by counting all individuals in the 1 m 2 plot. The density of Arctostaphylos uva-ursi, Festuca and Linnaea were not estimated (n/a) due to the difficulty in identifying distinct individuals. Species Achillea millefolium ssp. borealis Anenome parvitiora Antennaria spp. Arabis spp. Arctostaphylos rubra Arctostaphylos uva-ursi Artemisia norvegica Aster spp. Betula spp. Calamagrostis spp. Carex spp. Corpus canadensis Delphinium glaucum Draba spp. Epilobium angustifolium Festuca altaica Gentiana spp. Linnaea borealis Lupinus arcticus Mertensia paniculata Moneses unitlora Orthilia secunda Pedicularis spp. Picea glauca Polemonium spp. Pyrola spp. Salix spp. Senecio lugens Sheperdia canadensis Solidago multiradiata Stellaria longipes Trisetum spicatum  Frequency (%) 84.13 11.11 31.75 3.17 7.94 58.73 6.35 1.59 11.11 25.40 23.81 20.63 14.29 7.94 53.97 100.00 39.68 90.48 85.71 57.14 20.63 11.11 1.59 12.70 1.59 1.59 3.17 80.95 9.52 90.48 4.76 7.94  Mean percent cover ± S.E. 1.58 ± 0.26 0.79 ± 0.31 ± 0.04 0.18 0.02 ± 0.01 0.67 ± 0.57 6.26 ± 1.29 0.06 ± 0.04 0.01 ± 0.01 ± 0.04 0.10 0.15 ± 0.04 0.25 ± 0.07 0.28 ± 0.10 0.10 ± 0.04 0.04 ± 0.02 1.17 ± 0.25 14.21 ± 1.25 0.25 ± 0.05 25.33 ± 2.68 ± 0.42 2.68 3.29 ± 0.81 0.14 ± 0.04 0.14 ± 0.07 0.01 ± 0.01 0.16 ± 0.07 0.03 ± 0.03 0.02 ± 0.02 0.02 ± 0.01 1.36 ± 0.21 0.07 ± 0.04 1.20 ± 0.15 0.02 ± 0.01 0.06 ± 0.03  Mean density (no./m2) ± S.E. 21.97 ± 2.83 8.14 ± 3.15 1.11 ± 0.27 0.06 ± 0.05 5.83 ± 4.73  n/a  0.90 0.03 0.24 2.43 3.87 3.05 0.27 0.57 3.54  ± 0.68 ± 0.03 ± 0.10 ± 0.82 ± 1.20 ± 1.11 ± 0.10 ± 0.30 ± 0.65  7.64  ± 2.64  n/a n/a  27.38 12.59 1.67 1.14 0.02 0.16 0.03 0.02 0.05 8.52 0.21 11.97 0.49 0.79  ± 4.59 ± 2.44 ± 0.59 ± 0.49 ± 0.02 ± 0.06 ± 0.03 ± 0.02 ± 0.04 ± 1.22 ± 0.11 ± 1.36 ± 0.34 ± 0.42  83  EXPERIMENTAL DESIGN In the spring (late May) of 1999, 63 plots of 1 m x 1 m were delineated in a n area of approximately 25 m x 75 m. These plots were located in patches with representative samples of the vegetation com mon to the understory community in this forest. Plots were in small groups of 2 to 5, with a minimum of 1 m between adjacent plots. Each group of plots was surrounded by 1 m high 2.5 cm mesh galvanized chicken wire fence. All plants in each of the 63 plots were identified to species and counted. Individuals were identified as separate shoots or ramets and may have been connected underground. Percent cover of all species in the plots was estimated using a point quadrat frame, held above the top of the understory vegetation on stakes, with 100 pin drops (Table 5.1). The nine most abundant species were chosen to b e included in the community density series (CDS). Seven of the species are herbaceous perennials: Achillea millefolium L. ssp. borealis (Bong.) Breitung (yarrow), Epilobium angustifolium L. s.l. (fireweed), Festuca altaica Trin. (northern rough fescue), Lupinus arcticus Wats. (arctic lupine), Mertensia paniculata (Ait.) G. Don var. paniculata (bluebells), Senecio lugens Richards. (black-tipped groundsel), Solidago multiradiata Ait. (goldenrod). The remaining two species are woody perennials: Arctostaphylos uva-ursi (L.) Spreng. s.l. (bearberry) and Linnaea borealis L. ssp. americana (twinflower). Hereafter I will refer to species using their generic name. All nomenclature follows Cody (1996). A geometric series of six Initial Community Densities (ICDs) was constructed: 1/16, 1/8, 1/4, 1/2, 1 and 2 x the natural field density. The x1 density closely approximated the density of the natural vegetation estimated from the initial survey of the community. All plots in the CDS were manipulated by transplanting and removing plants such that the relative proportion of the 9 most common species was consistent for all densities. Without exception, every plot in the CDS had some plants added and some removed to obtain the proper proportions of the 9 study species. To increase the density in plots, transplants were taken from the surrounding vegetation as either large sods containing many individuals (and sometimes many species) or as single shoots. Removal was accomplished by cutting the unwanted shoots off at ground level. In the lower density plots, an attempt was made to keep the remaining plants approximately equidistant from each other. Density manipulations began in mid-June, 1999, and were completed by mid-July. Some regrowth of removed plants occurred but this was  84  removed before the end of the season survey done in the final 2 weeks of August. No plants were added at this time. In 2000, minor weeding was completed in June to adjust to the desired densities. No other density manipulation was required. Fifty four of the 63 1 m 2 plots were randomly assigned to the CDS and the fertilizer treatments. The remaining 9 plots were used as unmanipulated controls meaning they did not have any density manipulation. They were randomly assigned to one of the three fertilizer levels. The soil surrounding all plots was cut to a depth of approximately 25 cm just outside of the 1 x 1 m perimeter to sever any belowground connections between plant inside and ou tside the plots. Three levels of fertilizer addition were used — low (control), medium and high. Granular fertilizer (N-P-K; 21-7-7) was added after snowmelt, at the end of May or early June for each of the 4 years of the study. Fertilizer was added at a rate of 13.125 g N m -  4.375 g  m -2 y1 and 4.375 g K m y' -2 for the medium fertilizer treatment and at a 2 y -1 , 8.75 g p m -2 y 1 rate of 26.25 g N m" and 8.75 g K m"  2 y',  P  2  y -1 for the high treatment.  These are within the range of application rates from other studies in this area that demonstrated a significant effect of fertilizer add ition (John and Turkington 1997; Turkington et al. 1998; Dlott and Turkington 2000; Arii and Turkington 2002; Turkington et al. 2002). The low fertilizer treatment ( control) had no fertilizer added. Because I wanted to estimate interannual variation in biom ass within the plots, yet could not destructively harvest, surrogate measures of biomass were used to approximate the biomass for each species. In July, 1999, 20 randomly placed plots were sampled for percent cover of each species. The plots were clippe d to ground level, sorted to species, dried, and the relationship between each species cover and biomass was determined. The biomass of Arctostaphylos, Festuca, and Linnaea was accurately estimated by percent cover. The width of the widest leaf, the long est leaf, the number of leaves and maximum height were measured fo r random individuals of the remaining herbaceous species. These were then cut at ground level and dried. These measurements were then related to shoot mass and the best fitting relationships were determined (Table 5.2). All relationships were statistically significant and R 2 values ranged from 0.62 to 0.97. For each year of the study, during peak biomass, which occurs approximately at the end of July, all individuals in each CDS plot and control plot were measured and the plot biomass estimated. At the end of August, 2002, all plants were counted in the CDS and control plots. All aboveground biomass was removed and each individual was bagged separately  85  before being air dried for transport to the University of British Columbia in Vancouver, Canada. All samples were oven dried at 60 °C for 48 hours and weighed to the nearest mg.  Table 5.2 The equations used to estimate biomass for the years 1999 through 2001 for all species growing in the CDS and control plots. These equations were the best fitting curves between biomass and various surrogates for biomass and were based on destructive sampling done in 1999. In the equations: C = cover, H = height, L = length of the longest leaf, N = number of leaves, and W = width of longest leaf. All equation components are measured in mm and all estimated masses are in grams. Species Achillea millefolium ssp. borealis Arctostaphylos uva-ursi Epilobium angustifolium Festuca altaica Linnaea borealis Lupinus arcticus Mertensia paniculata Senecio lugens Solidago multiradiata  Plant form non-flowering flowering n/a  non-flowering flowering n/a n/a  non-flowering flowering non-flowering flowering non-flowering flowering non-flowering flowering  Biomass equation L x 0.0005 H x 0.001 C x 3.234 H x 0.0021 H x 0.0032 C x 1.099 C x 2.013 W x 0.0045 W x 0.0065 + H x 0.0005 L x 0.0011 H x 0.0012 N x 0.016 + L x 0.0006 L x 0.0009 + H x 0.0019 L x 0.0011 L x 0.021 + H x 0.001  n 34 10 17 17 6 17 16 19 11 21 14 21 9 20 10  R2 0.816 0.701 0.637 0.672 0.974 0.878 0.867 0.618 0.800 0.814 0.808 0.870 0.745 0.870 0.658  P  <.0001 0.0025 <.0001 <.0001 0.0003 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.0165 <.0001 0.0234  ANALYSIS The effects of density and fertilizer addition on the CDS were examined using analysis of covariance (AN COVA) with density as the covariate (because it is a continuous variable) and fertilizer as a categorical variable with 3 levels. All analyses were done using JM P 4 (SAS 1995). The effect of density on the unde rstory community was analyzed using an indivi dual performance approach (Goldberg et al. 2001). By examining the average performance of the individual in relation to density, any non-zero slope would indicate den sity dependence. For example, an increase in performance with density, whether linear or nonlinear, would indicate positive density dependence or facilitation. A negative slope would indicate neg ative density dependence or competition. Because the ANCOVA requires a linear covariate, fou r transformations were used to linearize the data: linear, power, sem ilog and reciprocal. The best fitting model, with the highest R 2 , is reported. These transformations also assisted in making the ANCOVAs better meet the usual statistical assumptions. If density was not  86  significantly related to the response variable, standard ANOVA was used to determine the effect of fertilizer addition. At the community level, the effect of density was examined on the mean plant mass for all 4 years of the experiment. Because I could not count the exact num ber of individuals for Arctostaphylos, Festuca, and Linnaea, and percent cover was used as the measure of abundance, an index of plant mass was used. The mean plant size index was the total mass of the plot divided by the density. The value obtained was higher than the actual mean plant mass, but was linearly related to it and was independent of density. Therefore, the same conclusions of density dependence as described can be reached if the index demonstrates a relationship with density (i.e. a non-zero slope). As reported in Chapter 4, and also in Goldberg et al (2001) and Lortie and Turkington (2002), the choice of which covariate measure of density to use made little difference to the interpretation of the results. In Chapter 4, I used the final plant density as the covariate for analyzing the final mean biomass responses. In that case, density did change from one life history stage to the next. In this study, the density of most species did not change after the first two field seasons. There was some death in the highest density plots resulting in fewer plants. The overall effect of this would be to weaken the relationship and make it slightly more difficult to detect negative density dependence. Therefore, the relative densities from the ICD were used as the covariate for all ANCOVAs. The density at which competition began to reduce the mean plant m ass was determined by regressing the lowest density plots' mean plant mass (where corn petition is presumed to be least important) against density. Initially, only the lowest two densities (x1/16 and x1/8) were included in the regression. If the regression had a significant slope, it was assumed that corn petition was responsible for the reduction in plant mass at that density. If the regression was not significant, successively higher densities were added until the regression had a significant slope; this was the density that competition began to have an effect. The density at which a final constant yield was reached was determined by regressing the final plot yield (total biomass) of the two highest densities against density. Lower density plots were added until the slope became significant. The last density that had a no n-significant slope is the density where a final co nstant yield was reached. The effect of density on the diversity in the CDS plots was also examined. The species richness (number of species in each plot) did not change in the 4 years of the  87  study; however, if density affects species differently, we may expect to see changes in their relative abundance, or evenness. The evenness in the final year was determined based on each species' mass in the plots to determine if density or fertilizer addition had any effect on the relative abundance of species. The index used was Eva, (Smith and Wilson 1996). Evenness ranges from 0 to 1. Evenness values close to 1 indicate that species are nearly uniformly abundant and values close to 0 indicate that one or a few species are much more abundant tha n the others. Therefore, if we detect any change in evenness with changing density, there would have to be species-specific changes in abundance. The effect of fertilizer addition on the plot biomass of the unmanipulated controls, those plots not part of the CDS, was also examined using ANOVA. The controls were also compared to the x1 CDS plots to determine if the final plot mass was similar between the manipulated plots and the unmanipulated plots and whether there was any difference in their response to fertilizer addition. Species-specific effects of density and fertilizer ad dition were examined on the final mean biomass for each plant species using ANCOVA with the relative planting density being used as the covariate. If density was not related to the final mean plant mass, ANOVA was used to determine the effect of fertilizer addition.  RESULTS COMMUNITY-LEVEL RESPONSES Negative density dependence of the mean plant size index (total plot mass divided by the density) was observed e ach year of the CDS experiment with the negative slope becoming increasingly steeper with each subsequent year (Table 5.3, Fig. 5.1). Therefore, the intensity of competition (the slope) increased each year. Similarly, the importance of competition (R 2 ) also increased each year. In all years the relationship between mean plant size and density was nonlinear (Table 5.3). The mean plant mass was significantly higher below the x1 density (the natural density of the community) (Fig. 5.1). The density that competition began to reduce mean plant mass was at x 1/8 for all years. Final constant yield was only reached in 2001 and 2002 at x1, the natural density observed in the field.  88  The mean plant size index was affected each year by plant density; however, fertilizer only had a significant effect in the final year, although there was a significant fertilizer and density interaction in the first year of the study (Table 5.4, Fig.5.2a). In the final year, fertilizer surprisingly had a negative effect on mean plant size with the highest growth in the unfertilized plots (Fig. 5.2a). Species evenness in the community in the final year was also affected by density (Table 5.3, Fig. 5.3) with the h ighest evenness in the higher density plots. The significant nonlinear relationship between evenness and density means that the relative proportion of each species' biomass was not constant, i.e. the species' responses were not consistent along a gradient of density. Evenness was also significantly different between fertilizer levels, irrespective of density, with the lowest evenness in the unfertilized plots (Table 5.4, Fig. 5.2b). There was no difference in the overall plot biomass between the unmanipulated controls and the x1 density plots (Table 5.4, Fig. 5.2c). There was also no effect of fertilizer addition on the plot biomass for either the control or the lx density plots (Table 5.4, Fig. 5.2c).  Table 5.3. Regression coefficients for the mean plant size index and density relationships for the  years 1999 through 2002 and the evenness and density relationship in 2002. These data are plotted in Figures 5.1 and 5.3. The mean plant size index is the total plot mass divided by the density. In the absence of interactions, there should be no relationship between the mean plant size index and density. The negative slopes indicate negative density dependence (or competition) for the plant size index. Model type refers to the data transformation that best linearizes the data. The degrees of freedom are for the model and error combined. Significant values (P < 0.05) are in bold.  Variable Mean plant size index 1999 Mean plant size index 2000 Mean plant size index 2001 Mean plant size index 2002 Evenness (Evar)  Model power power power power semiloq  df 53 53 53 53 53  Intercept 4.658 4.488 4.427 5.096 0.338  Slope -0.113 -0.306 -0.407 -0.663 5.97x10-2  R2  P  0.495 0.617 0.613 0.742 0.309  <.001 <.001 <.001 <.001 <.001  89  •  500  500 ^  ^ 1999  x  43 400 c .rA 300  2000  rs 400 (1)ti.. 300 - •  • • 200 •  200 -  SS  co c& 100  (I) 100 0 ^  •  0  0^0.5^1^1.5^2  0  0.5^1^1.5  Relative plant density  -8  500 -•— • 400 -1  •  N 300- 1 •  <a  2  Relative plant density  2500  •  2001  2002  •-(S) 2000 • .c  •  1500 C  ja o. 1000 - •^• 500  •  ^  ^ ^ 1.5 2 0^0.5^1^1.5^2 ^ Relative plant density Relative plant density  0^0.5^1  Figure 5.1. The effect of density on the mean plant size index (the total plot mass divided by density) for the years 1999 through 2002. All curves shown are statistically significant (P < 0.001) and the coefficients for the best fit curve are given in Table 5.3. In the absence of interactions, there should be no relationship between the mean plant size index and density. A negative slope indicates negative density dependence (competition). The y-axis for 2002 has a different scale than the graphs for other years. The natural field density is x 1. The density that competition began to reduce mean plant mass was at x 1/8 (i.e. 0.125) for all graphs. The vertical dashed line represents the density that constant final yield is reached for 2001 and 2002. Constant final yield was not reached in 1999 or 2000.  90  • •  ^  500 450 -a) 400 350 .N 300 U) • 250 a 200 c 150 cti 100 50 0 x  a  I  Low  ^  Med  ^  High  Fertilizer Treatment  b  ^0.4  ^ 0.35 0.3  I-L 0.25 -  a  0.2  g 0.15 > ui^0.10.05 0 ^ Low  ^  Med  ^  High  Fertilizer Treatment  C  300 ^ Low Fert • Med Fert • High Fert  250 O 200 O 150  n 100 To o 50  -  .  -  Controls  ^  x1 Density  Figure 5.2. The effect of fertilizer level on (a) mean plant size index (total plot mass divided by density) in the CDS, (b) evenness in the CDS and (c) the total plot biomass for the controls (density not manipulated) and the x1 density plots in the CDS experiment. Error bars are ± 1 S.E.  All data are for 2002. Columns that share the same letter are not statistically different (Tukey's HSD, P > 0.05).  91  Table 5.4. Summary of ANCOVAs for the mean plant size index in the CDS for 1999 to 2002 and ANOVA for total plot biomass in the control plots in 2002. The control treatment for the total plot biomass ANOVA compares the mean of the unmanipulated control plots to the x1 density in the CDS plots. Significant values (P < 0.05) are in bold.  P  Variable Mean plant size index 1999  Source Density Fertilizer Density x Fertilizer Error  df 1 2 2 48  SS 0.961 0.018 0.158 0.804  F-ratio 57.362 0.539 4.712  <.001 0.587 0.014  Mean plant size index 2000  Density Fertilizer Density x Fertilizer Error  1 2 2 48  7.077 0.421 0.250 3.723  91.225 2.715 1.612  <.001 0.076 0.210  Mean plant size index 2001  Density Fertilizer Density x Fertilizer Error  1 2 2 48  12.546 0.080 0.172 7.665  78.564 0.251 0.538  <.001 0.779 0.588  Mean plant size index 2002  Density Fertilizer Density x Fertilizer Error  1 2 2 48  33.254 1.603 0.124 9.825  162.465 3.916 0.303  <.001 0.027 0.740  Evenness (Evan)  Density Fertilizer Density x Fertilizer Error  1 2 2 48  0.269 0.084 0.011 0.508  25.455 3.977 0.526  <.001 0.025 0.595  Controls (Total Plot Biomass)  Control Fertilizer Control x Fertilizer Error  1 2 2 11  3197 104.858 28025 52406  0.671 0.011 2.941  0.430 0.989 0.095  92  ^ ^  0.6 0.5 -  •  0.4 Lir^!^  1  ^co^ • 03] ^co^. ^a)  ^i •  •  c^ c^ a) °'Il >^' 0.2^ ^•  u..i  1•  •  •  •  0.1 OS • 0 0  ^  0.5^1  ^  1.5  ^  2  Relative plant density  Figure 5.3. The effect of plant density on Smith and Wilson's (1996) index of evenness (Evar). Lower values of E„r indicate that the species are very unequally represented in the community. The natural field density is x 1. The best fit curve is statistically significant (P < 0.001) and the coefficients for the curve are given in Table 5.3. SPECIES-LEVEL RESPONSES Like the community-wide responses, most species' mean plant mass were negatively and non linearly density dependent (Tables 5.5, 5.6, Fig. 5.4). Only two species' mean mass, Epilobium and Senecio were not related to density, although Mertensia and Solidago were only related to density at P < 0.10 (Table 5.5). The intensity of competition was highest for Linnaea and Arctostaphylos, (slopes of -0.897 and -0.851, respectively) and was also high for Festuca (-0.687). The importance of competition was highest for Festuca (R 2 of 0.719) with high values also for Arctostaphylos (0.402) and Linnaea (0.477). No species displayed positive density dependence (facilitation). Species-specific responses to fertilizer addition were more varied than the response to density. The prostrate woody shrubs, Arctostaphylos and Linnaea, were negatively affected by fertilizer add ition while Epilobium and Mertensia responded favorably to fertilizer addition (Table 5.6, Fig. 5.5). The remaining 5 species had no response to fertilizer addition, although Achillea and Senecio had responses that were marginally significant (0.05 < P < 0.10). A switch between no density dependence to de nsity dependence at low densities was observed for three species, Achillea, Arctostaphylos, and Festuca (Fig. 5.4). The  93  density that com petition began to be important could not be determined for the other species because all densities needed to be included before the regression had a significant slope. The final constant yield was reached for three species, Achillea, Festuca and Lupinus.  Table 5.5. Regression coefficients for the relationship between each species mean plant mass and density. The model type is the transformation that best linearized the data. The degrees of freedom (df) are for the model and error combined. Significant values (P < 0.05) are in bold and values where P< 0.10 are in italics. These data are plotted in Figure 5.4. Variable Achillea millefolium ssp. borealis Arctostaphylos uva-ursi Epilobium angustifolium Festuca altaica Linnaea borealis Lupinus arcticus Mertensia paniculata Senecio lugens Solidago multiradiata  Model semilog power linear power power power semilog linear linear  df 53 51 53 53 53 51 53 53 53  Intercept 0.226 1.965 2.016 3.892 3.492 -0.762 1.215 0.354 0.424  Slope -0.119 0.851 -0.355 0.687 -0.897 -0.211 -0.263 -0.114 -0.107 -  -  R2  P  0.235 0.402 0.026 0.719 0.477 0.161 0.068 0.045 0.063  <.001 <.001 0.247 <.001 <.001 0.003  0.056  0.124 0.068  94  •  •  ••  •  1.6  •  1.4  250 Achillea fnillefolium ssp. borealis  ';: 1.2  • •  'a' 0.8  •  7.0 Arctostaphylos uva-ursi  .8 200 g  5.0^•  150 ▪  - 0.6 c 2 0.4 0  5  • 0.2 0  •  100  • •  50  0.5^1^1.5  800 ▪  600  Festuca altaica  e• •  o.o^• ^  -• 2  0^0.5^1^1.5 Relative plant density  Linnaea borealis  • 700 a, 600  E  8 500 • E. 400^• •  la 300  • 300^• 1 200  200 2  3.0  „^  Relative plant density  •  8 500 -.E. 400  •  800  •  x 7 00  •g.  •  •  0^0.5^1^1.5  Relative plant density  Epilobium angustifolium  E 4.0  •  0 O  •  - 6.0  100  M 100  0 0^0.5^1^1.5  2  Relative plant density  6.0  0^^ • ^ ^ 0^0.5^1^1.5 2 ^ Relative plant density  1.8  E  • 3.0 •^•  2  1.0 I.^•  2.0  •  ^:^  0.5  Senecio lugens  1.4^  F; 0.8 • 0.6 • • 0.4 0/1 • 0.2  ••  1  1.5  Relative plant density  ^  2  0 SO^ " 0  Relative plant density  •  Solidago multiradiata  • 1.2  2 •  0.0 -A. O  1.6  Mertensia paniculata  F„- 5.0 co • 2 4.0  0.5^1^1.5  1  E 0.8 • •  • •  •  •^ •  0- 0.6 .• 00 •• 2 0.4  0. 2^/  0  •  0.5^1^1.5^2^0^0.5^1^1.5 Relative plant density^  Relative plant density  Figure 5.4. The effect of plant density on the mean plant mass or mean plant size index (total plot mass divided by density) for species in the CDS plots in 2002. All curves shown are statistically significant at P < 0.05, except for Mertensia and Solidago, which are significant at P < 0.10. The coefficients for the best fit curve are given in Table 5.5. In the absence of interactions, there should be no relationship between the mean plant mass or mean plant size index and density. A negative slope indicates negative density dependence (competition). The natural field density is x 1. Solid vertical lines indicate the density that competition begins to reduce the mean plant mass. Dashed vertical lines indicate the density where the final constant yield for that species was reached.  95  1  Table 5.6. Summary of ANCOVAs and ANOVAs on each species' mean plant mass in the CDS in 2002, in response to manipulations of density and fertilizer. When the effect of density was not significant (P < 0.05, Table 5.5) on the mean plant mass, an ANOVA with just fertilizer as an effect was performed. Significant values (P < 0.05) are in bold. Source Density Fertilizer Density x Fertilizer Error  df 1 2 2 44  SS 1.231 0.313 0.265 2.441  F-ratio 22.178 2.819 2.389  P <.001  Density Fertilizer Density x Fertilizer Error  1 2 2 44  53.322 22.371 0.154 55.591  42.204 8.853 0.061  <.001 <.001  0.941  Epilobium angustifolium  Fertilizer Error  2 51  36.434 84.914  10.941  <.001  Festuca altaica  Density Fertilizer Density x Fertilizer Error  1 2 2 44  779900 22318 12363 959761  35.754 0.512 0.283  0.603 0.755  Density Fertilizer Density x Fertilizer Error  1 2 2 44  63.621 9.488 3.585 48.573  57.632 4.297 1.624  Density Fertilizer Density x Fertilizer Error  1 2 2 44  3.261 0.462 0.572 15.259  9.404 0.666 0.825  0.519 0.449  Mertensia paniculata  Fertilizer Error  2 51  26.201 50.305  13.281  <.001  Senecio lugens  Fertilizer Error  2 51  0.620 6.532  2.422  0.099  Solidago multiradiata  Fertilizer Error  2 51  0.018 4.489  0.101  0.904  Species Achillea millefolium ssp. borealis  Arctostaphylos uva-ursi  Linnaea borealis  Lupinus arcticus  0.071 0.104  <.001  <.001 0.020  0.209 0.004  96  • •  Achillea millefolium sop. borealis  ^  Arctoslaphylos uva-ursi  ^  Epiloblum angustifolium  0.6  -0.; 0.5 • 0.4 F t 0.3 0.2 0.1 0 Low  ^ ^ ^ ^ ^ ^ High Low Med High Low Med High ^ ^ Fertilizer level Fertilizer level Fertilizer level  ^  Festuca altaica  Med  ^  ^  Unnaea borealis  250  250  -0  200  200  g  150  .11 150  100  100  g  3  g 50 0  ^  Lupinus ftfalCUS  50 0  Low^Med^High  Low^Med  Mertensia particulate  High  Low  ^  Soma(' lugens  Med  High  Fertilizer level  Fertilizer level  Fertilizer level  ^  Solidago muldradiata  2.5  0.5  2  162 0.4  E 1.5  0.3  S  0.2  a  0.1  • 0.5 0  Low  Med Fertilizer level  High  Low  0 Med^High^ Low^Med Fertilizer level^  High  Fertilizer level  Figure 5.5. The effect of fertilizer level on the mean plant mass (±1 S.E.) or mean plant size index (±1 S.E.) for each species in the CDS in 2002. Columns sharing the same letter are not statistically different (Tukey's HSD, P > 0.05).  DISCUSSION There are now a number of examples of entire plant comm unities following the same pattern as observed in traditional single-species yield-density experiments (Rajaniemi and Goldberg 2000; Zamfir and Goldberg 2000; Goldberg et al. 2001; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 2005; Chapter 4). This experiment is the first experiment to demonstrate the yield-density response in a perennial community in a multi-year experiment. This understory community demonstrated density dependence in all years of the study with both the intensity and importance of competition increasing each subsequent year. Unfortunately, none of the  97  previous CDS experim ents have purposely presented information on the intensity or importance of competition at the community level, though Zamfir and Goldberg (2000) presented traditional competitive intensity indices calculated for each species in the experimental community. In a CDS, the intensity and importance of competition is simply the slope and the coefficient of determination (R2 ), respectively, of the regression of yield on den sity (Chapter 3). If these data are presented in previously published papers, we can examine the intensity and im portance of competition in the community, however, the regression model used must be consistent throughout the comparison (Chapter 3). For example, we can not compare either the intensity or importance of competition on various community parameters in Lortie and Turkington ( 2002, see Table 2) because the regression model changes from year to year. In Goldberg et al. (2001), the community responses are all analyzed as a log-log ( or power) relationships and corn parisons can be made. For example, both the intensity and importance of competition on the emergence index increased from one year to the next for the desert community (Goldberg et al., 2001, Table 3). Similar analysis could be completed comparing differing communities to determine if intensity or importance changes between communities or it could be done comparing differing prod uctivities, life stages or life forms 2 . Competition began to affect the structure of this experimental community at density levels much lower (x1/8 density) than the natural density and this was apparent in all years of the study. The density at which the com munity reached constant final yield occurred at the x1 density or natural density in the final two years of the study. In the first two years, constant final yield was not reached likely because insufficient tim e had elapsed since the densities were manipulated for the mean plant mass to show the effect of competition. Unfortunately, none o f the previous CDS experiments have presented information in the form of figures or data describing when competiti on begins to impact the community or at what point constant final yield is reached. The advantage of being able to exami ne both species-specific and communitylevel responses with the CDS has also been underutilized. Other than Chapter 4, Zamfir and Goldberg (2000) are the only researchers who have pr esented both species-specific responses and community-level responses using this technique, though many others have reported species diversity changes within a CDS (Goldberg and Estabrook 1998; See Chapter 3 for further discussion on the choice of model used in the analysis of CDS. 2  98  Rajaniemi and Goldberg 2000; Lortie and Turkington 2002). In order for competition to significantly affect community structure, and therefore diversity, it must affect species differently (Rajaniem i and Goldberg 2000). In this experiment, the community as a whole was negatively affected by increasing den sity with most species showing a decrease in mean plant mass, although two species, Epilobium and Senecio, were not affected. Similarly, the intensity and importance of competition differed between species. While the effect of competition on the community began at low densities (x1/8), only Festuca and Linnaea also demonstrated the effect of competition at x1/8 density while Achillea, Arctostaphylos and Lupinus did not respond until x1/2. The community reached constant final yield at x1, the natural field density, but only Achillea and Festuca reached constant final yield (also at x1). No changes in species richness occurred in this experiment, but there were changes in evenness which increased with increasing density. Because evenness expresses how equally abundant species a re in a sample (Magurran 1988), the lowest density plots, or null community without competitive interactions, had some speci es become much more abundant relative to others. In other words as density and competition increased, they were more affected than other species. Evenness was also affected by fertilizer rates with the higher evenness in the fertilized plots. This result is opposite to what has beenn observed in other research done in this plant community (Turkington et al. 2002) and is contrary to the usual observations (DiTomamaso and Aarssen 1989; Gough et al. 20 00; Rajaniemi 2003). In this boreal understory experiment, the addition of fertilizer had a negative effect on the mean plant mass in both the CDS and the unmanipulated controls. Although this system is generally considered to be nutrient limited, these results are not contradictory to other long-term studies done in this system which have observed both positive and negative effects of increased fertility (Turkington et al. 1998; Turkington et al. 2002). Some short-term studies have reported decreases in survival with increased fertilizer (Dlott and Turkington 2000, Chapter 2; Arii and Turkington 2002) while others indicate either no effect of fertilization (Graham and Turkington 2000, Chapter 2 and 4) or positive effects (Arii and Turkington 20 02). The species-specific responses to fertilizer addition here correspond well with responses observed by Turkington et al. (2002). Species that decreased with increased fertilizer in both studies include Arctostaphylos and Linnaea, which are both low-growing prostrate shrubs, while Epilobium and Mertensia, taller erect herbaceous species, increased in both studies.  99  The biggest difference here is the lack of response of some species that usually increase with added fertilizer such as Festuca and Achillea (Turkington et al. 1998; Arii and Turkington 2002; Turkington et a I. 2002). This lack of a positive response, especially for the graminoid, Festuca, may in part be due to an unusually high abundance of microtine rodents in 2002. There is evidence to suggest that these voles may be specifically attracted to the fertilizer added to our experimental plots (Appendix 2). Fertilizer level was positively related to the number of over-winter vole nests foun d in the experimental CDS and unmanipulated plots (Appendix 2). The voles, especially the  Microtus spp., are known to feed on gram inoid seeds and may be partially responsible for preventing the Festuca from becoming more abundant with increased fertilizer as previously observed in this system (Turkington et al. 1998; Arii and Turkington 2002; Turkington et al. 2002). The higher abundance of voles in the fertilizer plots may have also reduced other species' abundance by the direct damage caused by the creation of dens and paths under the snow by the voles. There was no evidence to support the idea that productivity or fertilizer addition affects the role of competition in structuring this community at either the community or species level. We would expect to observe an interaction between density and fertilizer addition if this were the case. Two major theories have developed on how com petition intensity changes with productivity. Grime's theory predicts that competition intensity increases along a gradient of productivity (Grime 1977; Grime 1979) whereas Tilman's model predicts that competition intensity remains the same, although there is a switch from predominately belowground to aboveground competition as productivity increases (Tilman 1982; Tilman 1988). Though I have not attempted to differentiate between below and abovegro and competition, there is no evidence to suggest that the intensity of competition changes with productivity. It should be noted that even if most of the competition that occurred in this system was belowground, it would not change the shape of the curves observed in the CDS yield to density relationship. Plants would still be regulated by competition and follow the same patterns in the CDS — however, our ability to detect competition m ay be somewhat reduced if we only measure aboveground responses. Only two other studies that have used the CDS also manipulated productivity. Lortie and Turkington (2002) manipulated productivity by adding water although they do not present any interaction terms in their AN OVAs. Therefore, it is unclear as to whether productivity affected density depende nce. Goldberg et al. (2001) also manipulated productivity with irrigation treatments and observed interactions  100  indicating that competition was most intense at the lowest level of productivity. T his result has also been reported in a meta-analysis of competition experiments (Goldberg et al. 1999). Unlike Chapters 2 and 4, this experiment offers no evidence to support t he idea that facilitative interactions may be important in this community and thereby providing no support for the idea that positive interactions become more important in stressful environments (Bertness and Callaway 1994; Callaway and W alker 1997; Brooker and Callaghan 1998; Callaway et al. 2002). However, this is not necessarily correct. Both Chapters 2 and 4 dealt with the interactions of individual plants within communities composed of juvenile plants grown from seed that were one or two years old. In this study, extant adult plants were transplanted into manipulated communities. Numerous studies have demonstrated that the strength of competitive interactions change as individuals age (Howard and Goldberg 200 1; Lamb and Cahill 2006) and that the sign of the interaction may change depending on the life stage of the individuals (Callaway and Walker 1997; Goldberg et al. 2001; Zanini et al. 2006; Chapter 4). Though no facilitative interactions were observed with the adu It plants in this study, a CDS using the same species grown from seed indicated that increased planting density resulted in increased survival (Chapter 4). Chap ter 2 also demonstrated facilitative effects with neighbours increasing both survival and mean biomass of transplants grown from seed. If positive interactions are important in this community it is likely that they are most important early in the establishment of new individuals. I have demonstrated that density dependence is important in structuring this boreal understory community utilizing the relatively new experimental technique called the community density series. This CDS approach allows us to quantify both the intensity and importance of plant competition at the community and species levels and to determine whether the importance of these biotic interactions depend on abiotic factors. While fertilizer addition did have minor effects on the comm unity, it did not change the intensity of competition. The results presented here clearly show that species-specific responses to biotic interactions are not necessarily the same as community level ones and if we are to understand what occurs at the corn munity level, it is necessary to use appropriate methodological approaches such a s the CDS.  101  REFERENCES Aarssen, L.W. and Epp, G.A., 1990. Neighbour manipulati ons in natural vegetation: a review. Journal of Vegetation Science, 1: 13-30. Abrams, P.A., 1995. Monotonic or unimodal diversity-productivity gradients: what does competition theory predict? Ecology, 76(7): 2019-2027. Aerts, R., Berendse, F., de Cal uwe, H. and Schmitz, M., 1990. Competition in heathland along an experimental gradient of nutrient availability. Oikos, 57: 310-318. Arii, K. and Turkington, R., 2002. Do nutrient availability and competition limit plant growth of herbaceous species in the boreal forest understory? Arctic, Antarctic, and Alpine Research, 34: 251-261. Bertness, M.D. and Callaway, R., 1994. Positive interactions in communities. Trends in Ecology and Evolution, 9(5): 191-193. Brooker, R., Kikvidze, Z., Pugnaire, F.I., Callaway, R.M., Choler, P., Lortie, C.J. and Michalet, R., 2005. The importance of importance. Oikos, 109: 63-70. Brooker, R.W. and Callaghan, T.V., 1998. The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oi kos, 81(1): 196-207. Callaway, R.M., 1995. Positive interactions among plants. Botanical Review, 61(4): 306349. Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., M ichalet, R., Paolini, L., Pugnaire, F.I., Newingham, B., Aschehoug, E.T., Armas, C., Kikodze, D. and Cook, B.J., 2002. Positive interactions among alpine plants incre ase with stress. Nature, 417: 844-848. Callaway, R.M. and Walker, L.R., 1997. Corn petition and facilitation: a synthetic approach to interactions in plant communities. Ecology, 78(7): 1958-1965. Carlsson, B.A. and Callaghan, T.V., 1991. Positive plant interactions in tundra vegetation and the importance of shelter. Journal of Ecology, 79: 973-983. Carrier, P., 2003. Effects of water addition on biotic and abiotic corn ponents of a dry boreal forest in the Yukon. Ph.D. Thesis, University of British Columbia, Vancouver, BC. Carrier, P. and Krebs, C.J., 2002. T rophic effects of rainfall on Clethrionomys rutilus voles: an experimental test in a xeric boreal for est in the Yukon Territory. Canadian Journal of Zoology, 80: 821-829. Cody, W.J., 1996. Flora of the Yukon. NRC Research Press, Ottawa, Ontario, Canada. Connell, J.H., 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. American Naturalist, 122(5): 661696. Darwin, C., 1859. On the origin of species by means of natural selection. John Murray, London. DiTomamaso, A. and Aarssen, L.W., 1989. Resource manipulations in natural vegetation: a review. Vegetatio, 84: 9-29. Dlott, F. and Turkington, R., 2000. Regu lation of boreal forest understory vegetation: the roles of resources and herbivores. Plant Ecology, 151: 239-251. Flores, J. and Jurado, E., 2003. Are nurse-protégé interactions more corn mon among plants from arid environments? Journal of Vegetation Science, 14: 911-916. Fowler, N., 1986. The role of corn petition in plant communities in arid and semiarid regions. Annual Review of Ecology and Systematics, 17: 89-110. Goldberg, D. and Novoplansky, A., 1997. On the relative importance of competition in unproductive environments. Journal of Ecology, 85: 409-418. Goldberg, D.E., 1994. Influence of competition at the community level: an experimental version of the null models approach. Ecology, 75(5): 1503-1506.  102  Goldberg, D.E. and Barton, A.M., 1992. Patterns and consequences of interspecific competition in natural communities: a review of field experiments with plants. American Naturalist, 139(4): 771-801. Goldberg, D.E. and E stabrook, G.F., 1998. Separating the effects of number of individuals sampled and competition on species diversity: an experimental and analytic approach. Journal of Ecology, 86: 983-988. Goldberg, D.E., Rajaniemi, T., Gurevitch, J. and Stewart-Oaten, A., 1999. Em pirical approaches to quantifying interaction intensity: competition and facilitation along productivity gradients. Ecology, 80(4): 1118-1131. Goldberg, D.E., Turkington, R. and Olsvig-Whittaker, L., 1995. Quantifying the community-level consequences of competition. Folia Geobotanica Phytotaxonomica, 30: 231-242. Goldberg, D.E., Turkington, R., Olsvig-Whittaker, L. and Dyer, A.R., 2001. Densitydependence in an annual plant community: variation among life history stages. Ecological Monographs, 71: 423-446. Gough, L., Shaver, G.R., Carroll, J., Royer, D.L. and Laund re, J.A., 2000. Vascular plant species richness in Alaskan arctic tundra: the importance of soil pH. Journal of Ecology, 88: 54-67. Graham, S.A. and Turkington, R., 2000. Population dynamics response of Lupinus arcticus to fertilization, clipping, and neighbour removal in the understory of the boreal forest. Canadian Journal of Botany, 78: 753-7 58. Grime, J.P., 1977. Evidence for the existence of three primary strategies in plants an d its relevance to ecologica I and evolutionary theory. American Naturalist, 111: 11 691194. Grime, J.P., 1979. Plant strategies and vegetation processes. John Wiley, Chichester, UK. Grubb, P.J., 1985. Plant populations and vegetation in relation to habitat, disturbance and competition: problems of generalization. In: J. White (Editor), The population structure of vegetation. Dr W. Junk Publishers, Dordrecht, pp. 595-621. Gurevitch, J., 1986. Com petition and the local distribution of the grass Stipa neomexicana. Ecology, 67: 46-57. Harper, J.L., 1977. Population biology of plants. Chapman & Hall, London, UK. Howard, T.G. and Goldberg, D.E., 2001. Competitive response hierarchies for germination, growth, and survival and their influence on abundance. Ecology, 82: 979-990. Huston, M., 1979. A general hypothesis of species diversity. American Naturalist, 113(1): 81-101. John, E. and Turkington, R., 1995. Herbaceous vegetation in the un derstorey of the boreal forest: does nutrient supply or snowshoe hare herbivory regulate species composition and abundance? Journal of Ecology, 83: 581-590. John, E. and Turkington, R., 1997. A 5-year study of the effects of nutrient availability and herbivory on two boreal herbs. Journal of Ecology, 85: 419-430. Kadmon, R., 1995. Plant competition along soil moisture gradients: a field experiment with the desert annual Stipa capensis. Journal of Ecology, 83: 253-262. Kareiva, P., 1994. Higher order interactions as a foil to reductionist ecology. Ecology, 75(6): 1527-1528. Keddy, P.A., 1989. Competition. Chapman and Hall, New York. Keddy, P.A., 1990. Competitive hierarchies and centrifugal organization in plant communites. In: J. Grace and D. T ilman (Editors), Perspectives on plant competition. Academic Press, San Diego, pp. 265-290.  103  Kira, T., Ogawa, H. and Sakazaki, N., 1953. Intraspecific competition among higher plants. I. Mean plant weight-den sity interrelationship in reg ularly dispersed populations. Journal of the Institute of Polytechnics, Osaka City University, Series D, 4: 1-16. Krebs, C.J., Boutin, S. and Boonstra, R., 2001. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York. Lamb, E.G. and Cahill, J.F., 2006. Consequences of differing competitive abilities between juvenile and adult plants. Oikos, 112: 502- 512. Lortie, C.J. and Callaway, R.M., 2006. Re-analysis of meta-analysis: support for the stress-gradient hypothesis. Journal of Ecology, 94: 7-16. Lortie, C.J. and Turkington, R., 2002. The effect of initial seed density on the structure of a desert annual plant comm unity. Journal of Ecology, 90: 435-445. Magurran, A.E., 1988. Ecological diversity and its measurement. Princeton University Press, Princeton, New Jersey. Maillette, L., 1988. Apparent corn mensialism among three Vaccinium species on a climatic gradient. Journal of Ecology, 76: 877-888. McLellan, A.J., Law, R. and Fitter, A.H., 1997. Response of calcareous grassland plant species to diffuse competition: results from a removal experiment. Morris, W.F. and Wood, D.M., 1989. The role of lupine in succession on Mount St. Helens: facilitation or inhibition. Ecology, 70(3): 697-703. Newman, E.I., 1973. Competition and diversity in herbaceous vegetation. Nature, 244: 310. Rajaniemi, T.K., 2003. Explaining productivity-diversity relationships in plants. Oikos, 101: 449-457. Rajaniemi, T.K. and Goldberg, D.E., 2000. Quantifying individual- and co mmunity-level effects of competition using experimentally determined null species pools. Journal of Vegetation Science, 11: 433-442. Sammul, M., Kull, K., Oksanen, L. and Veromann, P., 2000. Competition intensity and its importance: results of field experiments with Anthoxanthum odoratum. Oecologia, 125: 18-25. SAS, 1995. JM P. SAS Institute Inc., Cary, North Carolina. Schoener, T.W., 1983. Field experiments on interspecific competition. American Naturalist, 122(2): 240-285. Shevtsova, A., Ojala, A., Neuvonen, S., Vieno, M. and Haukioja, E., 1995. Growth and reproduction of dwarf shrubs in a subarctic plant community: annual variation and above-ground interactions with neighb ours. Journal of Ecology, 83: 263-275. Shilo-Volin, H., Novoplansky, A., Goldberg, D.E. and Turkington, R., 2005. Density regulation in annual plant corn munities under variable resource levels. Oikos, 108: 241-252. Smith, B. and Wilson, J.B., 1996. A consumer's guide to evenness indices. Oikos, 76: 70-82. Taylor, D.R., Aarssen, L.W. and Loehle, C., 1990. On the relationship between r/K selection and environmental carrying capacity: a new habitat template for plant life history strategies. Oikos, 58: 239-250. Tilman, D., 1982. Resource competition and community structure. Princeton University Press, Princeton. Tilman, D., 1988. Plant strategies and the dynamics and structure of plant communities. Princeton University Press, Princeton, NJ. Turkington, R., Goldberg, D.E., Olsvig-Whittaker, L. and Dyer, A.E., 2005. Effects of density on timing of emergence and its consequences for survival and growth in two communities of annual plants. Journal of Arid Environments, 61: 377-396.  104  Turkington, R., John, E., Krebs, C.J., Dale, M.R.T., Nams, V.O., Boonstra, R., Boutin, S., Sinclair, A.R.E. and Smith, J.N.M., 1998. The effects of NPK fertilization for nine years on boreal forest vegetation in northwestern Canada. Jour nal of Vegetation Science, 9(3): 333-346. Turkington, R., John, E., Watson, S. and Seccombe-Hett, P., 2002. The effects of fertilization and herbivory on the herbaceous vegetation o f the boreal forest in north-western Canada: a 10- year study. Journal of Ecology, 90: 325-337. Weigelt, A. and Jolliffe, P., 2003. Indices of plant competition. Journal of Ecology, 91: 707-720. Welden, C.W. and Slauson, W.L., 1986. The intensity of competition versus its importance: an overlooked distinction and some implications. Quarterly Review of Biology, 61(1): 23-44. Zamfir, M. and Goldberg, D.E., 2000. The effect of initial density on interactions between bryophytes at individual and community levels. Journal of Ecology, 88: 243-255. Zanini, L., Ganade, G. and Hiibel, I., 2006. Facilitation and competition influence succession in a subtropical old field. Plant Ecology, 185: 179-190. Zobel, M., 1992. Plant species coexistence: the role of historical, evolutionary and ecological factors. Oikos, 65: 314-320.  105  Chapter 6 CONCLUSIONS This thesis describes the first example of the CDS being used in a perennial plant community over multiple years. The CDS has proven to be an effective method for examining density dependence in annual plant communities (Goldberg et al. 2001; Lortie and Turkington 2002; Shilo-Volin et al. 2005; Turkington et al. 200 5), in experimental communities of bryophytes (Zamfir and Goldberg 2000 ), and in a single growing season of an old-field community (Rajaniem i and Goldberg 2000). Previous studies i n northern unproductive environments have detected both positive and negative plant interactions (Chapter 1, Chapter 2). However, due to a lack of suitable e xperimental methods, no other study has been able to demonstrate that d ensity dependence operates at both the species level and the entire community level (Chapter 5). Whether competition structures unproductive communities, such as northern plant communities, has been intensely debated, with some arguing that it is unimportant (Grime 1977; Grime 1979; Huston 1979; Keddy 1989; Keddy 1990) while others argue that it is important (Newman 1973; Tilman 1982; Grubb 1985; Tilman 1988; Taylor et al. 1990). There is also a growing body of theory sugge sting that facilitation is likely to be more important in these same unproductive systems (Bertness and Callaway 1994; Belcher et al. 1995; Brooker and Callaghan 1998 ). Existing evidence suggests that both competition and facilitation occur in unproductive systems (Chapter 1) and the results presented in this thesis confirm this evidence. In the ecosystem where this study took place, the plant communities are nutrient limited (Turkington et al. 1998; Turkington et al. 2002) and competition has been demonstrated for a single species (Achillea  millefolium) but not for others (Arii and Turkington 2002). However, these results do not tell us whether competition is important in structuring this community. The CDS experiment presented in Chapter 5 demonstrates that whil e the effect of fertilizer addition is not as pronou nced as in other studies, the effect of competition on the plant community is both intense and important and, the very important conclusion, that species-level responses are not identical to corn munity-level ones. . That responses measured at the individual-level do not equate with corn munitylevel ones makes it difficult to draw conclusions about community structure from competition experiments with either a single or few species included such as the majority  106  of studies described in Chapter 1 (Table 1). Competition and facilitation will only affect aspects of community structure, such as biodiversity, if species respond differently to biotic interactions (Rajaniemi and Goldberg 2000). In Chapter 2, the presence of neighbours increased survival and biomass of six species indicating a facilitative effect of neighbouring plants. Water increased the bio mass of three species and fertilizer increased the mass of one species and decreased the mass of another. In the sandbox experiment (Chapter 4), species-specific responses varied between the density, water and fertilizer treatments and were dependent on life stage with some positive, others negative, with almost half showing no effect. Species-specific responses to density in the multi-year CDS were much more consistent than the previous study with seven species negatively related to density. No facilitative effects were o bserved. The effect of fertilizer was negative at the community level and inconsistent at the species level. If only a few species were examined in a traditional removal experiment, it would be very difficult to come to any conclusion regarding the role of biotic interactions in the boreal understory community. Both the theories of Grime and Tilman predict that competitive exclusion is important at high levels of productivity so the major controversy between them concerns the role of competition at low productivity levels (Abrams 1995). The debate surrounding some of the conflicting predictions of these two theories has been one of the most enduring and controversial in plant ecology (Goldberg and Novoplansky 1997; Craine 2005). The results of this thesis support the idea that competition regulates this plant community which has been shown to be nutrient limited (Turkington et al. 1998; Turkington et al. 2002). Given this, in the Chapter 5 CDS we would expect that competition would be most intense and important at the lowest nutrient level; however, increasing fertilizer did not affect mean plant mass except in the final year of the study and the results were counterintuitive (fertilizer negatively affected plant mass). It might be argued that other factors become limiting (i.e. light) as nutrients become more abundant (Tilman 1982, 1988). This was not tested, although it seems unlikely that light became limiting in the sandbox experiment of Chapter 4, where final mean mass did not have any response to fertilizer (or water) addition yet did demonstrate intense and important density regulation. After one growing season plants were still so short (all plants were < 10 cm and most were < 5 cm) it is unlikely that light limitation is a feasible hypothesis. Similarly, the idea that facilitation becomes more important as productivity  107  decreases (Turkington et al. 1998; Turkington et al. 2002) was not supported by either Chapter 4 or 5, yet facilitation was observed in both Chapters 2 and 4. The results presented in this thesis offer surprisingly similar conclusions to the CDS of Goldberg et al. (2001) even though their study system was an annual plant community in the Negev desert in Israel compared to the perennial understory community in the boreal forest in the Yukon. Both studies found that facilitation (positive density dependence) was important in the survival of plants grown from seed, that negative density depend ence was important initially at the germination stage, and that once established, corn petition restricted growth of the plants. Neither th is thesis nor Goldberg et al. (2001) fully support Grime's or Tilman's theories regarding the role of competition in unproductive environments. Both studies, which were d one in unproductive environments, found that competition was both intense and important and regulated the plant community in terms of mean plant mass, yet when resources were increased, there was no clear relationship with corn petition intensity (or im portance) and resource availability. These results tend to support the ideas of Tilman (1982, 1988). The facilitative response of plant survival to increased density demonstra ted in Goldberg et al. (2001) and in Chapter 4 is not completely consistent with the idea that facilitation is more likely under stressful conditions systems (Bertness and Callaway 1994; Belcher et al. 1995; Brooker and Callaghan 1998) because there were no consistent changes in the density effect on survival with increased resource availability. Whether all of the differing responses to increased density at different life stages are characteristic of plant communities in more productive regions has not yet been tested and should beco me a focus of current research. Most studies are focused on single species and few exam ine more than one stage (Goldberg et al. 2001; Howard and Goldberg 2001). In Chapter 4, and in Goldberg et al. (2001), all life stages examined showed density dependence. While survival and final plant mass are usually included, more studies also need to include emergence. For example, in another CDS (Lortie and Turkington 2002), the most consistent density effect was a de crease in the emergence with increasing density. The negative effects of increased seed density are related, or often assumed to be related, to chemical inhibition (Murray 1998; Dyer et al. 2000; Goldberg et al. 2001; Lortie and Turkington 2002). If this is a common observation, it is crucial to measure both initial effects of density and the subseque nt effects of density in order to understand plant population dynamics and the structure of the resultant community.  108  Though the CDS has many benefits over traditional corn petition studies (Chapters 3, 4 and 5), there are some drawbacks, especially when done in a perennial system over many years. However, the advantage of being able to characterize the species- and community-level effects of biotic interactions overwhelms any of these disadvantages. The initial setup of the CDS is reasonably easy when started by removing all vegetation, letting the community restart from seed, and then "weeding" to a lower density to characterize the null community as was done by Rajaniemi and Goldberg (2000). This is true for both perennial and annual plant comm unities. Annual communities have the advantage of being easy to increase a nd decrease the plant density by seed addition and plant removal. In Chapter 4, the entire experimental plant community was begun by seed, and although it was comprised of perennial species, it was only allowed to grow for one season . The CDS in Chapter 5 was constructed by adding and removing adult plants. This was a tremendous undertaking, and took the entire first, and part of the second, season to establish. A II of the plant species in the CDS are clonal and it was very time consuming to remove or add only a few "individuals" when manipulating the density. The establishment of this sort of CDS study can only be considered for experiments planned to run for multiple years. Much research effort has been devoted to studying the effects of neighbours, but to date, few studies have examined the effect at the community level. With the CDS, it is now possible to examine the role of biotic interactions at the species- and communitylevel and to determine what life stages are most important for determining population dynamics and community structure.  109  REFERENCES Abrams, P.A., 1995. Monotonic or unimodal diversity-productivity gradients: what does competition theory predict? Ecology, 76(7): 2019-2027. Arii, K. and Turkington, R., 2002. Do nutrient availability and competition limit plant growth of herbaceous species in the boreal forest understory? Arctic, Antarctic, and Alpine Research, 34: 251-261. Belcher, J.W., Keddy, P.A. and Twolan-Strutt, L., 1995. Root and shoot competition intensity along a soil depth gradient. Journal of Ecology, 83: 673-682. Bertness, M.D. and Callaway, R., 1994. Positive interactions in communities. Trends in Ecology and Evolution, 9(5): 191-193. Brooker, R.W. and Call aghan, T.V., 1998. The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oi kos, 81(1): 196-207. Craine, J.M., 2005. Reconciling plant strategy theories of Grime and Tilman. Journal of Ecology, 93: 1041-1052. Dyer, A.E., Fenech, A. and Rice, K.J., 2000. Accelerated seedling emergence in interspecific competitive neighbourhoods. Ecology Letters, 3: 523-529. Goldberg, D. and Novoplansky, A., 1997. On the relative importance of competition in unproductive environments. Journal of Ecology, 85: 409-418. Goldberg, D.E., Turkington, R., Olsvig-Whittaker, L. and Dyer, A.R., 2001. Densitydependence in an annual plant community: variation among life history stages. Ecological Monographs, 71: 423-446. Grime, J.P., 1977. Evidence for the existence of three primary strategies in plants an d its relevance to ecologica I and evolutionary theory. American Naturalist, 111: 11 691194. Grime, J.P., 1979. Plant strategies and vegetation processes. John Wiley, Chichester, UK. Grubb, P.J., 1985. Plant populations and vegetation in relation to habitat, disturbance and competition: problems of generalization. In: J. White (Editor), The population structure of vegetation. Dr W. Junk Publishers, Dordrecht, pp. 595-621. Howard, T.G. and Goldberg, D.E., 2001. Competitive response hierarchies for germination, growth, and survival and their influence on abundance. Ecology, 82: 979-990. Huston, M., 1979. A general hypothesis of species diversity. American Naturalist, 113(1): 81-101. Keddy, P.A., 1989. Competition. Chapman and Hall, New York. Keddy, P.A., 1990. Competitive hierarchies and centrifugal organization in plant communities. In: J. Grace and D. Tilman (Editors), Perspectives on plant competition. Academic Press, San Diego, pp. 265-290. Lortie, C.J. and Turkington, R., 2002. The effect of initial seed density on the structure of a desert annual plant community. Journal of Ecology, 90: 435-445. Murray, B.R., 1998. Density-dependent germination and the role of seed leachate. Australian Journal of Ecology, 23: 411-4 18. Newman, E.I., 1973. Competition and diversity in herbaceous vegetation. Nature, 244: 310. Rajaniemi, T.K. and Goldberg, D.E., 2000. Quantifying individual- and co mmunity-level effects of competition using experim entally determined null species pools. Journal of Vegetation Science, 11: 433-442.  110  Shilo-Volin, H., Novoplansky, A., Goldberg, D.E. and Turkington, R., 2005. Density regulation in annual plant corn munities under variable resource levels. Oikos, 108: 241-252. Taylor, D.R., Aarssen, L.W. and Loehle, C., 1990. On the relationship between r/K selection and environmental carrying capacity: a new habitat template for plant life history strategies. Oikos, 58: 239-250. Tilman, D., 1982. Resource competition and community structure. Princeton University Press, Princeton. Tilman, D., 1988. Plant strategies and the dynamics and structure of plant communities. Princeton University Press, Princeton, NJ. Turkington, R., Goldberg, D.E., Olsvig-Whittaker, L. and Dyer, A.E., 2005. Effects of density on timing of emergence and its consequences for survival and growth in two communities of annual plants. Journal of Arid Environments, 61: 377-396. Turkington, R., John, E., Krebs, C.J., Dale, M.R.T., Nams, V.O., Boonstra, R., Boutin, S., Sinclair, A.R.E. and Smith, J.N.M., 1998. The effects of NPK fertilization for nine years on boreal forest vegetation in northwestern Canada. Jour nal of Vegetation Science, 9(3): 333-346. Turkington, R., John, E., Watson, S. and Seccombe-Hett, P., 2002. The effects of fertilization and herbivory on the herbaceous vegetation o f the boreal forest in north-western Canada: a 10-year study. Journal of Ecology, 90: 325-337. Zamfir, M. and Goldberg, D.E., 2000. The effect of initial density on interactions betwe en bryophytes at individual and community levels. Journal of Ecology, 88: 243-255.  111  Appendix 1 HOW TO GROW AND KILL THE NATIVE PLANTS OF KLUANE There are few resources available on how to grow the native plants from the Kluane region. One exception is "Growing Alaskan Natives" (Baldwin 1997). Also, traditional gardening books can he 1p with general plant propagation principles and for specifics on germinating seeds you can consult specialized texts (Baski n and Baskin 1998). Here I present some observations from my work while trying to grow and kill plants in the Yukon.  Growing from seed  The germination rates of seeds co Ilected from the Kluane region are q uite variable both in terms of interspecies comparisons and interannual variability within a single species (Table A 1.1). The interannual variability is likely due to variation in seed quality due to climatic conditions.  Table A 1.1. The mean (± 1 SE) percent germination of the most common understory species at the research site in Kluane described in chapters 2-4. These data are for seeds collected from 1999 to 2002. Seeds were sown onto wet sand in Petri plates with 50 seeds per Petri plate (n = 3 plates). Species Achillea millefolium var. borealis Anenome parviflora Epilobium angustifolium Festuca altaica Lupinus arcticus Mertensia paniculata Senecio lugens Solidago multiradiata  Mean % germination  63.2 ± 16.9 21.6 ± 8.6 23.6 ± 11.3 40.2 ± 17.3 18.7 ± 4.3 7.2 ± 2.8 69.8 ± 12.7 65.1 ± 5.0  As long as you wait for natural disper sal of the seeds (and therefore seed maturity), many of the species (Achillea, Senecio, and Solidago) need no special treatment to have reasonably high germination success. Other species have high germination in some years and low in others (for example Lupinus, Epilobium and  Festuca). Lupinus germination was quite varia ble from year to year. One of the problems of trying to grow lupines from seed is collecting enough seed (since it has a low  112  germination rate) before the seed eating larvae get to them. These might be seed beetles (Family Bruchidae) which are known to e at peas and other legumes. These larvae eat an amazing amount of the seed. One trick to get lupine seeds to germinate is to place them in warm water for 24 hours before planting. I don't think that this increases the germination rate, but it speeds up the emergence of the radical, probably because the warm water has softened the hard seed coat. As soon as lup ine seeds geminate, and radicle emerges, the young seedling need s to be removed from the Petri plate before the onset of damping off (due to Pythium and other fungi). Spraying the seedling with a product like Plant Prod® NoDamp seems to help a little. Mertensia also benefits from some extra special treatment. Like Lupinus, Mertensia has a very tough seed coat that softens when soaked in warm water. It helps to remove the seed coat entirely with a pair of forceps and a scalpel, although th is is very time consuming and still only increases the percent germination from 4.4% (SE=5.1) to 13.3% (SE=8.8). There were other species which I tried to germinate (Table A 1.2) that were found nearby the research site. Many of these species had quite high germination rates, however, seeds were collected only in 1999 and therefore these results may not be representative given the high interannual variation of some of the other species (Table A 1.1).  Table A 1.2. The mean (± 1 SE) percent germination of some of the less common species collected near the research site in 1999. Seeds were sown onto wet sand in Petri plates with 50 seeds per Petri plate (n = 3 plates). Species Agrostis scabra Anenome multifida Elymus spp. (likely trachycaulus) Epilobium latifolium Hedysarium douglasii Hordeum jubatum Linum lewisii Poa spp. (likely arctica) Polemonium acutiflorum Trisetum spicatum  Mean % germination  54.0 63.3 78.0 36.0 20.0 82.6 65.0 30.0 28.9 83.3  ± 4.2 ± 3.5 ± 7.2 ± 6.1 ± 1.9 ± 2.7 ± 14.4 ± 4.2 ± 11.3 ± 2.7  113  General seed sowing tricks: 1. Wait for the seeds to reach maturity before beginning to collect them. This will ensure the highest possible germination rate. For some species this might not be possible. For instance, Achillea seeds do not mature until late Septem ber. Most researchers would have left for the south by this time. Therefore, I collected seeds the following spring from intact plants surviving from the previous year. 2. Dry the seeds before they go into storage. Note that some species will not germinate if they are stored. For example, Salix seeds remain viable for only very short periods of time. 3. Store the seeds under "natural conditions". This might normally be a frozen wet substrate — but I had reasonable success with just keeping the air dried seeds in plastic bags in a freezer from November until May. 4. The light regime may be important for germination success (Densmore 1997). The germinating success in Vancouver (in May) was less than it was at Kluane (in June). I always germinated seeds in the Yukon in early June. Normally seedlings are seen in the field in late June to early July so this seemed like a reasonable time to begin sowing. I started the seeds in Petri plates on be ach sand and I have transplanted them to peat plugs or seedling trays once both the shoot and root emerged. I have also germ inated them directly in the peat plugs. 5. Sufficient water is very important to germinate many species (Oberbauer and Miller 1982). Ensure that the sand (or whatever substrate you choose) is saturated. However, once the seedling is esta blished you can have problems with too much moisture; damping off fungus can be a problem especially with Lupinus and Mertensia.  Growing from cuttings Woody species such as Linnaea borealis and Arctostaphylos uva-ursi are reasonably easy to start from cuttings with survival after 2 weeks ranging from approximately 50 for Linnaea to nearly 90% for Arctostaphylos. I usually collected fresh growing tips approximately 5 cm long and removed the leaves from half of the length closest to the cut. The cutting was then dipped in a corn mercially available rooting  114  compound and placed in moist sand and kept in a moist environment. I preferred to use Wilson Roots® Liquid Root Stimulator (a rooting stimulant gel which also contains a fungicide) rather than the StimRoot powder. Sometimes, before I dipped the cutting in the rooting compound, I used a clean sharp razor blade to slice the bark of the cutting 46 times in the region where I removed the leaves. Some non-woody species can als o be propagated from fresh cutti ngs. Table A 1.3 lists the rooting success of fresh cuttings of some common non-woody under story species. Though rooting success ranged from as high as 22.6 % to none, all species had some individual cuttings that remained green, though they did not develop new rooting tissue (ranging from 12% to 96%).  Table A 1.3. The mean (± 1 SE) percentage rooting of fresh cuttings after 30 days. All cuttings were from new leaves. The cut end of the leaf was dipped into commercially available rooting hormone and cuttings were placed in planting trays filled with moist sand. The tray was covered with plastic and the sand kept moist. This experiment ran from July 8 to Aug 7, 2002.  Species  Mean % survival  Achillea borealis  22.0 ± 2.0  Anenome parviflora  0.0  Epilobium angustifolium  17.3 ± 3.6  Lupinus arcticus  0.0  Mertensia paniculata  3.0 ± 3.0  Senecio lugens  15.2 ± 1.5  Solidago multiradiata  22.6 ± 0.5  Transplanting Transplanting whole plants can be successful for all of the species I used in my experiments. The problem with transplanting, and the key for success, is that a fairly large clump of soil must be moved with the transplant to avoid disturbing the plant's roots. The most problematic species is Lupinus arcticus. A clump of dirt surrounding a lupine transplant with a diameter equal to 2x (or more) the height of the plant is necessary to ensure any chance of the transplant surviving. The clump also has to be as deep as it is wide. Many times during the transplanting of lupines, we noticed a large and broken tap root coming straight out from the bottom of our transplant sod. Within hours the entire transplant begins to wilt. The next most important key for success is to  115  water transplants. During the first couple of weeks of establishing my large experiment with transplants (this covered a total of 63 m 2 ), I trucked approximately 150 L a day to the site. Both Arctostaphylos uva-ursi and Linnaea borealis were successfully transplanted as large sods in my large experiment. Many species such as Achillea borealis, Mertensia paniculata, Epilobium angustifolium, Senecio lugens and Solidago multiradiata can be transplanted as sm all plugs obtained with 1.5" soil cores as long as they have sufficient root cores, though they will be happier being transferred in larger clumps of soil. Obviously you can't transplant large individuals this way. Samantha Hicks (Hicks and Turkington 2000) had very high success (>80%) transp lanting Festuca, Achillea and Mertensia. All were transplanted as small soil cores (4.5 cm in diameter and 7 cm deep), but all were covered with plastic "tents" (to keep humidity high) for a week and watered twice a day.  Killing with Glyphosate I used Glyphosate (RoundUpTM), one of the more common herbicides, to remove plants in Chapter 2. This is a non-selective and systemic herbicide that is quickly broken down in the soil. It was quite effective at killing most plants, although some were very resistant (Table A 1.4). It can be applied with a pump sprayer (Roy has one) or you can apply it with a paint bush if you are trying to be really selective about which plants to kill. See the note below on cl onal plants. You can't kill an Achillea with RoundUp that is anywhere close to an Achillea that you want to keep. By close I mean within a meter or perhaps more. This applies to most of the species in the and erstory; they are almost all clonal. Glyphosate has been criticized in the popular press as being detrimental to the environment. As previously mentioned, glyphosate stro ngly bonds to soil particles which limits its phytotoxicity in soil (WHO 1994; Ahrens 1994). It h as an average half-life in the soil of 32 days (Giesy et al. 2000) an d even if detected in the soil, it is bound to particles and is no longer able to kill plants. It is eventually broken down by soil microorganisms (WHO 1994). For summary information on glyphosate and its toxicity and health impact, see the fact sheet provided by the US EPA and Oregon State University at <npic.orst.edu/factsheets/glyphotech.pdf>.  116  Table A 1.4 The number of applications of a 1:20 (Glyphosate to water) concentration necessary to see significant die back in the named plant species. Applications were approximately 1 week apart and were applied with a pump sprayer. The leaves were soaked with the solution until dripping wet. The higher the number beside the species, the higher the resistance to the Glyphosate. Plant species^No. of applications Achillea borealis^2-3 Anenome parviflora^2-4 Arctostaphylos uva-ursi^4 Artemesia norvegica^>5 Betula spp.^ 1-2 Carex spp.^ 3-5 Danthonia intermedia^2-3 Delphinium glauca^3 Draba spp.^ 2-3 Epilobium angustifolium^2-3 Festuca altaica^2 Linnaea borealis^2-3 Lupinus arcticus^1 Picea glauca^ 3-5 Salix myrtilloides^5 Salix reticulata^3-5 Senecio lugens^3-4 Solidago multiradiata ^3-4  The problem of clonal plants The problem with transplanting and trying to kill plants in the Yukon is that most of them are clonal. Probably most of the new growth and horizontal spread of the understory vegetation is due to clona I growth; we see relatively few seedli ngs in the understory. Transplanting isn't such a big problem because individual shoots or ramets can be taken without any ill effects so long as the shoot has sufficient roots. The biggest problem is trying to selectively kill the plants or keep them out of plots. In my big community density series experiment (the one done in the field), I took a spade and severed all the roots around my plots so what was going on root-wise in the soil was restricted to the plots. I then removed all the ramets of those species that I didn't want in the plot. Regrowth was not a m ajor issue. Roy also has had his m inions spade around the perimeter of his 5x5m plots since 1990. When using RoundUp to kill unwanted vegetation, you not only need to be really careful about overspray landing on your desirable target plants, but you can't spray  117  plants that are close to ones you want to keep if they are the same species. RoundUp is a systemic herbicide — it goes into the plant via the stomata and will "travel" throughout the plant and kill all shoots that are connected underground. The entire genet will die. I had the summer students do a little experiment for me one summer where I had them repeatedly apply RoundUp with a paint brush to Achillea, Epilobium, Lupinus, Mertensia, Senecio and Solidago. They observed that for all species, individuals (obviously connected underground) died for up to 1 m from the sprayed target plants.  118  REFERENCES  Ahrens, W. H. (Ed.). 1994. Herbicide Handbook, 7 th Edition. Weed Science Society of America. Champaign, IL. Baldwin, R.L., 1997. Growing Alaskan Natives: the propagation of Alaska's native plants. Appears to be self published. Baskin, C.C. and Baskin, J.M., 1998. Seeds: ecology, biogeography and evolution of dormancy and germination. Academic Press, San Diego, CA. Densmore, R.V., 1997. Effect of day length on germination of seeds collected in Alaska. American Journa I of Botany, 84: 274-278. Giesy, J.P., Dobson, S., and Solomon, K. R. 2000. Ecotoxicological risk assessment for Roundup herbicide. Reviews of Environmental Contamination and Toxicology 167:35-120. Hicks, S. and Turkington, R., 2000. Corn pensatory growth of three he rbaceous perennial species: the effects of clipping and nutrient availability. Canadian Journal of Botany, 78: 759-767. Oberbauer, S. and Miller, P.C., 1982. Effect of water potential on seed germination. Holarctic Ecology, 5: 218-220. World Health Organization. 19 94. Environmental Health Criteria 159: Glyphosate. Geneva, Switzerland.  119  APPENDIX 2 Are Voles attracted to fertilizer? A cautionary tale' INTRODUCTION As field ecologists we strive to design well designed, properly replicated, controlled experiments that permit us to unambiguously test our hypotheses; sometimes we get it wrong. This research was prompted by another project that did n't adequately control all external factors, specifically the activities of herbivores. In the boreal forest, the structure of the understory vegetation is thought to be primarily governed by nutrient limitation (Bonan and Shugart 1989; Turkington et al. 1998) and competitive interactions, and less so by herbivores. Indeed, Turkington et al. (2002) demonstrated that the exclusion of herbivores had negligible impact on the community composition in a boreal forest understory. Therefore, in a rather complicated experiment (see Field CDS experiment — Chapter 5) I decided to simplify our experimental design and exclude the effects of herbivores and address only nutrients and plant co mpetition. I constructed a simple 1 m high fence made of 2.5 cm aperture chicken wire sim ilar to fences used to exclude herbivores in of her experiments. Although the fences were effective i n excluding the primary herbivore, the snowshoe hare (Lepus americanus Erxleben), from the 63 plots, nevertheless, in May 2002, a surprising amount of over- winter damage was observed in some of the plots. The fences were ineffective at excluding other herbivores in the system, particularly the microtine rodents, mice and voles.  ' A version of this chapter will be submitted for publication. Treberg, M. A., Edwards, K. and Turkington, R. (2008). Are voles attracted to fertilizer?  120  Figure A 2.1. An over-winter vole's nest in one of the plots in the main experiment of Chapter 5. The white dots are fertilizer pellets from the previous summer that have not yet dissolved. The nests are surrounded by long pieces of cut grass (Festuca altaica) and often contain many short (-2.5 cm) pieces of grass formed in small haystacks. These stacks are created by voles (primarily Microtus spp.) that cut the stems of Festuca in order to reach the seed heads (Forsyth 1999).  Through casual observation, it was apparent that there was higher winter activity and nesting of voles in the fertilized plots. In fact, many of the vole nests contained caches of undissolved fertilizer that had been applied the previous spring (Fig. A 2.1). Voles may prefer nesting in fertilized plots because they offer a higher quality and abundance of food, and these plots may also provide better cover and protection. A third option is that the voles are sim ply attracted to fertilizer and it is this hypothesis I test in this chapter.  121  METHODS  SITE DESCRIPTION The experiments were conducted near Kluane Lake in southwestern Yukon in the same area as previous experiments investigating the role of biotic and abiotic factors on plant community structure (Turkington et al. 1998; Turkington et al. 2002) and is within the area where the Kluane Boreal Forest Ecosystem Project (KBFEP) took place (Krebs et al. 2001). The research was conducted in open and closed stands of white spruce (Picea glauca) which also has some balsam poplar (Populus balsamifera) and aspen (Populus tremuloides) in the overstory. The shrub layer is do minated by willows (various Salix spp.) and dwarf birches (Betula nana and B. glandulosa). The understory is dominated by grass (mostly Festuca altaica), some low-growing woody species such as twinflower (Linnaea borealis) and bearberry (Arctostaphylos uva-ursi and A. rubra) and mixed herbaceous vegetation. The site is described in detail in Krebs et al. (2001) and Turkington et al. (2002). The most common microtines in the Kluane region a re the northern red-backed vole (Clethrionomys rutilus although C. gapperi may also be in this range), the deer mouse (Peromyscus maniculatus), and various Microtus spp., of which, the meadow vole, M. pennsylvanicus, is the most abundant (Boonstra et al. 2001). While every effort was made to identify all captured voles and mice to species, all are grouped together for the subsequent analyses. Of all the 210 captured voles and mice, approximately 58% were identified as Microtus spp., 39% Clethrionomys spp., 2% Peromyscus maniculatus, and 1% were unknown either becau se they escaped before identification or I couldn 't identify them to species. Hereafter, all species will be referred to as voles.  VOLE NESTS OBSERVED IN FERTILIZED CDS PLOTS In 1999 an experiment was designed to determine the role of density dependence and increased productivity on the and erstory plant community (Chapter 5). The experimental treatments used in this Community Density Series (CDS) were manipulations of density from 1/16X the naturally observed density up to 2X the normal plant density crossed with three levels of fertilizer ad dition. Plots were 1 m  2  squares and  were surrounded by 1 m high chicken wire fencing with a 2.5 cm mesh size. In June 2002, all over-winter vole nests were counted in each plot. These were easily identified as small depressions surrounded by cut pieces of grass. The number of nests per plot  122  was analyzed using ANOVA with density and fertilizer addition as treatments. The fertilizer used in this study, and all of the following experiments, was a standard N-P-K granular fertilizer (Circle H Farms, 21-7-7 Spring Lawn Fertilizer, Nu-Gro Corporation, Woodstock ON).  VOLES CAPTURED IN FERTILIZED PLOTS A 2 x 2 factorial experiment (± fertilizer and ± fences; n= 4) using 5 m x 5 m plots at 2 sites, Boutellier Summit and Microwave Road, has been running since 1990 (Turkington et al. 2002). In 2002 I trapped voles in these plots in m id-June and at the end of August. A Longworth mouse trap was placed at opposite ends of each plot (but more than 1 m from the edge) in wire cages (to prevent squirrels from disturbing the traps) with a piece of plywood placed on top to offer some shade to the trap. Prebaiting was done for 3 days before traps were set. Traps were baited with barley and a small slice of apple and were stuffed with a wad of cotton batting. Traps were set in the evening, checked and reset if necessary the next morning, checked and reset in the evening and checked the last morning and then removed. All microtines were identified to species and sex, were tagged and then released. The number of vole s and mice caught per treatment was analyzed using ANOVA. All species were combined and each trapping session (June o r August) was analyzed separately. All analyses were done using JMP 4 (SAS 1995).  VOLES CAPTURED IN TRAPS WITH FERTILIZER AS BAIT To test if voles were attracted to fertilizer, 2 sets of 24 Longworth traps were set out at each of three different locations during July 2002. Half the traps contained th e regular barley and apple bait while half had a small amount of fertilizer added (< 1.0 g) to the bait. The trapping protocol was similar to the experiment described above. Traps were set out approximately 10 m apart in a grid, alternating between traps with regular bait and traps with regular bait + fertilizer. Traps were set for 3 consecutive days. All microtines were identified to species and sex, tagged and released. The number of voles and mice caught per treatment was analyzed with AN OVA and all species were combined. Site was considered as a treatment effect.  123  VOLES ENTERING LOCKED-OPEN TRAPS TO REMOVE FERTILIZER In an additional test of whether voles are attracted to fertilizer, 10 Longworth traps containing only fertilizer, were set out for a 2 week period, placed 10 m apart along a transect. No grain or apple bait was add ed to the traps and the traps were lockedopen to allow the voles to enter and leave freely. Each trap contained 5 g of fertilizer and in the entrance tunnel to each trap, a small plastic plate covered with a thin smear of vegetable oil and dusted with talcum powder was placed to determine how many times voles entered and left the trap (Fig. A 2.2). Traps were checked twice daily for activity and categorized as "low" activity (trap entered 1 or 2 times), "medium" activity (entered 3 or 4 times), and "high" activity (entered more than 5 tim es). If fertilizer had been removed from a trap, the remaining fertilizer was weighed. On twelve occasions traps had fertilizer changed because fertilizer was lost during heavy rainfall events; all ten traps needed replacing after one heavy rainfall and 2 others needed replacing on a separate day. On three occasions traps ha d to be repaired and new fertilizer added due to damage caused by grizzly bears. Once, fertilizer needed to be replaced because ants removed pe Ilets. Ants were present in the trap during the morning check of the tra ps and there was a trail left in the talcum powder.  124  ^'It  11111  ^ 0 ^1^ 2^3^4^5^6  Figure A 2.2. Plastic plate covered with vegetable oil and dusted with talcum powder. These were placed in the entrance tunnels of the Longworth traps to determine if mice and voles had entered the traps to remove fertilizer.  RESULTS VOLE NESTS OBSERVED IN FERTILIZED PLOTS There was significantly more over-winter nests in the fertilized plots of the CDS experiment than the unfertilized (zero treatment) plots (Fig. A 2.3a, Table A 2.1). There was a trend towards increased nests with increased plant density but this was only marginally significant (ANOVA, P = 0.058; Fig. A 2.3b, Table A 2.1).  125  (a)  7  ,  6_  0 5 CL  H  4,  b  3 a)  z 2  -I  a  1 0  0  1  2  Fertilizer treatment  (b)  9 8 7  -5 6  43 5 0_ co 4 17') a) 3  z  2 1 0 x1/16^x1/8^x1/4^x1/2^x1^x2^control Plant density  Figure A 2.3. The mean number of nests per plot and 95% confidence interval as a function of (a) fertilizer treatment (averaged over all plant densities) and (b) plant density (averaged over the three fertilizer levels) in the CDS experiment. In (a), columns sharing the same letter are not significantly different (P<0.05, Student's t). The control shown in (b) are those plots that did not undergo density manipulation but did have the three fertilizer treatments.  Table A 2.1. Summary of ANOVA of the number of over-winter nests observed in the CDS experiment. Significant effects (P < 0.05) are shown in bold.  Source Fertilizer Density Fertilizer x Density Error  df 2 6 12 42  SS 147.841 77.492 41.937 242.667  F-ratio 12.794 2.235 0.605  P  <.001 0.058 0.826  126  VOLES CAPTURED IN FERTILIZED PLOTS During the first trapping session in June, more voles were captured in the fertilized plots than the unfertilized plots (Fig. A 2.4a, Table A 2.2a), but this effect was not detected in the August census (Table A 2.2b). Curiously, there was also a significant fence and site interaction effect during the first census (Fig A 2.4b, T able A 2.2a), that also was not detected in the August census (Table A 2.2b). As is usually the case, fewer mice and voles were ca ptured in the June census (mean = 2.7 voles per plot per trapping session) than in the August census (mean = 3.9 voles per plot pe r trapping session; t-Test, P = 0.002)  ^  Boutellier  im Microwave  0 ^ Unfertilized  Fertilized  Unfenced  Fenced  Figure A 2.4. The mean number of voles captured per plot and 95% confidence interval during the June trapping session in (a) the fertilized and unfertilized plots and the (b) fenced and unfenced plots at the Boutellier Summit and Microwave Road sites. Columns sharing the same letter are not significantly different (Student's t, P<0.05).  Table A 2.2. Summary of ANOVAs on the number of mice and voles caught in the fertilized plots at the Boutellier and Microwave Road sites for the 2 trapping sessions. Significant effects (P < 0.05) are shown in bold.  (a) June census  SS 3.781 13.781  F-ratio 1.824 6.648  P 0.189 0.017  1  1.531  0.739  0.399  1 1 1 1 24  0.781 11.281 0.031 0.281 49.750  0.377 5.442 0.015 0.136  0.545 0.028 0.903 0.716  Source Site Fertilizer  df 1 1  Fence  Site x Fertilizer Site x Fence Fertilizer x Fence Site x Fertilizer x Fence Error  127  ^  (b) August census  Site Fertilizer Fence Site x Fertilizer Site x Fence Fertilizer x Fence Site x Fertilizer x Fence Error  1 1 1 1 1 1 1 24  0.781 0.781 0.031 0.781 1.531 1.531 1.531 55.750  0.336 0.336 0.014 0.336 0.659 0.659 0.659  0.567 0.567 0.909 0.567 0.425 0.425 0.425  VOLES CAPTURED IN TRAPS WITH FERTILIZER AS BAIT Adding fertilizer to the standard barley + apple bait had no effect on the mean number of voles captured (Fig. A 2.5) and there was no effect of site (Table A 2.3). 8-  co u)  3-  02-  0 Traps with^Traps without fertilizer^fertilizer Figure A 2.5. The mean number of voles captured per trap and the 95% confidence interval for traps with and without fertilizer included in the standard barley + apple bait. These are not statistically different (P<0.05, Table A 2.3).  Table A 2.3. Summary of ANOVA on the number of voles captured in traps with and without fertilizer included in the bait.  Source Fertilizer Site Fertilizer x Site Error  df 1 2 2 30  SS 18.778 32.056 22.389 224.000  F-ratio 2.515 2.147 1.499  P 0.123 0.135 0.240  128  VOLES ENTERING LOCKED-OPEN TRAPS TO REMOVE FERTILIZER The traps that were locked open , with known masses of fertilizer, were entered beginning the first day they were set up ( Figs. A 2.6, A 2.7). The total amount of fertilizer removed, either by consumption or by being carried away, was not related to the number of traps entered (Fig. A 2.7; r = 0.356, P = 0.212) and the mean m ass of fertilizer removed per trap was not related to the estimated number of times that the traps were entered (Fig. A 2.8; Wilcoxon Test, P = 0.471).  Figure A 2.6. Evidence inside a Longworth mouse trap that voles had entered the trap. This trap was locked open and initially contained 5.0 g of fertilizer.  129  8-0 7H 2 2 6()  - 0.8  —•-- Number of traps entered - - - e- - - Fertilizer removed (g)  0.7  rn  0.6 '^1:3 0 T0.5  5^-1  8  H 0.4  Lu 4  0 3_  0.3 ",,,1"  0  _c) E 2 Z 1^-  L 0.2  6. 0 0  0 0  5  10  ,•  t  _ 0.1 0 15  Day Figure A 2.7. The number of traps entered per day (solid line and solid circle) and the total fertilizer removed (g) per day (dotted line and open circle). There was no correlation between number of traps entered and total mass of fertilizer removed (r = 0.356, P = 0.212).  0.25 0- 0.2 E'3. rn 0 0.15 > 0  7  E  ,c2^0.1 N a)  u_  0.05 -  I  I  0 low^medium^high Number of tracks Figure A 2.8. Mean mass of fertilizer removed (g trap - ') and 95% confidence interval depending on the number of tracks going into and out of the traps. There is no statistical difference between these classes (Wilcoxon Test, P = 0.471)  130  DISCUSSION At the end of the Kluane Boreal Forest E cosystem Project (KBFEP), it was concluded that only two groups of vertebrates responded to the fertilization treatments, passerine songbirds (Smith and Folkard 2001) and microtines (Boonstra et al. 2001). Here I also observe a significant response by voles to the addition of fertilizer to vegetation both in terms of over-winter nests and more voles captured early in the season in fertilized vegetation plots. I also observed that voles remove fertilizer from traps that were locked open to provide unrestricted access, although I don't have data to determine if the voles and mice ate the fertilizer, or if they simply removed it. There has been relatively little work done on the effect of fertilizer addition to microtine populations. Most work has focused on the changes in the quality of plant food with increased fertilizer. For example, laboratory experiments have shown that fertilized plant material is more palatable to Microtus spp. than unfertilized feed (Rousi et al. 1993; Hartley et al. 1995). Similarly, Clethrionomys rufocanus voles in tundra plant communities preferred to feed and live in fertilized areas than unfertilized areas (Grellmann 2002). However, in one of the few field experiments that examined the effect of fertilizer addition on the herbivore community, and microtines specifically, Ball et al. (2000) reported that there was a decrease in observed tracks and usage of fertilized plots during their winter sampling. Boonstra et al. (2001) also reported a decrease in Clethrionomys populations due to fertilization during the KB REP. It was postulated that with fertilization, there was a reduction in low berry-producing shrubs (Turkington et al. 1998) and this coincided with a reduction in Clethrionomys abundance. Microtus spp. had an opposite response and increased with fertilization (Boonstra et al. 2001). It was postulated that this occurred because the meadows and grasslands, which Microtus prefer, is nutrient limited and fertilization stimulated grass growth (Nams et al. 1993; Turkington et al. 1998). This pattern was also observed in the CDS experiment (Chapter 5). There was increased grass growth in the fertilized plots and more over-winter nests were observed in these plots. However, the question arises, does the increase in grass abundance due to fertilizer addition cause the increased vole activity or are voles simply attracted to the fertilizer? No data were collected on the amount of fertilizer cached in the CDS plots — the fertilizer was left in place because it was part of the experimental treatment. It was evident, however, that based on ne st counts, the fertilized plots had more vole activity. 131  The trapping of voles using fertilizer as bait was a poorly designed experiment for answering any questions on whether mice and voles are attracted to fertilizer. The voles that went into a "fertilized" trap were just as likely to be recaptured in a "non-fertilized" trap and vice versa (data not presented). Similarly, the August census in the 5 m x 5 m plots that had been fertilized for 13 years did not detect a difference in number of voles captured in each treatment partially because the density of voles was so high that all the traps were usually filled each night. If more traps were set out, it may have been possible to determine if the voles preferentially chose fertilized or unfertilized traps. In 2002, vole abundance was the highest ever recorded at Kluane (39.50 voles/ha, unpublished data, C. J. Krebs, UBC). In fact, I may not have observed any of these patterns if it were not for the exceedingly high abund ance of voles that year. In 2001, in 6 nights of trapping, I only caught a single red- back vole — twice. The removal of fertilizer from the traps that were locked open and baited with only fertilizer, points to the idea that the voles were influenced by more than quality food alone. A plausible, yet untested, hypothesis is that the voles were consuming the fertilizer as a source of micro- or macronutrients. One nutrient that many mammals, including microtine rodents, "crave" is sodium (Robbins 1993). In a field experiment, sodium soaked sticks were actively sought out and chewed by meadow voles (Hansson 1990; Hansson 1991). There should not be excessive quantities of sodium in commercial fertilizer; however, other mineral constituents may attract the voles to the fertilizer. The results presented here are intriguing in that they join a small set of experiments demonstrating a relationship between fertilizer and voles. Whether voles are attracted to the fertilizer alone or the higher quality feed it produces remains unclear.  132  REFERENCES Ball, J.P., Danell, K. and Sunesson, P., 2000. Response of a herbivore community to increased food quality and quantity: an experiment with nitrogen fertilizer in a boreal forest. Journal of Applied Ecology, 37: 247-255. Bonan, G.B. and Shugart, H.H., 1989. Environmental factors and ecological pro cesses in boreal forests. Annual Review of Ecology and Systematics, 20: 1-28. Boonstra, R., Krebs, C.J., Gilbert, S. and Schweiger, S., 2001. Voles and mice. In: C.J. Krebs, S. Boutin and R. Boonstra (Editors), Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, Oxford, New York, pp. 215239. Forsyth, A., 1999. Mammals of North America: temperate and arctic regions. Firefly Books Ltd., Willowdale, ON. Grellmann, D., 2002. Plant responses to fe rtilization and exclusion of grazers on an arctic tundra heath. Oikos, 98: 198-204. Hansson, L., 1990. Mineral selection in microtine populations. Oikos, 59: 213-224. Hansson, L., 1991. Bark consumption by voles in relation to mineral contents. Journal of Chemical Ecology, 17: 735-743. Hartley, S.E., Nelson, K. and Gorman, M., 1995. The effect of fertilizer and shading on plant chemical composition and palatability to Orkney voles, Microtus arvalis orcadensis. Oikos, 72: 79-87. Krebs, C.J., Boutin, S. and Boonstra, R., 2001. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York. Nams, V.O., Folkard, N.F.G. and Smith, J.N.M., 1993. Effects of nitrogen fertilization on several woody and nonwoody boreal forest species. Canadian Journal of Botany, 71: 93-97. Robbins, C.T., 1993. Wildlife feeding and nutrition. Academic Press, San Diego, California. Rousi, M., Tahvanainen, J., Hentton en, H. and Uotila, I., 1993. Effects of shading and fertilization on resistance of winter-do rmant birch (Betula pendula) to voles and hares. Ecology, 74: 30-38. SAS, 1995. JM P. SAS Institute Inc., Cary, North Carolina. Smith, J.N.M. and Folkard, N.F.G., 2001. Other herbivores and small predators: arthropods, birds and mammals. In: C.J. Krebs, S. Boutin and R. Boonstra (Editors), Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, Oxford, New York, pp. 261-274. Turkington, R., John, E., Krebs, C.J., Dale, M.R.T., Nams, V.O., Boonstra, R., Boutin, S., Sinclair, A.R.E. and Smith, J.N.M., 1998. The effects of NPK fertilization for nine years on boreal forest vegetation in northwestern Canada. Jour nal of Vegetation Science, 9(3): 333-346. Turkington, R., John, E., Watson, S. and Seccombe-Hett, P., 2002. The effects of fertilization and herbivory on the herbaceous vegetation o f the boreal forest in north-western Canada: a 10-year study. Journal of Ecology, 90: 325-337.  133  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0066297/manifest

Comment

Related Items