UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Sources of variation in larval survival, growth and development rates and their consequences for adult… Anholt, Bradley Ralph 1988

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

Item Metadata

Download

Media
831-UBC_1989_A1 A53.pdf [ 4.71MB ]
Metadata
JSON: 831-1.0098136.json
JSON-LD: 831-1.0098136-ld.json
RDF/XML (Pretty): 831-1.0098136-rdf.xml
RDF/JSON: 831-1.0098136-rdf.json
Turtle: 831-1.0098136-turtle.txt
N-Triples: 831-1.0098136-rdf-ntriples.txt
Original Record: 831-1.0098136-source.json
Full Text
831-1.0098136-fulltext.txt
Citation
831-1.0098136.ris

Full Text

Sources of variation in larval survival, growth and development rates and their consequences for adult survival and reproductive success in Enallagma boreale Selys (Odonata: Coenagrionidae) By B R A D L E Y R A L P H A N H O L T B.Sc, The University of Alberta, 1979 M.Sc, The University of Calgary, 1982 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December 1988 © Bradley Ralph Anholt, 1988 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Zeotogyr l ' . r l!v • The University of British Columbia Vancouver, Canada Date December 12, 1988 DE-6 (2/88) Abstract The sources of variation in larval survival, growth, and development rates and their consequences for adult survival and reproductive success of the coenagrionid damselfly Enallagma boreale Selys were examined in two years with contrasting weather. The effect of larval density, food availability, and interference competition was studied in field enclosures. Habitat complexity was manipulated using artificial macrophytes to decouple exploitative and interference competition. At high larval densities or low habitat complexity, larvae were more evenly spaced among the artificial macrophytes than expected if they were distributed independently of each other. Lower survival, growth and development rates were exhibited by larvae that experienced high density and low food availability. There was no effect of the habitat complexity manipulation on these rates. Interference competition therefore has a low cost. Adults that had been experimentally manipulated as larvae were individually marked to assess whether larval conditions affected adult survival and reproduc-tive success. Females had lower survival to sexual maturity than males. Females increased their mass by nearly 30 percent independent of their mass at emergence. Small males increased their mass more than large males but the mean mass gain was less than five percent of mass at emergence. Female survival to reproductive maturity was independent of size at emergence in both years. Large males had better survival in 1985, a warm dry year, but there was no relationship in 1986, a cool wet year. Male survival was independent of date of emergence in 1985, but late emerging males had better survival in 1986 (after the weather had improved). Early emerging females survived better in 1985, but late emerging females did better in 1986. Lower survival of females and small males suggests that mortality is correlated with mass gain in the pre-reproductive period, possibly due to increased risk of ii predation while foraging. The target of selection may not be body size but be-haviour which is correlated with body size. Increased mass due to foraging almost certainly contributes to reproductive success of females by providing resources for eggs. Ignoring the mortality associated with this increase in mass underestimates the variance in reproductive success of females. While large males may have had better survival, of those males that did survive, small males had higher reproductive success. Reproductive success was independent of size at or date of emergence for females in both 1985 and 1986. iii Table of Contents Page Abstract i i Table of Contents iv List of Tables vi List of Figures vii Acknowledgements viii Chapter One Introduction 1 Chapter Two Larval Competition 3 Methods 5 Study Site 5 Cages 5 Experimental Design 5 Experimental Animals 7 Sampling of Cages 8 Analysis 8 Results 10 Distribution of damselflies 10 Survival 10 Date of emergence 16 Mass at emergence 16 Overall effects 20 Variability of the response 20 Discussion 20 Chapter Three The Pre-reproductive Period 27 Methods ..28 Weather 28 Measurements and marking 28 Recapture 29 Analysis 29 i v Results 30 Weather 30 Survival 30 Pre-reproductive period 30 Pre-reproductive mass gain 30 Dispersal 35 Discussion 40 Chapter Four Adult Survival and Reproductive Success 43 Methods 45 Field methods 45 Phenotypic measurements and their principal components 45 Selection Analysis 51 Results 53 Opportunity for selection 53 Selection for survival to reproductive maturity 53 Mating success 65 Discussion 70 Differences between males and females 70 Year to year variation 80 The influence of dispersal 81 Variance in lifetime reproductive success 81 Chapter Five General Discussion 83 Separating interference and exploitative competition 83 Measurement of selection on adults 85 The pre-reproductive period and odonate mating systems 86 Literature Cited 88 Appendix 96 v List of Tables Table I—Analysis of variance table for the probability of larval survival 15 Table II—Analysis of variance table for date of adult emergence 19 Table III—Analysis of variance table for mass at emergence 23 Table IV—Mean and standard deviations of unstandardized phenotypic characters before and after selection 47 Table V—Proportion of variance in phenotypic characters accounted for by their first two principal components : 48 Table VI—Estimates of variance in relative individual fitness 54 Table VII—Selection differentials of the phenotype for survival from emergence to sexual maturity 64 Table VIII—Selection differentials of the phenotype for number of matings 77 Table IX—Estimates of curvature selection for both survival and reproductive suc-cess 78 vi List of Figures Figure 1—Relationship between mean and variance of the number of damselflies per perch 11 Figure 2—Daily probability of survival for the three larval experiments 13 Figure 3—Mean date of emergence of adults from the three larval experiments . 17 Figure 4—Mean mass at emergence of adults from the three larval experiments 21 Figure 5—Daily maximum temperature and mm of precipitation while adults were active 31 Figure 6—Length of the pre-reproductive period 33 Figure 7—Relationship between mass gain and mass at emergence 36 Figure 8—Mass at emergence of dispersing and returning adults 38 Figure 9—Loadings of the phenotypic characters on the principal components .. 49 Figure 10—Distribution of male phenotypes at emergence and at sexual maturity in 1985 . . . . 55 Figure 11—Distribution of female phenotypes at emergence and at sexual maturity in 1985 58 Figure 12—Distribution of male phenotypes at emergence and at sexual maturity in 1986 60 Figure 13—Distribution of female phenotypes at emergence and at sexual maturity in 1986 62 Figure 14—Distribution of all male phenotypes and phenotypes of mated males in 1985 66 Figure 15—Distribution of all female phenotypes and phentoypes of mated females in 1985 68 Figure 16—Distribution of all male phenotypes and phenotypes of mated males in 1986 71 Figure 17—Relationship between number of matings and mass for males in 1985 and 1986 73 Figure 18—Distribution of all female phenotypes and phentoypes of mated females in 1985 75 v i i Acknowledgements A thesis only gets finished with the help of many people and the forebearance of many more. I wish to thank my supervisory committee: C.J. Krebs, J.D. McPhail, J. Myers, D. Schluter, J.N.M. Smith, and especially my supervisor W.E. Neill. I valued their contributions, especially when we disagreed, their support and friend-ship during times that were occasionally trying. John Richardson helped in the field and the lab and turned what could have been a competition into a collaboration. Gordon Haas, S wizard, helped more than he will admit. The remaining graduate students, too numerous to name, staff, post-docs, and hangers-on of the Institute of Animal Resource Ecology and then the Ecology Group gave advice, abuse, help, and friendship when most needed. The Biosciences Data Center, especially Alistair Blachford, helped with analysis and presentation of the thesis. The University of British Columbia Research Forest built access to the study pond, provided accom-modation one year, rescued me from snow drifts and were generally helpful and agreeable. Funding for this research came from a Natural Sciences and Engineering Research Council operating grant to W.E. Neill and scholarship to myself. A sum-mer graduate fellowship was also provided by the University of British Columbia. The majority of the analysis and writing was done on a computer bought by my parents Dr. L.M. and Mrs. P.E. Anholt. For my spouse Beth I reserve special thanks. This thesis would not have been possible without her love, support, and patience. Finally, Maggot, Tiger, and Dragon kept reminding me that there were more things to life than my science. viii Chapter One INTRODUCTION The life-history of most animals undergoes a radical shift as they grow. For many this is characterized by a metamorphosis from larva to adult. For others profound shifts in life-history occur without morphological reorganization. Piscivorous fish begin life as planktivores and herbivorous lizards are carnivorous when small. Al-though such complex life-cycles (Istock 1967, Wilbur 1980) are widespread, they have received comparatively little theoretical attention. Most models of population regulation of organisms with complex life-cycles have only considered regulation within one part of the life-cycle (Istock 1967, Fretwell 1972, Kot and Schaffer 1984; Crowley et al. 1987, but see Prout 1986 for an exception). In these models, density-dependence due to larval conditions is treated as acting only in the larval stage and similarly density-dependence due to adult conditions acts only on adults. If density-dependence in the larval stage is only expressed as mortality then this is an appropriate formulation. However, density-dependence in the larval stage can also affect adult characters (Prout and McChesney 1985). If emergence at smaller sizes or later in the season affects sur-vival and reproductive success, then density-dependence is not restricted to a single portion of the life-history. However, few studies of factors limiting or regulating larvae have considered the consequences for adults. Harvey and Corbet (1985) and Smith (1983, 1987) are notable exceptions. Semlitsch et al. (1988) also examined the consequences of variation generated in the larval stage (although not its source) on adult survival and growth rates. Coenagrionid damselflies are common inhabitants of the weedy margins of per-manent water bodies. Larvae are often food-Hmited (Macan 1964, 1974; Lawton 1971; Baker 1986). Overt aggression between individuals may also reduce feeding 1 efficiency (Uttley 1980) or increase metabolic costs. Density dependent effects on larval survival, growth and development rates could be the consequence of prey depletion or interference. When damselflies are food-limited or raised at higher densities, they become adult at smaller sizes (Pickup et al. 1984; Harvey and Cor-bet 1985), and somewhat later in the season. Both effects may lower reproductive success (Harvey and Corbet 1985; Banks and Thompson 1987). In this thesis I examine the sources of variation in larval survival, growth and development rates and its consequences for adult survival and reproductive success in the coenagrionid damselfly Enallagma boreale Selys. Chapter Two describes experiments that disentangle density effects due to food depletion and interference among larvae. Chapter Three presents data on the interval after emergence and prior to sexual maturity. Chapter Four presents an analysis of the consequences of the variation produced by the larval regime on adult survival and reproductive success. 2 Chapter Two LARVAL COMPETITION Competition, whether interspecific or intraspecific, is usually defined phenomeno-logically. Competition is said to occur when increased density is associated with a reduction in the survival, growth rate, or reproduction of individuals. However, as Tilman (1987) has pointed out, the mechanism of competition is usually unclear. Competition can occur because there is less of a resource available per individual (exploitative competition). In addition, behavioural interactions between individ-uals can reduce fitness by decreasing the efficiency of food acquisition, increasing metabolic costs (food-related interference competition), or by increasing the risk of injury and mortality (mortality-related interference competition). Because the expected population dynamics of exploitative and interference competition differ (Schoener 1973, 1976; Jensen 1987), their relative strength has usually been inferred by examining the rate of change of some performance variable (usually individual or population growth rate) as a function of density and food supply (Schoener 1973, 1976, 1978; Rasmussen 1983). While investigators have de-clared their data to be more consistent with one form of competition or another, a stronger inference could be made by manipulating the intensity of the two processes independently of each other. Coenagrionid damselflies (Odonata:Zygoptera) are ideally suited to such a ma-nipulation. Strong behavioural interactions are known to occur between larval dam-selflies on perches (usually macrophytes) which are preferred hunting sites (Baker 1980, 1981, 1983; Rowe 1980; Uttley 1980). During encounters individuals assume aggressive stances and then may escalate, striking at each other with their labia. This results in one or both animals swimming away from the encounter site. In-dividuals with prior residence or a size advantage are less likely to be displaced 3 (Baker 1983). This behaviour may incur substantial metabolic and time costs that should reduce growth and development rates. If there are different costs for winners and losers of interactions, then growth and development rates should become more variable. The influence of interference to population processes of odonates was first suggested by Macan (1964) and has been repeatedly invoked to explain observed density effects (Johnson et al. 1984, 1985; Harvey and Corbet 1985; Pierce et al. 1985; Crowley et al. 1987; Van Buskirk 1987). Crowley et al. (1987) argue that density-dependent mortality due to interference among larvae has a larger effect than exploitation of food or non-mortality costs of interference in determining the dynamics of damselfly populations. Because damselfly larvae prefer to sit on perches, the frequency of interactions can be manipulated independently of food by altering perch density while main-taining a fixed bottom area. Increasing the abundance of perches should reduce the frequency of encounters and thereby reduce the intensity of interference competition without affecting the supply of planktonic prey. If interference competition affects population processes of damselflies then in-creased habitat complexity, with its concomitant reduction in interference compe-tition should: 1) reduce mortality rates 2) increase growth rates 3) increase development rates 4) increase prey consumption 5) reduce variation in growth and development rates among individuals. The occurrence of any or all of these effects would be evidence for the presence of interference competition. 4 METHODS Study Site This study was carried out in Thesis Pond, a small oblong (25 m x 13 m) bog pond located in the University of British Columbia Research Forest (49° 19' N, 122° 34' W) near Haney, B.C. Canada. Experimental animals were collected from Placid Lake a small (1.87 ha) bog lake near the primary study site. Enallagma boreale is the most abundant odonate in Thesis pond and Placid Lake. The majority of damselflies overwinter in the penultimate (F-l) or antepenultimate (F-2) instar. Cages Thirty-six cages (0.5 m x 0.5 m x 0.5 m) were constructed of 4 x 4 cm lumber with bottoms of 9 mm plywood. Two opposing sides were of sunlight-resistant polyethy-lene sheeting which also lined the bottom. The remaining sides were covered with 225 /xm mesh polyester cloth and the top was covered with 1 mm mesh insect screening supported on a wood frame. All edges were sealed with silicone sealant. Artificial macrophytes were evenly spaced 40 cm long strands of polypropylene rope stapled to the bottom of the cage. Groups of six cages were connected in two rows of three cages 85 cm apart so that the cloth sides of opposing cages faced each other. Plastic sides faced the neighbor 4 cm away. Two groups of cages were supported on a float made of four sealed ABS plastic pipes (3 m long, 15 cm wide). Plywood covering the float acted as a working platform. Three floats were moored to a log in the pond so that cages were not adjacent to the shore. Experimental Design 1985 The design was a fixed effects analysis of variance with four factors completely crossed. Factors were food availability, larval density, habitat'complexity and larval starting instar. Two replicates were stocked with F- l larvae and one was stocked 5 with F-2 larvae. There were two food levels, 10 or 30 plankters L _ 1 . The low food level was chosen because this allows 95% of the larvae of the coenagrionid Ischnura elegans (Van der Lind.) to complete metamorphosis (Lawton et al. 1980). Damselfly density was 24, 48, or 96 animals per cage (divided equally between the sexes), i.e. .25, .5 and 1 damselfly L _ 1 . Habitat complexity was either 'simple' (five perches) or 'complex' (forty-nine perches). These ranges of density and habi-tat complexity encompass observed field densities and numbers of damselflies per macrophyte. The experiment was set up between May 11 and May 18, 1985. The first animal emerged June 7, 1985. 1986 In 1985 interactions between main effects were small. The experiments were there-fore repeated as two separate two-factor analysis of variance experiments, a larval density xfood availability experiment and a perch availability x starting instar ex-periment. More treatment levels covering a wider range could be tested and the magnitude of the treatment could be included in the analyses. Because the interac-tion terms in 1985 had contributed such a small amount to the explained variance the experiments were not replicated beyond crossing of the factors. Four food levels (5, 10, 20, or 30 plankters L _ 1 ) were crossed with five densities (24, 48, 72, 96, or 120 F- l instar damselflies per cage). Twenty perches were provided in each of the 20 cages. The experiment was set up between April 27 and May 29, 1986. The first animal emerged on June 23, 1986. The habitat complexity experiment had 5, 20, 35, or 49 perches per cage. This factor was crossed with a size factor of 96 F- l , or 96 F-2, or 48 F- l and 48 F-2 damselflies for a total of 12 cages. Twenty plankters L _ 1 was the food density. The experiment was set up on April 10, 1986. The first emergence was June 7, 1986. 6 Experimental Animals Damselflies Over 2000 animals were required to set up the experiments in 1985 and in 1986. Collecting from Thesis Pond would have severely disturbed the population so larvae were collected by dip-net from Placid Lake 2 km away. The animals from Placid Lake were in the same instars and the same mass at each instar as those from Thesis Pond. To minimize the variation between experimental treatments, the size range of damselflies in the experiments was restricted to a single instar. Larvae in the F-l or F-2 instars were used because of their abundance. Animals were taken to the lab where they were narcotized in CO2 super-saturated water from a soda syphon. The animals were measured from the tip of the labium to the tip of the abdomen (excluding lamellae) with Vernier calipers to determine the instar, sexed, assigned to treatments using random numbers, and added to the experiment the following day. Observations in the first week after the animals had been added to the cages indicated minimal mortality (four deaths out of 2016 in 1985) due to stocking. An analysis of the starting sizes showed that there were no differences in the size of animals within treatments (ANOVA p > 0.19). Pre y Crustacean zooplankton were collected from nearby Ashless Shirley Lake with a plankton net. Collections were passed through a 1.3 mm opening sieve to remove almost all Chaoborus larvae. Only a few potential competitors were added to the experiment as a result. Prey smaller than 1 mm were not included in the estimates of prey density but contributed only a tiny amount to biomass in any case. Copepods (Diaptomus kenai, D. leptopus) and cladocerans (Daphnia rosea) were the principal prey. 7 Sampling of cages Food Food levels were restored to experimental levels twice weekly. Prey density in the cages was estimated by enclosing five replicate 1.2 L columns of water in a cylinder sampler fitted to a foam-covered base. Crustacean zooplankton prey larger than 1 mm were counted in a white enamel pan. No attempt was made to distinguish between different species of prey. The sampling method was calibrated by replicated sampling of a cage to which a known number of prey had been added. Additional prey were added by volume to the cages from collections from Shirley Lake of known prey concentration to restore densities to experimental levels. Dispersion For the habitat complexity manipulation to be meaningful, convincing evidence that the damselflies respond behaviourally to the number of perches is required. I counted the number of damselfly larvae on each perch in treatments with 20 or fewer perches, and on 20 randomly chosen perches in treatments with more than 20 perches. The same perches were counted at each census of a cage. Adults Adults were collected from the cages on the day of their emergence and taken to the lab in ventilated 1L plastic bottles. The animals were left overnight to harden and then weighed the following morning on an analytical balance to ±0.1mg while restrained in a glassine envelope. Analysis Dispersion The relationship of the variance to the mean number of damselflies per perch can be used to determine if the distribution of damselflies among perches changes with 8 perch abundance. If the likelihood of another individual using a perch is indepen-dent of the number already on the perch then a Poisson distribution is expected with the mean number per perch equal to the variance. I predicted that aggres-sive interactions among individuals would lead to a more even distribution among perches (variance less than the mean) as the mean number of damselflies per perch increased. A weighted regression of logevariance against logemean was done (PROC GLM of PC-SAS) using the inverse of the mean number per perch as weights. This overcomes the heteroscedasticity of the logevariance (Draper and Smith 1981). Survival The daily rate of survival was calculated from the beginning of the experiment to the median date of emergence using the formula Proportion surviving = (daily probability of survival)days. This removed the confounding effect of a prolonged larval period reducing total survival but assumes that survival rate did not vary during the experiment. A fixed effects analysis of variance was performed using the arc-sine square-root trans-formed values (PROC GLM of PC-SAS). The level of replication permitted inclusion of two-way interaction terms in 1985 but not 1986. Emerging Adults 1985 The analysis of date of emergence and mass at emergence was a simple fixed effects ANOVA with the two-way interaction terms included (PROC GLM of PC-SAS). Males and females had identical responses to the treatments therefore mean values for each cell were calculated without reference to sex. The mean emergence date (counting from the first emergence: June 8, 1985 or June 2, 1986) of the survivors from a cage was used for each cell value in the analysis. Because survival, date 9 of emergence and mass at emergence are correlated, a multivariate ANOVA was performed to determine if there was an overall treatment effect (Wilbur et al. 1983). The variability of the phenotypic characters at emergence was examined by using the coefficient of variation of the characters of the emerging animals. This measurement was normally distributed with stable variance in a quantile-quantile plot against standard normal deviates. 1986 Data were analyzed as in 1985 but without interaction terms because the experiment was not replicated beyond the crossing of the factors. RESULTS Distribution of damselflies When the number of damselflies per perch was low, the variance was approximately equal to the mean, suggesting that animals were distributed independently of each other among perches. As density increased the variance increased more slowly than the mean; damselflies were more uniformly distributed among the available perches than at low numbers per perch (Figure 1). The slope of a weighted log-log regression of variance against the mean was 0.52 when pooling all censuses in 1985 and 0.72 in 1986. The slopes of the regressions for every census are less than one, and seven of the nine regressions were significantly less than one. Survival The daily probability of survival was high, with no mortality in some cages. When F-2 larvae were raised with only 10 prey L - 1 they had much lower survival than other treatments (Figure 2a). Increasing damselfly density to more than 96 per cage (1 L - 1 ) and reducing prey density to as low as 5 L _ 1 further reduced survival of F-l larvae in 1986 (Figure 2b) but never to the level observed for F-2 larvae in 1985. Manipulating habitat complexity had almost no effect on survival. These 10 F i g u r e 1 — T h e relationship between log e mean number of damselflies per perch and loge variance of the number of damselflies per perch on 5 censuses i n 1985. The solid Une is a weighted regression and the dashed line is the expectation of mean = variance if damselflies choosing perches independently of the presence of other damselflies. 11 F i g u r e 2 — D a i l y p robab i l i t y of surv iva l for the three l a rva l exper iments , a) 1985 F o o d X D e n s i t y x S t a r t i n g i n s t a r x H a b i t a t comp lex i t y exper iment . F o o d levels were 10 or 30 prey L - 1 . Damse l f l y dens i ty was 24, 48, or 96 an imal s per cage (0.25, 0.5, a n d 1.0 L _ 1 ) . S ta r t ing instars were penu l t ima te ( F - l ) a n d an tepenu l t imate ( F -2). V e r t i c a l bars are s t anda rd errors of the est imate us ing the hab i ta t comp lex i t y t rea tment s (5 or 49 perches per cage) as repl icates, b) 1986 F o o d x Dens i t y exper-imen t . P r e y dens i ty was 5, 10, 20, or 30 prey L - 1 . Damse l f l y dens i ty var ied f r o m 0.25 to 1.25 L _ 1 . H a b i t a t comp lex i t y was f ixed at 20 perches per cage, c) 1986 S ta r t ing ins tar x H a b i t a t comp lex i t y exper iment . N i n e t y - s i x damself l ies began each exper iment . T h e ' m i x e d ' t rea tment was set u p w i t h 48 F -2 a n d 48 F - l damselfl ies. D a i l y probab i l i t ies of surv iva l were ca lcu la ted f r o m the p r o p o r t i o n surv iv ing a n d the in terva l f r o m the start of the exper iment to the m e d i a n date of emergence. N o t e that the range of surv iva l probabi l i t ies are not i dent i ca l be tween figures. 13 F-1; 30/L F-2; 30/L F-1;10/L A F-2; 10/L 24 48 96 number of damselflies per cage 24 48 72 96 120 number of damselflies per cage F-1 A V £ A • 0 F-2 s mixed 5 20 35 49 number of perches per cage 14 TABLE I—Analysis of daily probability of survival (arc-sine square-root trans-formed) for all three experiments with main effects and all two-way interaction effects where replication warrants. R 2 values are the proportion of the total variation accounted for by that component. 1985 FoodxDensity XStarting instarxHabitat complexity experiment Sum of Source i df Squares R 2 F P< Model 14 3.0 X 10" -2 0.69 3.3 0.007 Food 1 4.5 X 10" -3 0.10 6.8 0.02 Density 2 9.0 X io--3 0.20 6.8 0.006 Instar 1 2.8 X 10" -3 0.06 4.2 0.06 Perches 1 7.6 X 10" -5 < 0.01 0.1 0.74 FoodxDensity 2 6.2 X 10" -3 0.14 4.7 0.03 Foodxlnstar 1 1.4 X 10" -3 0.03 2.2 0.16 FoodxPerches 1 1.6 X 10" -4 < 0.01 0.2 0.63 Density xlnstar 2 3.7 X 10" -3 0.08 2.8 0.09 Density x Perches 2 2.8 X io--3 0.06 2.1 0.15 Instar x Perches 1 2.4 X io--3 0.05 3.6 0.08 Residual 21 1.4 X io--2 1986 FoodxDensity Experiment Sum of Source df Squares R 2 F P< Model 7 3.8 x 10- 4 0.83 8.1 0.001 Food 3 1.4 x 10~ 4 0.26 6.0 0.01 Density 4 2.4 x 10- 4 0.56 9.7 0.001 Residual 12 4.6 x 10~ 4 1986 Instar xPerches Experiment Sum of Source df Squares R 2 F P< Model 5 8.8 x 10-6 0.77 4.0 0.06 Instar 2 5.6 x 10-6 0.52 6.8 0.03 Perches 3 3.2 x 10~6 0.25 2.2 0.19 Residual 6 2.6 x 10-6 15 treatments had so little effect on survival (Table I) that they were used as replicates when calculating the standard error of food and density treatments in Figure 2a. Reduced survival was observed when F-l and F-2 larvae were mixed together in 1986 (Figure 2c). Increasing habitat complexity improved survival slightly for the 'mixed' treatment but not significantly. Date of emergence Increasing damselfly density or decreasing prey abundance delayed emergence date in both 1985 and 1986 (Figures 3a and 3b). Low food levels had more of an effect at high damselfly density than at low density (Figure 3a; Table II). The mean emergence dates were considerably later in 1986 compared to 1985 due to cold, rainy weather (see Chapter Three) after emergence had begun. Habitat complexity did not affect the date of emergence in 1985 where it accounted for a miniscule proportion of the total variation (Table II). No effect was observed in 1986 either (Figure 3c; Table II). The mean date of emergence for the mixed treatment was closer to that of F-l than F-2 larvae, probably indicating that the majority of surviving individuals were stocked as F-l larvae. Mass at emergence High damselfly density and low prey density both reduced the mass at emergence for F- l and F-2 larvae (Figure 4a; Table III). Animals that started the experiment as F-2 larvae emerged at lower mass than those started as F-l larvae under the same experimental conditions (Figure 4a). It appears by inspection that F- l larvae were more affected by reduced food availability at high damselfly density than F-2 larvae but I did not include three-way interaction terms in the analysis. Mean emergence mass was about 5 mg less in 1986 than 1985 but the pattern is qualitatively the same with high damselfly density and low prey density reducing emergence mass (Figure 4b). The mean emergence mass of treatments with both F-2 and F- l larvae was consistently higher than in treatments with only one instar. This, in addition 16 Figure 3—Mean day of emergence for the three larval experiments. a) 1985 FoodxDensityxStarting instarxHabitat complexity experiment, b) 1986 FoodxDensity experiment, c) 1986 Starting instar xHabitat complexity experiment. Vertical bars in a) are the standard errors of the mean using habitat complexity treatments as replicates. The day the first E. boreale emerged (June 8, 1985 or June 2, 1986) is defined as day 1. Symbols are fully explained in the legend of Figure 2. 0 17 40 CD A O 35 F-2; 10/L c - • erge 30 A A V F-2; 30/L E CD • | 25 m F-1; 10/L O day 20 m day $ F-1; 30/L 15 0 70 • O en 60 -mei 50 -CD M— O 40 • >^  da 30 -20 -60 -CD O m c CD 50 -o> CD E -CD 40 ->% CO • T5 30 -24 48 96 number of damselflies per cage 24 48 72 96 120 number of damselflies per cage -v V F-2 V V XX mixed A a A F-1 5 20 35 49 number of perches per cage 18 T A B L E II—Analysis of the mean date of emergence for all three experiments with main effects and all two-way interaction effects where replication warrants. R 2 values are the proportion of the total variation accounted for by that compo-nent. 1985 Food x Density x Starting instar x Habitat complexity experiment Sum of Source df Squares R 2 F V < Model 14 1.4 x 103 0.98 70.9 0.0001 Food 1 1.4 x 102 0.10 103.2 0.0001 Density 2 1.3 x 102 0.09 45.2 0.0001 Instar 1 1.1 x 103 0.75 758.0 0.0001 Perches 1 3.6 x 10- 3 < 0.01 0.0 0.96 FoodxDensity 2 2.2 x 101 0.02 ! 45.2 0.004 Food x Instar 1 2.8 x 10° < 0.01 2.0 0.18 Food x Perches 1 4.9 x 10 - 1 < 0.01 0.4 0.56 Density xlnstar 2 5.2 x 10° < 0.01 1.9 0.19 Density x Perches 2 2.0 x 10" 1 < 0.01 0.1 0.93 Instar x Perches 1 3.5 x 10 - 2 < 0.01 0.0 0.88 Residual 21 2.9 x 101 1986 FoodxDensity Experiment Sum of Source df Squares R 2 F P < Model 7 2.0 x 103 0.95 32.1 0.0001 Food 3 7.2 x 102 0.34 26.6 0.0001 Density 4 1.3 x 103 0.61 36.2 0.0001 Residual 12 1.1 x 102 1986 Instarx Perches Experiment Sum of Source df Squares R 2 F V < Model 5 8.5 x 102 0.99 135.8 0.0001 Instar 2 8.4 x 102 0.98 334.9 0.0001 Perches 3 1.1 x 101 0.01 3.0 0.12 Residual 6 7.5 x 10° 19 to the low survival and early emergence in the 'mixed' treatment suggests that cannibalism by F-l larvae on F-2 larvae was significant. Habitat complexity had no effect on mass at emergence in either 1985 or 1986 (Figures 4a and 4c; table III). Overall effects A multivariate analysis of variance was entirely consistent with the univariate anal-yses. Damselfly density, prey density, starting instar, and their interaction terms all had strong effects. Habitat complexity, or any interaction including habitat complexity had no effect. Variability of the response Analysis of variance using the coefficient of variation in the date of emergence or mass at emergence rather than the mean value of the emergers from a cage, revealed no differences in the variability of the responses between treatments. All probability values were greater than 0.15 D I S C U S S I O N All coenagrionid damselflies that have been tested show a preference for some sort of perch, and interact aggressively with conspecifics (Baker 1980, 1981, 1983; Rowe 1980; Johnson et al. 1984; Harvey and Corbet 1985; Pierce et al. 1985). Damselfly growth rates are often density-dependent (Macan 1964; Johnson et al. 1984) but evidence for prey depletion is lacking (Thompson 1982; Pierce et al. 1985). Similar results have been found for anisopteran species (Folsom and Collins 1982; Johnson et al. 1985; Johnson et al. 1987; Van Buskirk 1987). These authors have concluded by default that interference is the primary cause of density-dependent survival, de-velopment and growth in experiments with larval odonates. None of these studies has tried to disentangle the effects of interference and exploitation by manipulat-ing the rates of the two processes. Experiments designed to test for population 20 Figure 4—Mean live mass at emergence for the three larval experiments. a) 1985 FoodxDensityxStarting instarxHabitat complexity experiment, b) 1986 FoodxDensity experiment, c) 1986 Starting instar xHabitat complexity experiment. Vertical bars are the standard errors of the mean using habitat complexity treat-ments as replicates. Symbols and treatment levels are fully explained in the legend to Figure 2. 21 o o c <D D) i _ CD E CD CO D) "ST CO CO E CD O c CD O) CD E CD • CO E, c/> CO E CD O c CD D) k _ CD E CD ••—• CO c/> CO E 35 30 25 20 30 r 25 20 r 15 4 4 A T F-1; 30/L S F-2;3 - ; 30/L 0/L F-2; 10/L 24 48 96 number of damselflies per cage v O • • V V V 30/L O o A <> 20/L • A 10/L H 5/L 24 48 72 96 120 number of damselflies per cage 30 25 E A H A H A Jj^ mixed 20 V V V V F-2 15 20 35 49 number of perches per cage 22 T A B L E III—Analysis of the mean mass at emergence for all three experiments with main effects and all two-way interaction effects where replication war-rants. R 2 values are the proportion of the total variation accounted for by that component. 1985 Food x Density x Starting instar x Habitat complexity experiment Sum of Source df Squares R 2 F P< Model 14 3.3 x 102 0.88 11.3 0.0001 Food 1 1.4 x 102 0.38 67.9 0.0001 Density 2 7.2 x 101 0.19 17.3 0.0001 Instar 1 8.3 x 101 0.22 39.9 0.0001 Perches 1 4.0 x 10- 2 < 0.01 0.0 0.90 FoodxDensity 2 2.7 x 10° < 0.01 0.6 0.54 Foodxlnstar 1 1.0 x 10° < 0.01 0.5 0.50 FoodxPerches 1 3.4 x 10° < 0.01 1.6 0.22 D e nsity x Inst ar 2 4.8 x l O " 1 < 0.01 0.1 0.90 DensityxPerches 2 1.5 x 10- 1 < 0.01 0.0 0.97 Instar x Perches 1 2.7 x l O " 1 < 0.01 0.1 0.73 Residual 21 4.4 x 101 1986 FoodxDensity Experiment Sum of Source df Squares R 2 F P< Model 7 1.6 x 102 0.95 33.1 0.0001 Food 3 8.3 x 101 0.49 40.1 0.0001 Density 4 7.8 x 101 0.46 27.9 0.0001 Residual 12 8.4 x 101 1986 Instar x Perches Experiment Sum of Source df Squares R 2 F P< Model 5 3.2 x 101 0.89 9.5 0.009 Instar 2 2.9 x 101 0.81 21.6 0.002 Perches 3 2.8 x 101 0.08 1.4 0.34 Residual 6 4.0 x 10° 23 level effects of interference (Baker 1986, 1987) have been unable to establish their presence. Uttley (1980) and Crowley et al. (1987) made the specific prediction that increasing habitat complexity should decrease the level of interference competition. In these experiments I increased the total surface area of the cage by up to one-third. The increased surface area does not adequately reflect the increase in habitat complexity. Added perches reduced encounter rates effectively because the mean distance between two randomly chosen points on a stem is much larger than on a plane. I was unable to alter growth, development, or survival rates or the variability of these rates with this manipulation. It cannot be argued that a small response was present but inadequate replication failed to detect a statistically significant response. The F values for the perch treatment in the analysis were all small and in some cases were zero (Tables I to III). Lower overall survival was found when F-l and F-2 larvae were reared together and those that did: emerge weighed more than adults emerging from treatments stocked with only F- l larvae. This argues that cannibalism may have some effect on the population dynamics of this species as suggested by Crowley et al. (1987) but I see no evidence for density-dependent changes in growth or development rates attributable to feeding-related interference competition (Figures 3c, 4c). The densities used in these experiments are in the same range as those where interference has been implicated. Damselfly density was between 100 and 500 m~ 2 or 0.25 and 1.25 L _ 1 i n my experiments. These values are exceeded by those of Johnson et al. (1984) and Pierce et al. (1985) but their density estimates included animals of all instars. The biomass of my enclosures was actually greater than in the above experiments. The availability of the perches did, however, affect the distribution of the an-imals. Damselflies were more evenly distributed at combinations of density and perch availability treatments that increased the mean number of animals per perch. 24 Changes in distribution were also observed by Uttley (1980) and Convey (1988) in their laboratory populations. Although, behavioural interactions bring about this change in distribution, these interactions have very little cost associated with them. Why, do larval damselflies behave aggressively and defend perches? Perches could serve to reduce mortality due to predation. Significantly increased preda-tion rates have been observed for both invertebrate (Thompson 1987) and verte-brate (Convey 1988) predators of damselflies when habitat complexity was reduced. Whether the distribution of damselflies among perches affects their vulnerability to predation is not yet known. Cannibalism of smaller larvae by later instars may be a potent force in dam-selflies as argued by Crowley et al. (1987). When larvae of adjacent instars were raised together mortality rate of the smaller instar was high. The impact of canni-balism on the population could be assessed by comparing species that differ in the range of larval instars present at any one time. Comparing lestid damselflies, which have relatively synchronized life-cycles, with coenagrionid damselflies or comparing populations that are univoltine with those that are semivoltine may help to resolve this question. The density-dependent reduction in odonate growth rates which has been at-tributed to interference competition by numerous workers (Johnson et al. 1984, 1985, 1987; Pierce et al. 1985; Crowley et al. 1987; Van Buskirk 1987) may be attributable to localized prey depression (Charnov et al. 1976). Organisms will re-duce their foraging effort or alter foraging habitat when predation risk increases (Sih 1980; Ydenberg and Houston 1986; Gilliam and Fraser 1987). Behavioural changes by the prey that reduce their risk of predation will reduce the availability of prey to damselflies. As predation risk increases with predator density, the behavioural reduction of prey availability to damselflies may then be density-dependent. When sampling is done on too coarse a scale (whole water column, or entire enclosure) 25 resource depression will not be apparent. Density-dependent reduction in growth and development observed in my experiments could be due to both resource depres-sion and resource depletion. Manipulating habitat complexity, however, allowed me to rule out feeding-related interference competition. Damselflies were strongly food-Hmited in these experiments and the strength of the limitation was density-dependent. Although I used enclosures and a single class of prey, I consider that food-Umitation probably also applies in the field. Animals emerging from the cages encompassed the full range of sizes observed emerging from the pond. Faecal pellets produced by damselflies in the cages were also of similar mass to those produced by animals from the pond. Habitats are even more complex in the pond than in the cages. Eleocharis sp., a preferred perch in the pond, is more finely dissected than the simple polypropylene ropes used in the experiment. Thus, my manipulations of habitat complexity were biased towards showing a density-dependent effect due to interference if it existed. The absence of alternative benthic prey may have had some effect but damselflies are generalist predators that use whatever prey are available (Pearlstone 1973; Thompson 1978), switching hunting modes when one prey type becomes more abundant (Akre and Johnson 1979). Determining the extent that field populations of damselflies are limited in the larval stage will require larger scale manipulations of whole ponds or partitioned ponds. The separation of either feeding-related or mortality-related interference competition from exploitative competition in such future experiments will have to be done explicitly by the manipulation of individual behaviour. The assumption that the behaviour must have significant consequences because the behaviour is pronounced cannot be justified without more direct evidence. 26 Chapter Three THE PRE-REPRODUCTTVE PERIOD The time between emergence and the onset of sexual activity is the most poorly known stage in the life-history of odonates (Corbet 1962, 1980). At emergence, gonadal tissue is not fully developed, the exoskeleton is soft, and colouration is more cryptic than at sexual maturity. The maiden flight of odonates is oriented away from the natal pond (Moore 1954; Pajunen 1962; Parr 1976; Corbet 1980), and may lead the animal to another body of water. Dispersal between water bodies has rarely been recorded in odonates (Mitchell 1962; Koenig and Albano 1987; Van Buskirk 1986). There are also no estimates of whether dispersal is related to sex, size, or larval conditions. Crowley et al. (1987) have speculated that adult dispersal from crowded larval conditions was a possible density-dependent factor. The external dimensions of insects are fixed at the final moult but mass is not. Male and female Anisoptera can double their mass prior to reproductive ma-turity, with mature adult female mass exceeding that of males by up to 45 percent (Jenkins 1981). Even so, adult size has usually been treated as fixed at emergence (Fincke 1982; Hafernik and Garrison 1986). Increased gonadal mass may increase female fecundity. Increased muscle mass may improve mate-searching efficiency, or mate-guarding abilities of males. Increased storage products can be diverted to reproductive tissues, allow a male to spend more time in reproductive activity, or allow either sex to better survive inclement weather. The benefits of having stored reserves will depend on the mating system, prevailing environmental conditions, and the costs of acquiring these additional resources. In this chapter I present data on dispersal between adjacent water bodies, the length of the pre-reproductive period, survival to sexual maturity, and mass gain 27 prior to sexual maturity of a marked population of damselflies in two years with contrasting weather patterns. M E T H O D S Weather Daily maximum temperature and precipitation data were collected at the Univer-sity of British Columbia Research Forest Headquarters 8 km south of Thesis Pond in 1985 and 1986. Data from 1986 were strongly correlated (r2 > 0.9) with precipi-tation and temperature measured at a weather station 400 m south of Thesis Pond that began operation in 1986. Measurements and marking Larvae were raised in covered cages under varying density, food availability and habitat complexity (see Chapter Two for details). Adults were collected at emer-gence. Newly emerged wild adults were also collected from the margins of the study pond and nearby Shirley Lake. All adults were returned to the laboratory in venti-lated one liter jars and held overnight while the exoskeleton hardened. The following morning, wing length and abdomen length were measured with Vernier calipers to ± 0.1 mm. Body mass was measured on an analytical balance to ± 0.1 mg while the animal was restrained in a glassine envelope. Each animal was uniquely marked with a combination of four coloured paint dots (FaberCastell Uni-Paint Marker®), two on the left forewing, one on the thorax and one on the abdomen. The combined mass of the marks was less than 0.2 mg (0.5 to 1.0 percent live weight). After being marked, the animals were returned to the pond where they were released in the vegetation surrounding the pond. Animals that showed any difficulty in flying were not released. While the marks are small they may have contributed to increased metabolic costs of flight or subjected marked animals to increased risk of predation. No behavioural differences were noted between marked and unmarked animals. 28 Recapture Returning animals were identified by two observers equipped with binoculars who patrolled the study pond beginning in the morning as the first males arrived at the pond and continued until the evening when reproductive activity had ceased. Observers were present on all warm sunny days from the beginning of emergence in early June until the end of the breeding season in early September. When first recaptured, sexually mature adults were taken to the lab to be reweighed and then returned to the study site the following day. Animals that were not resighted may have died or moved to another body of water. Neighbouring Shirley Lake (0.1 km distant) or Katharine Lake (0.2 km distant) were searched for dispersing animals, 30 to 60 minutes each working day by one observer. Few dispersing animals were recovered from these nearby water bodies; most animals were recaptured at the pond. Analysis The length of the pre-reproductive period and the amount of mass gain during this time were both analyzed with a fixed effects model two-way analysis of variance using PROC GLM of PC-SAS. No distinction was made between reared and wild animals. Sex and year were the factors included in the model. Multiple regression analysis was used to determine if changes in mass during the pre-reproductive period were associated with its duration, date of emergence, or mass at emergence. The sexes were tested separately. Differences between animals that dispersed to other waterbodies and those recaptured at the natal pond were tested with a Wilcoxon signed-rank test af-ter standardizing for date of emergence using a robust locally-weighted regression (Cleveland 1979). 29 RESULTS Weather The weather was considerably warmer and drier in 1985 than 1986 (Figure 5). Between June 15 and July 15, 1985, when the majority of emergence took place, there were only three days with precipitation. For the same period in 1986, however, there were 16 rainy days. Daily maximum temperature, was rarely less than 20 °C, in 1985, but averaged less than 20 °C for the same time in 1986. Survival Fewer reproductively mature females than males were recaptured. In 1985, I re-captured 40 of 580 marked males (6.9%) and 26 of 576 females (4.5%). In 1986, 31 of 878 marked males (3.5%) and only 14 of 933 females (1.5%) were recap-tured. Female survival was not significantly lower than male survival in 1985 (X2 = 2.62; p < 0.11), but was in 1986 (x2 = 6.88; p < 0.009), and for the two years combined (x2 = 9.53; p < 0.002). The proportion recaptured was signif-icantly lower in 1986 than 1985 for both males (x2 = 6.42; p < 0.05) and females (X2 = 10.31; p < 0.002). Pre-reproductive Period The interval between emergence and return to the pond at sexual maturity was slightly longer for females than males in both 1985 (10.6 vs 9.6 days) and 1986 (14.7 vs 12.4 days) (Figure 6). A two-way analysis of variance showed that while the pre-reproductive period was significantly longer in 1986 than 1985 (F = 9.01; p < 0.005) the difference between the sexes was not significant (F — 1.9; p = 0.17). Pre-reproductive Mass Gain In both years females gained considerably more mass than males between release and recapture (Figure 7). Females gained (x±S.E.) 10.4 ± 0.7 mg in 1985 and 6.3 ± 1.8 mg in 1986. Males gained only 1.8 ± 0.6 mg in 1985 and lost 0.9 ± 0.4 mg in 1986. 30 Figure 5—Daily maximum temperature (°C) (continuous line) and mm of precipitation (vertical bars) for June 1 to August 31 of 1985 and 1986. The scale applies to both temperature and precipitation. 31 Figure 6—Interval between emergence and first resighting as sexually mature adults. Plots are of lower quartile, median and upper quartile. Extensions are the range of the data, to a maximum of 1.5 times the inter-quartile distance. Outliers are plotted individually. 33 £40 "D O "v_ CD Q_ 0 Z3 •4—» CC E 30 [ 20 'Z 10 C D 0 0 £40 "D T3 O " i _ 0 CL 0 v_ "3 •*—» co E E 30 I 20 I 10 C D C 0 0 females males females males A two-way analysis of variance of the data showed that the differences between the sexes were significant (F = 64.4; p < 0.0001) as were the differences between the years (F = 5.0; p < 0.03). There was no interaction effect (F = 0.03; p < 0.88). Males that emerged at a low mass tended to gain more mass in the pre-reproductive period than heavier males in both 1985 (r = —0.57; p < 0.006) and 1986 (r = —0.43; p < 0.02) (Figure 7). The smallest males gained almost 5 mg in 1985 but less than 2 mg in 1986. The largest males actually lost mass in both 1985 and 1986. There was a weak relationship between mass at emergence and mass gain for females in 1986 (r = —0.70; p < 0.02). This relationship was dependent on a single outlying data point which is probably a data recording error. Eliminating this point from the analysis greatly reduced the correlation (r = —0.40; p < 0.16). There was no relationship between pre-reproductive mass gain and mass at emer-gence for females in 1985 (r = 0.24; p < 0.22). Multiple regression showed that pre-reproductive mass gain was not associated with the date of emergence (number of days since the first adult emerged), or the length of the pre-reproductive period for males or females in 1985 or 1986 once the influence of mass at emergence was accounted for. Dispersal Of the 67 animals recovered at sexual maturity in 1985, two females and eight males were found at nearby water bodies. These ten animals were on average 2.1 mg heavier at emergence than animals recovered at the natal pond (Z = 2.9; p < 0.01) (Figure 8). The two groups did not differ in date of emergence, wing or abdomen length. Only two dispersing animals were recovered in 1986. 35 Figure 7—Relationship of mass gain with mass at emergence of males and females recaptured and reweighed in 1985 and 1986. 36 cn 3 w oo OO S-<—»• CD 3 CD - r CQ CD 13 O CD ro o ro cn CO o CO cn J 5 » o o o — I — o — I — ro o mass (mg) change o 3 cn _ ro 3 o 0 0 0 0 3 ro CQ cn CD 3 CD C O o CQ CD Zi o <D _^ 8 w mass (mg) change • o ro o i o cn ro 3 ° 0 0 0 0 3* ro CQ cn w - r CD 3 CD CO o CQ CD 13 O CD CO cn o 3 - . © S • CO ° > • • • cn ro o 3 CO 0 0 3 rO CQ cn CD 3 CD CO o CQ CD OS o CD _^ B 0 0 CD 0 > CO Figure 8—Mass at emergence of all resighted damselflies in 1986. Individuals returning to the natal pond (+) and those recaptured at nearby lakes (•) were both used in the calculation of the robust locally weighted regression plotted through the data. 38 40 F OS CO CD O c 35 CD ° ° O ) i _ CD E30 CD ' 03 o)25 E CO 83 20 E 15 h 0 10 20 30 date of emergence 40 DISCUSSION The reduction in survival in 1986 was associated with cooler, wetter weather when the majority of animals were still immature. Even if daily survival rates were not affected in cool weather, the prolonging of the pre-reproductive period (Buchholtz 1951; Corbet et al. 1960; Pajunen 1962; Ubukata 1974) would reduce overall sur-vival. Pre-reproductive mass gain was also lower in 1986 for both sexes. Either reduced activity on the part of damselflies or lower prey availability in the cool, wet weather could cause this. Few dispersing animals were recovered in either year, but more were recaptured in 1985. Dispersing animals were significantly larger than animals recaptured at the natal pond in 1985. Dispersal is therefore not likely to be density-dependent as proposed by Crowley et al. (1987) because animals emerging from crowded conditions are smaller than animals raised in uncrowded conditions (Chapter Two). The possibility exists that large animals from crowded conditions are more likely to disperse, but I have insufficient data to test this. One hypothesis to explain the relationship between body mass and the likelihood of dispersal is that heavier animals take a longer maiden flight than light ones and are therefore more likely to encounter another body of water and not return to the natal pond. This hypothesis remains to be tested. Females survived consistently less well than males. This might be due to sex-biased dispersal, as observed in some vertebrates (Greenwood 1980), but my evidence does not support this hypothesis. The only published estimates of pre-reproductive survival in odonates are for coenagrionid damselflies and all have lower female survival (Fincke 1982, E. hageni; Banks and Thompson 1985a, Coenagrion puella; L. Hamilton, pers. comm. Argia chelata) Mature, adult female coenagrionid damselflies are heavier than males by 30 to 80 percent (Appendix I). The association of lower survival to reproductive maturity and higher mass gain by females suggests 40 that reduced survival is a consequence of increased risk of predation while foraging. The widely observed male biased sex-ratio at the breeding site in zygopterans (Bick and Bick 1961, 1963, 1965; Bick 1972; Parr and Palmer 1971; Parr 1976; Hafernik and Garrison 1986) is therefore not simply the result of males spending more time at the breeding site but is at least partly due to differential mortality. Mass gain during the pre-reproductive period is widespread in the odonates and females usually gain more mass than males (Jenkins 1981, Appendix I). Sexual dimorphism in mass is not apparent until after emergence, although the strongly territorial and sexually dimorphic neotropical damselfly Mecistogaster coerulatus is probably an exception (Fincke 1984). In territorial species, heavy males may be better able to control access to breeding sites, or have more reserves to increase tenure on a territory. In the absence of territoriality, heavy males may not have a clear advantage, especially if there are mortality costs of gaining mass. My observa-tion that light male E. boreale gain more mass than heavy males, but not enough to compensate for emerging at a lower mass, suggests that such a trade-off is present. Females, on the other hand, increase their mass regardless of mass at emergence. The head and thorax of mature males and females weigh about the same, but fe-male abdomens are typically twice the mass of male abdomens, probably due to egg load (Jenkins 1981). Anax Junius and Plathemis lydia are exceptions, where males weigh more than females (Jenkins 1981); male abdomens were lighter but the thoraxes heavier than those of females. P. lydia is strongly territorial (Jacobs 1955; Koenig and Albano 1987). Personal observations of A. Junius suggest that it is also territorial. Wider investigation will probably find that mass gain of immature males is gen-erally more similar to gains in females in territorial than in non-territorial species, and that male survival is also more similar as a result. Three species of the Asian damselfly genus Mnais provide an ideal test of this hypothesis. Territorial orange-winged males are generally larger than non-territorial hyaline-winged males. Male 41 form is determined at least as early as emergence (S. Nomakuchi and H. Ubukata pers. comm.). I predict higher mass gain and lower survival for territorial males than non-territorial males. Higher mating success for territorial males suggests the possibility of an evolutionary stable strategy conditional on mass at emergence. An alternative hypothesis has been suggested by Singer (1982), who interprets the sexual dimorphism in mass of butterflies as selection for protandry. As there are no differences in the date of emergence of males and females or in the length of the pre-reproductive period, this hypothesis seems unlikely. 42 Chapter Four ADULT SURVIVAL AND REPRODUCTIVE SUCCESS The assertion that a particular morphology or behaviour is an adaptation can be tested by measuring the covariance of survival or reproductive success and that phenotype. Since selection tends to reduce additive genetic variance through both directional and stabilizing selection, the available phenotypes may not cover a suf-ficiently wide range for any experiment to have the power to detect the covariance. One solution to this difficulty is to broaden the range of phenotypes available for selection, either by selective breeding or using phenotypic plasticity due to envi-ronmental variation (eg. food availability, temperature). Because natural selection acts on phenotypes and is blind to the underlying genotype, the results of selection on an experimentally produced phenotype can be used to understand selection on natural variation. By raising larvae of Enallagma boreale under a wide range of larval densities and food availability I was able to markedly increase the variation in the size (wing and abdomen lengths) and mass at emergence, and the date of emergence (Chap-ter Two). While it is tempting to assume that changes from the norm will lead to reduced fitness, it is precisely such assumptions that have led to many of the criticisms of the adaptationist program (eg. Gould and Lewontin 1979). Actually measuring the covariance of fitness with morphology is a test of the hypothesis. In addition, the measurement of selection on these manipulated traits allowed me to test whether variation in the conditions experienced by larvae (Chapter Two) affected the adult stage of the life-cycle. Odonates are a particularly useful group for such a manipulation because of their abundance, relatively large size, localized reproductive sites, and ease of obser-vation. Considerable variance in reproductive success has been observed, especially 43 for males (Fincke 1982; Banks and Thompson 1985b; Koenig and Albano 1987; McVey 1988), but also females (Banks and Thompson 1987). Stabilizing selection for size has been found by both Fincke (1982) and Banks and Thompson (1985b) for male size and directional selection for female size (Banks and Thompson 1987). These studies concentrated on the sexually mature adults. However, little is known of the survival of individuals from emergence to reproductive maturity, and whether this is correlated with phenotype. Considerable argument has taken place over the separation of natural and sex-ual selection (Williams 1966, Otte 1979; Banks and Thompson 1985b; Koenig and Albano 1986) because of the contribution of survival to reproductive success. An al-ternative to examining the components of lifetime reproductive success is to estimate the strength of selection in the consecutive episodes that make up reproductive suc-cess. In this chapter I treat selection as having occurred during the pre-reproductive period (which deals exclusively with survival), then I examine whether there is se-lection for the number of mates acquired in those animals surviving to reproductive maturity. Lande and Arnold (1983) have extended the work of Pearson (1903) and pro-vided a theoretical basis for relating the covariance of a phenotypic trait with a fitness measure as a measure of natural selection. Arnold and Wade (1984a, 1984b) provide the basis for breaking up the life-cycle into episodes of selection. Because natural selection acts within a generation but the genetic transmission of the pheno-typic changes occurs between generations, the estimation of the strength of natural selection can be decoupled from the estimation of the heritability of the characters being selected. The consequences for evolutionary change in the population can be assessed when information on the heritability of the characters being investigated is known. Separating genetic transmission from the selective process is at variance with the terminology of Endler (1986) who would term this "phenotypic selection", 44 reserving the term "natural selection" for selection that occurs on phenotypic char-acters with a demonstrated genetic basis. It does, however, follow in the tradition of Fisher (1930) and Haldane (1954) who designated selection as occurring within generations and the evolutionary response to selection occurring between genera-tions. M E T H O D S Field methods Larvae were raised in covered cages under various levels of density, food availability and habitat complexity (see Chapter Two for details). Adults were collected at emergence. Newly-emerged wild adults were also collected from the margins of the study pond and nearby Shirley Lake. All adults were returned to the laboratory in ventilated one liter jars and held overnight while the exoskeleton hardened. The following morning, damselflies were measured and weighed (see below). Each animal was uniquely marked with a combination of four coloured paint dots (FaberCastell Uni-Paint Marker®), two on the left forewing, one on the thorax, and one on the abdomen. The combined mass of the marks was less than 0.2 mg (0.5 to 1.0 percent live weight). After marking, the animals were returned to the pond where they were released in the vegetation surrounding the pond. Animals that showed any difficulty in flying were not released. There are likely to be some additional costs of flight due to the weight of the marks but no behavioural differences were observed between marked and unmarked animals. Additional costs due to increased predation rates on marked animals are also possible. Returning damselflies were identified by two observers equipped with binoculars who patrolled the study pond beginning in the morning as the first males arrived at the pond and continued until the evening when reproductive activity had ceased. Observers were at the pond on all warm sunny days from the beginning of emergence in early June until the end of the breeding season in early September. 45 Phenotypic measurements and their principal components Wing and abdomen lengths were measured to ± 0.1 mm with Vernier calipers. Mass was measured to ± 0.1 mg on an analytical balance while the animal was restrained in a glassine envelope. Measurements of wing and abdomen lengths made at emergence and again at recapture were strongly correlated (r2 = 0.82 for wing length and r 2 = 0.80 for abdomen length). Differences in the measurements were uncorrelated between the two characters (v2 = 0.01). Unstandardized means of the four variables before and after selection are presented in Table IV. The three body size measurements and date of emergence are strongly corre-lated. Such highly collinear data poses special problems for the analysis of selection by multiple regression (Draper and Smith 1981; Lande and Arnold 1983; Mitchell-Olds and Shaw 1987), so the first two principal components were extracted to reduce the dimensionality of the data set after standardizing the characters to a mean of zero and a standard deviation of one (Lande and Arnold 1983; Schluter and Smith 1986). Sexes and years were treated separately. The characters met the assumptions of multivariate normality. The usual log transformation of morphological variables (Bookstein et al. 1985) and cube-root transformation for mass (Gould 1971) re-duced the fit to normality so the data were not transformed in this way. The first two principal components accounted for 85 to 90 percent of the variation in the data set (Table V). Character loadings on the first two principal components were nearly identical in the two years and between the sexes (Figure 9). The first principal component (PCI) is a generalized size variable with positive loadings for wing and abdomen lengths and mass at emergence. Date of emergence has a negative loading on PCI. The second principal component (PC2) is primarily a date variable. There is no loading of mass at emergence and light positive loadings of wing and abdomen lengths. 46 Table IV—Mean and standard deviation of phenotypic characters at three points in the selection history. 1. The released population. 2. Survivors. 3. Mated survivors. 1985 Released Survived Mated Males n= 580 n= 40 n=7 X SD X SD X SD Date 22.3 7.6 19.8 7.4 22.6 3.3 Wing 20.7 0.8 20.9 0.7 20.4 0.8 Abdomen 25.2 1.1 25.5 0.9 24.4 1.1 Mass 27.1 3.9 28.6 3.8 25.4 4.9 Released Survived Mated Females n= 565 n= :26 n=16 X SD X SD X SD Date 22.6 7.5 20.4 8.1 21.3 7.9 Wing 21.5 0.9 21.3 0.7 21.5 0.8 Abdomen 25.2 1.2 25.0 1.0 25.3 1.2 Mass 28.1 4.4 27.5 3.2 28.4 3.5 1986 Released Survived Mated Males n= 878 n= =31 n=16 X SD X SD X SD Date 38.0 17.0 42.3 18.4 51.3 10.9 Wing 20.1 1.0 20.0 1.0 20.0 0.8 Abdomen 24.6 1.2 24.4 1.2 24.4 1.3 Mass 23.7 4.8 22.9 4.4 22.6 3.1 Released Survived Mated Females n= 933 n= 14 n=9 X SD X SD X SD Date 38.3 17.6 48.2 17.0 50.8 14.5 Wing 20.8 1.1 21.0 1.1 20.8 0.7 Abdomen 24.4 1.3 24.5 1.2 24.4 1.3 Mass 24.6 5.2 24.7 4.2 23.7 3.0 47 Table V Proportion of variance in phenotypic characters accounted for by the first two principal components. 1985 Male Female P C I 0.668 0.669 P C 2 0.183 0.188 Total 0.851 0.857 1986 Male Female 0.776 0.779 0.113 0.118 0.889 0.897 48 Figure 9—Loadings of the four phenotypic characters on the first two principal components in 1985 and 1986. Characters are denoted by the first letter and year by the digits, d—date of emergence, w—forewing length, a—abdomen length, m— mass at emergence. 49 CM O Q_ CD • O CM • O CM • o I d85 males d86 -a85 w 8 i 8 6 - w86 m86 i , , ..; J _ . m85 -1.0 -0.5 0.0 PC1 0.5 1.0 CM O Q _ CO o CVJ • o CM • o d85 d86 -1.0 females -0.5 o.o PC1 0.5 1.0 50 Selection analysis Selection was treated as having occurred in two episodes. The first episode was the pre-reproductive period from the time of emergence to reproductive maturity (a minimum of four days). Damselflies were either resighted or not; these states were scored as one for survival and zero for dead/disappeared. The second episode was the number of mates acquired by those surviving to reproductive maturity. Individuals in tandem or in copula were deemed to have acquired a mate. While a male in tandem might be displaced by a competitor, I have never seen this occur although I observed hundreds of tandems. Females mated only once per visit to the pond and visits were separated by at least three days, so the number of mates was a reasonable measure of fitness. Tandems lasted between 30 and 90 minutes, so I am confident that I observed all matings at the pond that included marked females. Fitness was defined as relative fitness (absolute fitness divided by the population's mean absolute fitness). The opportunity for selection is the variance in relative fitness. It represents the maximum possible displacement of the mean phenotype by selection. Thus, if nearly all animals survive, a phenotypic character will not be much changed after selection. If only one individual survives then the potential difference after selection is at a maximum. The total opportunity for selection is the variance in the relative fitness as measured by number of mates regardless of whether individuals survived to sexual maturity or not. The two episodes do not sum to the total opportunity for selection because the two components of fitness (survival to sexual maturity and number of mates after sexual maturity is attained) must be multiplied to get lifetime fitness rather than summed. Directional selection differentials (S )were computed as the covariance between relative fitness and phenotype using principal components standardized to a mean of zero and a standard deviation of one. Selection differentials defined in this way are identical to selection intensity (i ) (Hartl 1980; Falconer 1981) and allow com-parisons of selection in units of standard deviation among characters with divergent 51 scales of measurement. Curvature selection on the phenotypic traits (c ) was tested by regressing the squared value of the standardized phenotypic trait and relative fitness as suggested by Lande and Arnold (1983). This procedure tests for non-linearities including the presence of stabilizing or disruptive selection in the rela-tionship between the phenotype and fitness (Mitchell-Olds and Shaw 1987; Schluter 1988). The relationship between fitness and morphology may be complex with dips, peaks or multiple peaks (eg. Schluter and Grant 1984). Linear or quadratic re-gression may then give misleading estimates of the relationship between fitness and the phenotype (Schluter 1988). A non-parametric estimate of the fitness function can be made using cubic polynomials joined seamlessly at the phenotypic values (splines). The best estimate of the fitness function given the data will fit all of the data points exactly but will have low predictive value for new data. The best value of the smoothing parameter is chosen so as to maximize the predictive power of the spline. This is achieved through the procedure of generalized cross-validation (Craven and Wahba 1979; Schluter 1988) outlined below. Splines of a given smoothness are calculated with one data point missing. The difference between the spline calculated without the data point and the value of the data point is the prediction error. This procedure is repeated for every data point and the sum of the squared prediction errors is calculated. The smoothing parameter is chosen so as to minimize this sum. 52 RESULTS Opportunity for selection Reduced survival noted in 1986 compared to 1985, and in females compared to males resulted in higher variance in relative fitness (Table VI). Variance in relative mating success was approximately equal to mean fitness for females in both years and for males in 1986. Substantially greater variance in mating success was observed for males in 1985 when a smaller proportion of males were mated. The total opportunity for selection was greater for males than females in 1985 but the reverse was true in 1986. Selection for survival to reproductive maturity 1985 Males—The forty males recaptured at sexual maturity were not a random subset of the 580 released (Figure 10). Large males were more likely to survive to sexual ma-turity than small males. Only two individuals with PCI scores less than -1 survived to sexual maturity. The mean PCI score of the population increased 0.35 standard deviations after selection (Table VII). Regression of the principal component scores against relative fitness showed that survival was significantly associated with PCI (t = 2.3; p < 0.03) but not with PC2 (t = 1.1; p > 0.3). Females—There was almost no difference in the mean size of survivors and the initial population. However, both the smallest and the largest individuals released were never recaptured. The reduction in variance was significant (c = —0.29;t = 2.04; p < 0.05) (Figure 11, Table VII). Females that emerged early for their size were more likely to survive than those that emerged late. Survival was distinctly bimodal with 7 of the 17 earliest emerging females surviving and a second group clustered around the mean. Not one of the latest emergers was recaptured. Females that survived to reproductive maturity had a mean PC2 score 0.42 standard deviations 53 Table VI—Opportunity for selection (variance in relative individual fitness) for males and females in 1985 and 1986 for both episodes of selection and the entire adult lifespan. 1985 Males Survival 13.5 Matings 5.6 Total 92.2 Females Survival 20.8 Matings 1.0 Total 41.1 1986 Males Survival 27.4 Matings 1.1 Total 56.8 Females Survival 65.7 Matings 0.8 Total 114.8 Figure 10—The distribution of principal component scores of the initial population (open histogram) and that of the surviving population (shaded histogram) of males in 1985. Solid vertical line is the mean of the initial population and the dashed vertical line is the mean of the survivors. The difference between the two lines is the selection differential. The two distributions are drawn to the same scale. See Table IV for sample sizes. A cross-validated spline of the probability of survival for a given phenotype is also presented. 55 less than the mean of females that were released (t = 2.2; p < 0.03), and the variance of the survivors was significantly increased (c = 0.27;p < 0.2). Selection on PC2 was significantly stronger for females than males (F = 4.6; p < 0.04) while selection on PCI was not significantly stronger for males than females (F = 0.3;p < 0.59). 1986 Males—The mean mass of emerging males was 3.4 mg less in 1986 than 1985 (Table IV). Nevertheless, selection for large body size was significantly stronger in 1985 than in 1986 (F = 8.1;p < 0.005). Indeed smaller males were more likely to survive than large ones in 1986, although this trend was not significant (t = 1.4; p < 0.69; Figure 12, Table VII). Selection on PC2 was also in the opposite direction from 1985. Late emerging males were more likely to survive to sexual maturity than early emerging males (S = 0.48; t = 2.7; p < 0.01). The slopes of the regressions of survival against PC2 were significantly different in the two years (F = 5.9; p < 0.02). There was no significant curvature selection on either PCI or PC2. Females—As in males, late emerging females were more likely to survive than early emergers (t — 3.1; p < 0.005; Figure 13, Table VII). Only two females with negative PC2 scores were recaptured at sexual maturity. Selection for early emergence in females in 1985 was significantly different from the selection for late emergence in 1986 (F = 8.3; p < 0.005). Although the estimate of curvature selection for PC2 was greater in 1986 than 1985, smaller sample sizes placed the probability in the neighbourhood of a (Table IX). As in 1985, there was no relationship between the probability of survival and body size of females. There were also no significant changes in the variability of PCI. The estimates of selection intensity in males and females were not significantly different. 57 Figure 11—The distribution of principal component scores of the initial population (open histogram) and that of the surviving population (shaded histogram) of females in 1985. Solid vertical line is the mean of the initial population and the dashed vertical line is the mean of the survivors. The difference between the two lines is the selection differential. See Table IV for sample sizes. A cross-validated spline of the probability of survival for a given phenotype is also presented. 58 Figure 12—The distribution of principal component scores of the initial population (open histogram) and that of the surviving population (shaded histogram) of males in 1986. Solid vertical line is the mean of the initial population and the dashed vertical line is the mean of the survivors. The difference between the two lines is the selection differential. The two distributions are drawn to the same scale. See Table IV for sample sizes. A cross-validated spline of the probability of survival for a given phenotype is also presented. 60 Figure 13—The distribution of principal component scores of the initial population (open histogram) and that of the surviving population (shaded histogram) of females in 1986. Solid vertical line is the mean of the initial population and the dashed vertical line is the mean of the survivors. The difference between the two lines is the selection differential. The two distributions are drawn to the same scale. See Table IV for sample sizes. A cross-validated spline of the probability of survival for a given phenotype is also presented. 62 Table VII—Selection differentials of phenotypic characters and their first two prin-cipal components for survival from emergence to reproductive maturity. Differen-tials significant at a < 0.05 are in bold face. Males P C I PC2 Females PCI PC2 1985 S ±SE 0.35 0.15 -0.16 0.15 -0.08 0.19 -0.42 0.19 Males PCI PC2 Females PCI P C 2 1986 S ±SE -0.24 0.18 0.48 0.18 -0.08 0.27 0.80 0.26 64 Mating Success 1985 Males—There were only nine matings observed in total by seven males. While larger males may have been more likely to survive to reproductive maturity, of those males that did survive, small males were more likely to acquire mates (Figure 14). Only a single mated individual had a positive PCI score. The mean PCI score was 1.03 standard deviations smaller after selection (Table VIII), and a regression of PCI against relative fitness was significant (t = 3.1; p < 0.005). There were no matings by animals emerging earliest or latest for their size and there was no significant directional or curvature selection for PC2 (S =• —0.24; t = 0.6; p > 0.5)(c = -0.34; t = 1.3; p > 0.2). Females—Few of the resighted females occurred without mates, and mated females were drawn from the entire size range of survivors (Figure 15). The mean PCI score was slightly but not significantly greater after selection. Selection on body size was significantly different between the sexes (F = 8.3; p < 0.006; Table VIII). The mean PC2 score was slightly but not significantly greater after selection. Although opposite in sign, there was no significant difference between the sexes for selection on PC2. 1986 Males—Although fewer males were recaptured in 1986, a larger proportion were mated (Figure 16). Unlike in 1985, smaller males did not get significantly more matings than large males (Table VIII). The largest and smallest males were not mated and estimates of curvature selection approached significance for selection on PCI (Table IX). However, surviving males were nearly 3.4 mg lighter than in 1985. Combining males from both years, a regression of mass against relative fitness was significant (t = 3.97; p < 0.005; Figure 17). The earliest and latest emerging males both acquired mates in 1986. There was no significant directional or curvature 65 Figure 14—The relationship between principal component scores and the number of mates acquired by males in 1985. Animals that survived to sexual maturity are represented by the open histogram. Animals that mated at least once are repre-sented by the shaded histogram. The phenotype of an individual that mated twice is denoted with a •. The solid vertical line is the mean principal component score of the whole population. The dashed vertical line is the mean principal component score of all the matings observed. The difference between the two lines is the selec-tion differential. Principal component scores have been restandardized to a mean of zero and a standard deviation of one between selection episodes. Raw phenotypic measurements are presented in Table IV. A cross-validated spline of the number of mates for a given phenotype is also presented. 66 Figure 15—The relationship between principal component scores and the number of mates acquired by females in 1985. Animals that survived to sexual maturity are represented by the open histogram. Animals that mated at least once are represented by the shaded histogram. The phenotype of an individual that mated twice is denoted with a •. The phenotype of an individual that mated three times is denoted with a ••. The solid vertical line is the mean principal component score of the whole population. The dashed vertical line is the mean principal component score of all the matings observed. The difference between the two lines is the selection differential. Principal component scores have been restandardized to a mean of zero and a standard deviation of one between selection episodes. Raw phenotypic measurements are presented in Table IV. A cross-validated spline of the number of mates for a given phenotype is also presented. 68 selection on PC2 (Tables VIII and IX). Nor was there a difference between 1985 and 1986 for the relationship between reproductive success and PC2 (F = 2.5; P < 0.2) Females—Few reproductively mature females were recaptured in 1986 so inferences on the relationship between phenotype and relative fitness are difficult to draw. There was no significant linear component to selection on either PCI or PC2 (Table VIII) . Successful females tended to have intermediate PCI and PC2 scores (Table IX) . Estimates of curvature selection were near a = 0.05 and cross-validated splines had distinct maxima at intermediate values (Figure 18). DISCUSSION The use of experimentally manipulated phenotypes is a potentially powerful tool to explore natural selection. The hypothesis that animals with phenotypes beyond the range normally observed are somehow less fit can be tested directly by measuring the survival and reproductive success of individuals with extreme phenotypes. In addition, expanding the range of phenotypes exposed to selection increases the power of analyses based on regression methods (Draper and Smith 1981; Schluter 1988). Male and female E. boreale vary the timing of emergence and mass at emergence in the same ways over a wide range of resource availability (Chapter Two). However, the consequences of emerging at a small or large size, early or late in the season are not the same for males and females in spite of identical responses to larval conditions. Further, similar responses to larval conditions in the two years also had quite different effects because of differences in weather. Differences Between Males and Females Female survival to reproductive maturity is independent of size at emergence and the number of reproductive episodes is also independent of size at emergence. The number of eggs produced might be a function of size but eggs are matured after 70 Figure 16—The relationship between principal component scores and the number of mates acquired by males in 1986. Animals that survived to sexual maturity.are represented by the open histogram. Animals that mated at least once are repre-sented by the shaded histogram. The phenotype of an individual that mated twice is denoted with a •. The solid vertical line is the mean principal component score of the whole population. The dashed vertical line is the mean principal component score of all the matings observed. The difference between the two lines is the selec-tion differential. Principal component scores have been restandardized to a mean of zero and a standard deviation of one between selection episodes. Raw phenotypic measurements are presented in Table IV. A cross-validated spline of the number of mates for a given phenotype is also presented. 71 Figure 17—Number of matings acquired by males of different masses in 1985 (+) and 1986 (o). A locally weighted moving average (lowess [Cleveland 1979] as implemented in S [Becker and Chambers 1984]) shows the expected number of mates for a given phenotype. Data at the center of the moving window are more heavily weighted than those at the edges. 73 Figure 18—The relationship between principal component scores and the number of mates acquired by females in 1986. Animals that survived to sexual maturity are represented by the open histogram. Animals that mated at least once are repre-sented by the shaded histogram. The phenotype of an individual that mated twice is denoted with a •. The solid vertical line is the mean principal component score of the whole population. The dashed vertical line is the mean principal component score of all the matings observed. The difference between the two lines is the selec-tion differential. Principal component scores have been restandardized to a mean of zero and a standard deviation of one between selection episodes. Raw phenotypic measurements are presented in Table IV. A cross-validated spline of the number of mates for a given phenotype is also presented. 75 females 1986 -2 -1 0 PC1 0.8 CO CD CTJ 0.6 E « • o 0.4 CD E 0.2 c 0.8 0) CO 0.6 E * • o 0.4 CD -Q E 0.2 c 76 Table VIII—Selection differentials of the first two principal components of the phenotypic characters for number of mates acquired. Principal components have been restandardized to mean of zero and a standard deviation of one. Differentials significant at a < 0.05 are presented in boldface. 1985 S ±SE Males P C I -1.03 0.34 PC2 -0.24 0.38 1986 S ±SE Males PCI -0.14 0.19 PC2 0.28 0.19 Females PCI 0.29 0.19 Females PCI -0.21 0.25 PC2 0.23 0.20 PC2 0.02 0.26 77 Table IX—Selection coefficients of the variance of PCI and PC2. Positive selec-tion coefficients indicate selection for extreme values (disruptive selection) while negative values indicate selection for mean values (stabilizing selection). Estimates are based on the covariance of fitness and the squared principal component scores. Differentials significant at a < 0.05 are presented in boldface. Males PCI PC2 S U R V I V A L 1985 c ±SE -0.09 0.11 0.14 0.09 1986 c ±SE Males PCI 0.03 0.11 PC2 0.14 0.11 Females P C I -0.29 0.14 Females PCI -0.09 0.18 PC2 0.27 0.11 PC2 0.29 0.16 Males PCI PC2 M A T I N G S U C C E S S 1985 c ±SE 0.45 0.22 -0.34 0.26 1986 c ±SE Males PCI -0.26 0.13 PC2 0.00 0.17 Females PCI PC2 -0.04 0.19 Females PCI -0.46 0.24 -0.08 0.15 PC2 -0.26 0.12 78 emergence (Corbet 1962). Resource acquisition for the production of eggs, as mea-sured by mass gain, was also independent of size at emergence (Figure 7, Chapter Three). In the range observed, adult female fitness was independent of larval con-ditions affecting female size. Adult female fitness was affected by changes in date of emergence, but not as a simple function of larval conditions. Early emergers were more likely to survive in 1985, but less likely to survive in 1986. Male survival was affected by size at emergence. Large males survived signifi-cantly better than small males in 1985. Small males also increased their mass while large males lost mass prior to sexual maturity. Thus, selection may not be acting directly on size, but on foraging behaviour which is correlated with size. Indepen-dent measures of foraging behaviour for the original population would be required to disentangle these two factors. The advantage of increased mass for females is probably the increased fecun-dity that can be achieved. Increased mass may be of some advantage to males when holding a territory is a prerequisite to acquiring mates, but in the absence of terri-toriality, large size may be a disadvantage. Small males may be more maneuverable than large males as suggested by Alexander et al. (1978) and observed for midges (McLachlan 1986) and therefore more likely to capture a female when she arrives at the mating site. Alternatively, small males may have lower flight costs than large males and thus spend more time at the pond before they run out of energy (Weis-Fogh 1977). If this is true, then the mating success of individual males should be better correlated with the total number of minutes spent at the pond than with the number of separate days seen at the pond or the reproductive lifetime of the individual. Some lower limit on male size must exist. Males may need to attain some minimum size to successfully metamorphose, or be large enough to form a tandem with a female. 79 Year to year variation When the weather was warm and dry in 1985, large males were more likely to survive to reproductive maturity than small males, but selection on body size was not apparent in 1986 (even though males were smaller in 1986) when weather was much colder and wetter. Overall survival was lower in 1986 and may have swamped selective mortality. Poor weather may also have acted to reduce selective mortalities indirectly. Mass gain was lower for both males and females in 1986. Therefore mortality correlated with this mass gain may also have been reduced. Selection may have been acting on PCI but my sample sizes were too small to detect it. Statistical significance of a difference of the magnitude observed given the observed variance would require increases in sample sizes on the order of three-fold on top of the nearly two-fold increase in sample size compared to 1985. Sample size requirements for females would be considerably larger because of the lower survival rate of females. Selection on the timing of emergence also varied between years. Females that emerged relatively early in 1985 had higher survival but late emergence was favoured for both males and females in 1986. Varying directional selection suggests that there is some long term intermediate optimum date for reproduction. If there is a long term optimum, why does improved nutrition lead to earlier emergence in E. boreale (Chapter Two)? A correlation between nutritional status and early breeding also exists for birds (Perrins 1965, 1970; Boyce and Perrins 1987; Arcese and Smith 1988). Such a pattern can persist in spite of stabilising selection for date of nesting (Fisher 1958; Price et al. 1988). The variability in selection differentials between years argues for caution on the interpretation of selection estimates based on a single season or generation. Where measurements of natural selection have been made over several years, considerable variability, as reported here, has been shown to exist, but changes in sign are usually not observed (Schluter and Smith 1986; Fincke 1988; Gibbs 1988). 80 The influence of dispersal Large damselflies were more likely to disperse than small ones in 1985. Estimates of selection on PCI in 1985 are therefore probably conservative, and not the result of differential dispersal by small damselflies. Lower survival by females is also unlikely to be the result of differential dispersal. Only two of the ten dispersing animals were female. The lower intensity of sampling where dispersing animals were recovered biases the recoveries towards males but the bias would have to be very large indeed to account for both the difference in female survival and the difference in recoveries. If the same pattern of dispersal was present in 1986, the estimated selection differential would be too low. Variance in lifetime reproductive success The contribution of survival to lifetime reproductive success can confound estimates of the relative strength of natural and sexual selection (Williams 1966, Otte 1979; Banks and Thompson 1985b; Koenig and Albano 1986). Low pre-reproductive survival coincident with mass gain used for egg production in female odonates con-founds sexual selection with natural selection. Estimating the variance in lifetime re-productive success on the basis of the reproductive period alone will under-estimate the variance in female reproductive success because it ignores mortality incurred to acquire resources essential to reproductive success. In territorial species where males gain mass the same arguments will apply if increased mass serves to make them more able to hold territories. The difficulty is in estimating the amount of mortality attributable to mass gain. If we include all the variance associated with survival it is no longer clear that males have a higher variance in reproductive suc-cess than females. Variance in lifetime mating success was higher for males in 1985 but higher for females in 1986. We need to include the entire adult portion of the life-cycle for a more complete understanding of the sources of variation in reproductive success. This is true 81 whenever resource acquisition influences survival and reproductive success of adults. The use of phenotypically manipulated individuals can increase the power of these studies and may provide insights unavailable to other approaches. 82 Chapter Five GENERAL DISCUSSION This thesis makes original contributions in four areas of damselfly population bi-ology. 1) I experimentally separated the effects of interference and exploitative competition for late stage damselfly larvae and demonstrated that interference is much weaker than previously thought. 2) The consequences of larval conditions to adult survival and reproductive success were measured by estimating the intensity of selection on the manipulated phenotypes of the adults. 3) I shed some light on the poorly known pre-reproductive period after emergence and prior to sexual maturity. 4) Data on adult survival and reproductive success and data on survival to reproductive maturity and mass gain suggested a general hypothesis about the significance of pre-reproductive mass gain in odonate mating systems. Separating interference and exploitative competition The energetic cost of agonistic behaviour (Hamilton 1964) has been thought to be sufficiently high to influence its expression (Maynard Smith and Parker 1976; Parker and Thompson 1980). It is not surprising then, that interference competition has been implicated in odonate population processes where agonistic behaviour is well developed (Rowe 1980; Baker 1981, 1983). Density-dependent reduction in growth and development in odonates (Johnson et al. 1984, 1985; Pierce et al. 1985; Van Buskirk 1987) has usually been attributed to reduced feeding in the presence of conspecifics (Uttley 1980; McPeek and Crowley 1987). Johnson et al. (1984, 1985) and Pierce et al. (1985) deduced that density effects in odonates must be due to interference because they were unable to detect a density-dependent reduction in prey abundance. Because of the low statistical power of their experiments Johnson et al. (1987) reanalyzed the data from all of the experiments in Bays Mountain Lake and found that prey depletion did occur in 83 several taxa, but that the level of depletion dependend on the presence or absence of odonates but not on their density. Van Buskirk (1987) concluded that while food supply had an effect, density effects were probably the result of interference among larvae because he was unable to detect a difference in the response of low density versus high density treatments to food addition. Van Buskirk did not discuss the power of his experiment to detect the expected interaction effect. The conclusion that interference competition must be the source of density-dependence when it has not been possible to demonstrate the effect of exploitative competition, should be treated with care. A more powerful test would be to ma-nipulate interference and demonstrate its effect. Uttley (1980) and Crowley et al. (1987) have both suggested manipulating habitat complexity to reduce encounter rates as a way to accomplish this. I have conducted these experiments and I found that interference competition did not contribute to the density-dependent reduction in growth and development of E.boreale in Thesis Pond. Reconciling these two points of view will, of course, require further experiments in a wider range of localities. Before considering alternative approaches, I would like to discuss what conditions would lead us to make opposite conclusions. Pro-ductivity is low in Thesis Pond and this could exaggerate the effect of food supply relative to interference competition. It may be that when food is less limiting, in-terference competition has a larger effect. It is also possible that our experiments are measuring the wrong variable. For instance, prey density was not affected in the experiments reanalyzed by Johnson et al. (1987) but prey availability may have changed. Just as damselflies alter their behaviour in the presence of predators (McPeek and Crowley 1987) so too may their prey. If the level of predator avoid-ance is density-dependent then damselfly growth and development rates may also be density-dependent in the absence of prey depletion. Deciding whether this is exploitative or interference competition quickly becomes an exercise in semantics. 84 Because these were enclosure experiments, the results may be artifacts of enclos-ing the system. Whole pond experiments, where productivity is altered or habitat complexity is manipulated are an obvious, if difficult, next step. Alternative approaches might be to measure the metabolic cost of the agonistic behaviour or measure an effect that can only be attributed to interference. Measure-ments of size-specific food limitation and dispersal (Baker 1985, 1986) in damselflies have also been unable to demonstrate interference effects. Actual measurements of the metabolic costs of interference are rare. Riechert (1988) has demonstrated that the cost of fighting in territorial disputes of the spider Agelenopsis aperta was tiny. Rasmussen (1983) concluded that there were significant metabolic costs of interfer-ence in larval Chironomidae. In general, it is difficult to study experimentally the effect of behaviour on pop-ulation dynamics because it is difficult to manipulate the expression of behaviour. Behaviour can be modified with the use of hormones (Searcy and Wingfield 1980) or drugs but care must be taken in the choice of an appropriate control. Otherwise it becomes an exercise in measuring the effect of a hormone or drug on population dynamics. Selective breeding for the expression of behaviour is also possible, but correlated responses to selection may also cloud interpretation of results from ex-periments using such animals. However, manipulations are possible. Attributing an effect to behaviour which has not been manipulated or directly tested when a manipulated factor has no effect, is an asymetry in the burden of proof that should and can be rectified. Measurement of selection on adults Regulation of odonate populations is likely to occur in the larval portion of the life-cycle because the larval habitat is restricted and the adults are relatively short-lived (Wilbur 1980; Van Buskirk 1987). It is generally assumed that density-dependent changes in the size at emergence and the timing of emergence reduce the fitness 85 of the adults. Results are not clear-cut. Large males may have higher mating success (Harvey and Corbet 1985) or they may not (Van Buskirk 1987; Koenig and Albano 1987; Fincke 1987). Large females may have a slight advantage in one species but this is because they live longer, not because they lay more eggs (Banks and Thompson 1987). In general, a fecundity advantage of large female odonates is assumed to exist because it exists in other insects. Mass gain of immature females suggests that there is little effect of skeletal size on fecundity, although there must be some lower limit. Grafen (1988) has made a distinction between investigating the adaptations of animals by experiment and investigating ongoing natural selection by regression methods (Lande and Arnold 1983). This seems to presume that existing phenotypes cannot be experimentally manipulated. As long as phenotypes respond to environ-mental variation or there is additive genetic variance, manipulation is possible. If there are no environmental effects on the phenotype or no additive variance with which to do selective breeding to manipulate phenotypes then alternative methods of investigating adaptation will have to be used. The pre-reproductive period and odonate mating systems In spite of their well-known feeding behaviour (Corbet 1962, 1980) mass gain in adult odonates has been all but ignored (Jenkins 1981 is the single exception that I am aware of). Wilbur (1980) has viewed the larval stage of anurans as an adaptation to take advantage of the ephemeral resources of ponds. Considering the rapid mass gain of immature odonates (about 30 percent of body mass is gained in less than 2 percent of the larval lifespan) this phase could also be viewed as an adaptation to take advantage of ephemeral resources. Mass gain is achieved at the cost of increased mortality. For females, mass gain can be used for the production of eggs the fitness value of which is similar across mating systems. For males, the value of mass gain will depend on the mating system and perhaps on the density of other 86 males. In non-territorial systems such as E. boreale mass gain may only be useful to provide sufficient energy to allow the male to spend more time in sexual activity. In territorial mating systems, mass gain may increase the competitiveness of the male in contests for mating territories. The value of this may increase with male density so that when males emerge from dense larval conditions they will gain more mass than in years or from ponds with lower densities. Similar conditions are likely to apply to anurans with contrasting mating sys-tems (Howard 1988). In male bullfrogs (Rana catesbeiana) the probability of mating increases with age and size and the sex-ratio is near 1. However, the mating success of male woodfrogs (Rana sylvatica) does not increase after the first year and the sex ratio favours males by up to 4:1. In the latter case males may be foregoing increased mass to reduce mortality risks. Females accept higher risks because increased mass increases fecundity. The increased mortality associated with mass gain can be thought of as a cost of sexual selection. This cost has been neglected when considering aspects of mating systems such as sexual dimorphism and sex-ratio. The increase in fitness due to increased mass and the costs due to risks incurred by foraging can be measured and used to test hypotheses about the outcome of sexual selection. Hypotheses that have so far resisted investigation. 87 L I T E R A T U R E C I T E D Akre, B.G. and D.M. Johnson. 1979. Switching and sigmoid functional response curves by damselfly naiads with alternative prey available. Journal of Animal Ecology 48:703-720. Alexander, R.D., J.L. Hoogland, R.D. Howard, and P.W. Sherman. 1978. Sexual selection and breeding systems in pinnipeds, ungulates, primates and humans. IN: Evolution-ary biology and human social behavior. Chagnon, N.A. and W.G. Irons (editors). Duxbury Press. North Sciutate Massachusetts, pp. 402-435. Arcese, P. and J.N.M. Smith. 1988. Effects of population density and supplemental food on reproduction in song sparrows. Journal of Animal Ecology 57:119-136. Arnold, S.J. and M.J. Wade. 1984a. On the measurement of natural and sexual selection: theory. Evolution 38:709-719. Arnold, S.J. and M.J. Wade. 1984b. On the measurement of natural and sexual selection: applications. Evolution 38:720-734. Baker, R.L. 1980. Use of space in relation to feeding areas by zygopteran nymphs in captivity. Canadian Journal of Zoology 58:1060-1065. Baker, R.L. 1981. Behavioural interactions and use of feeding areas by nymphs of Coena-grion resolutum (Coenagrionidae:Odonata). Oecologia 49:353-358. Baker, R.L. 1983. Spacing behaviour by larval Ischnura cervula Selys: effects of hunger pre-vious interactions, and familiarity with an area (Zygoptera: Coenagrionidae). Odona-tologica 12:201-207. Baker, R.L. 1986. Food limitation and population density of larval dragonflies: a field test of spacing behaviour. Canadian Journal of Fisheries and Aquatic Sciences 43:1720-1725. Baker, R.L. 1987. Dispersal of larval damselflies: do larvae exhibit spacing behaviour in the field? Journal of the North American Benthological Society 6:35-45. Banks, M.J. and D.J. Thompson. 1985a. Emergence, longevity, and breeding area fidelity in Coenagrion puella (L.) (Zygoptera: Coenagrionidae). Odonatologica 14:279-286. Banks, M.J. and D.J. Thompson. 1985b. Lifetime mating success in the damselfly Coenagrion puella. Animal Behaviour 33:1175-1183. Banks, M.J. and D.J. Thompson. 1987. Lifetime reproductive success of females of the damselfly Coenagrion puella. Journal of Animal Ecology 56:815-832. 88 Becker, R.A. and J.M. Chambers. 1984. S: an interactive environment for data analysis and graphics. The Wadsworth statistics and probability series. Belmont CA. 550 pp. Bick, G.H. 1972. A review of territorial and reproductive behaviour in Zygoptera. Con-tactbrief Nederlandse Libellenonderzockers Supplementum 10:1-14. Bick, G.H. and J.C. Bick. 1961. An adult population of Lestes disjunctus australis Walker. Southwestern Naturalist 6:111-137. Bick, G.H. and J.C. Bick. 1963. Behavior and population structure of the damselfly Enallagma civile (Hagen) (Odonata: Coenagrionidae). Southwestern Naturalist 8:57-84. Bick, G.H. and J.C. Bick. 1965. Demography and behavior of the damselfly Argia apicalis (Say). Ecology 46:461-472. Bookstein, F., B. Chernoff, R. Elder, J. Humphries, G. Smith, and R. Strauss. 1985. Morphometries in evolutionary biology. Special Publications 15. Academy of Natural Sciences of Philadelphia. Philadelphia. Boyce, M.S. and C. Perrins. 1987. Optimizing great tit clutch size in a fluctuating envi-ronment. Ecology 68:142-153. Buchholtz, C. 1951. Untersuchungen an der Libellen-Gattung Calopteryx—Leach unter besonderer Berucksichtigung ethologischer Fragen. Zeitschrift fur Tierpsychologie 8:273-293. Charnov, E.L., G.H. Oiians, and K. Hyatt. 1976. Ecological implications of resource depression. The American Naturalist 110:247-259. Cleveland, W.S. 1979. Robust locally weighted regression and smoothing scatter plots. Journal of the American Statistical Association 74:829-836. Convey, P. 1988. Competition for perches between larval damselflies: the influence of perch use on feeding efficiency, growth rate and predator avoidance. Freshwater Biology 19:15-28. Corbet, P.S. 1962. A biology of dragonflies. Witherby, London. Corbet, P.S. 1980. Biology of Odonata. Annual Review of Entomology 25:189-217. Corbet, P.S., C. Longfield, and N.W. Moore 1960. Dragonflies. Collins. London. Craven, P. and G. Wahba. 1979. Smoothing noisy data with spline functions: Estimat-ing the correct degree of smoothing by the method of generalized cross-validation. Numerische Mathematik 31:377-403. 89 Crowley, P.H., R.M. Nisbet, W.S.C. Gurney, and J.H. Lawton 1987. Population regulation in animals with complex life-histories: formulation and analysis of a damselfly model. Advances in Ecological Research 17:1-59. Draper, N.L. and H. Smith. 1981. Applied regression analysis (2nd ed.). John Wiley and Sons. New York. Endler, J.A. 1986. Natural Selection in the Wild. Princeton University Press. Princeton. Falconer, D.S. 1981. Introduction to quantitative genetics, 2nd ed. Longman, London. Fincke, O.M. 1982. Lifetime mating success in a natural population of the damselfly Enallagma hageni (Walsh) (Odonata: Coenagrionidae). Behavioral Ecology and So-ciobiology 10:293-302. Fincke, O.M. 1984. Giant damselflies in a tropical forest: reproductive biology of Megaloprepus coerulatus with notes on Mecistogaster (Zygoptera: Pseudostigmatidae). Advances in Odonatology 2:13-27. Fincke, O.M. 1988. Sources of variation in lifetime reproductive success in a non-territorial damselfly (Odonata:Coenagrionidae) IN Reproductive Success. T .H. Clutton-Brock (editor) University of Chicago Press. Chicago, pp.24-43. Fisher, R.A. 1930. The genetical theory of natural selection. Oxford University Press. Oxford. Fisher, R.A. 1958. The genetical theory of natural selection. 2nd ed. Dover, New York. Folsom, T .C. and N.C. Collins. 1982. Food availability in nature for the larval dragonfly Anax Junius (Odonata: Aeshnidae). Freshwater Invertebrate Biology 1:33-40. Fretwell, S.D. 1972. Populations in a seasonal environment. Princeton University Press. Princeton N.J. Gibbs, H.L. 1988. Heritability and selection on clutch size in Darwin's medium ground finches (Geospiza fortis) Evolution 42:750-762. Gilliam, J.F. and D.F. Fraser. 1987. Habitat selection under predation hazard: test of a model with foraging minnows. Ecology 68:1856-1862. Gould, S.J. 1971. Geometric similarity in allometric growth: a contribution to the problem of scaling in the evolution of size. The American Naturalist 105:113-136. Gould, S.J. and R.C. Lewontin. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist program. Proceedings of the Royal Society of London B. 205:581-598. 90 Grafen, A. 1988. On the uses of data on lifetime reproductive success. IN Reproductive success. T .H. Clutton-Brock (editor). University of Chicago Press. Chicago, pp.454-471. Greenwood, P.J. 1980. Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour 28:1140-1162. Hafernik, J.E. Jr., and R.W. Garrison. 1986. Mating success and survival rate in a population of damselflies: results at variance with theory? The American Naturalist 128:353-365. Haldane, J.B.S. 1954. The measurement of natural selection. Proceedings of the IX International Congress of Genetics 1:480-487. Hamilton, W.D. 1964. The genetical evolution of social behaviour. II. Journal of Theoret-ical Biology 7:17-52. Hartl, D.L. 1980. Principles of population genetics. Sinauer, Sunderland. Harvey, I.F. and P.S. Corbet. 1985. Territorial behaviour of larvae enhances mating success of male dragonflies. Animal Behaviour 33:561-565. Howard, R.D. 1988. Reproductive success in two species of anuran. IN: Reproductive success. T .H. Clutton-Brock (editor). University of Chicago Press. Chicago, pp. 99-113. Istock, C.A. 1967. The evolution of complex life cycle phenomena: an ecological perspec-tive. Evolution 21:592-605. Jacobs, M.E. 1955. Studies on territorialism and sexual selection in dragonflies. Ecology 36:566-586. Jenkins, D.M. 1981. Weight measurements of odonata in relation to teneral growth. M.Sc. thesis. University of Louisville. Louisville, KY. 67 pp. Jensen, A.L. 1987. Simple models for exploitative and interference competition. Ecological Modelling 35:113-123. Johnson, D.M., R.E. Bohanan, C.N. Watson, and T.H. Watson. 1984. Coexistence of Enallagma divagans and Enallagma traviatum (Zygoptera:Coenagrionidae) in Bays Mountain Lake: an in situ enclosure experiment. Advances in Odonatology 2:57-70. Johnson, D.M., P.H. Crowley, R.E. Bohanan, and T.H. Martin. 1985. Competition among larval dragonflies: a field enclosure experiment with Tetragoneuria cynosura and Celithemis elisa (Odonata: Anisoptera). Ecology 66:119-128. 91 Johnson, D.M., C.L. Pierce, T .H. Martin, R.E. Bohanan, C.N. Watson, and P.H. Crow-ley. 1987. Prey depletion by odonate larvae: combining evidence from multiple field experiments. Ecology 68:1459-1465. Koenig, W.D. and S.S. Albano. 1986. On the measurement of sexual selection. The American Naturalist 127: 403-409. Koenig, W.D. and S.S. Albano. 1987. Lifetime reproductive success, and the opportunity for selection in the White-tailed skimmer Plathemis lydia (Odonata: Libellulidae). Evolution 41:22-36. Kot, M. and W.M. Schaffer. 1984. The effect of seasonality on discrete models of popula-tion growth. Theoretical Population Biology 26:340-360. Lande, R. and S.J. Arnold. 1983. The measurement of selection on correlated characters. Evolution 37:1210-1226. Lawton, J.H. 1971. Maximum and actual field feeding-rates in larvae of the damselfly Pyrrhosoma nymphula (Sulzer)(Odonata:Zygoptera). Freshwater Biology 1:99-111. Lawton, J.H., B.A. Thompson, and D.J. Thompson. 1980. The effects of prey density on survival and growth of damselfly larvae. Ecological Entomology 5:39-51. Macan, T .T . 1964. The Odonata of a moorland fishpond. Internationale Revue der gesamten Hydrobiologie 49:325-360. Macan, T .T . 1974. Twenty generations of Pyrrhosoma nymphula (Sulzer) and Enallagma cyathigerum (Charpentier) (Zygoptera:Coenagrionidae). Odonatologica 3:107-119. Maynard Smith, J. and G.A. Parker. 1976. The logic of asymetric contests. Animal Behaviour 24:159-175. McLachlan, A. 1986. Life-history tactics of rain-pool dwellers. Journal of Animal Ecology 52:545-561. Mcpeek, M A . and P.H. Crowley. 1987. The effects of density and relative size on the aggressive behaviour, movement, and feeding of damselfly larvae (Odonata: Coena-grionidae). Animal Behaviour 35:1051-1061. McVey, M.E. 1988. The opportunity for sexual selection in a territorial dragonfly Erythemis simplicicollis IN: Reproductive Success. T.H. Clutton-Brock (editor) University of Chicago Press. Chicago, pp.44-58. Mitchell, R. 1962. Storm-induced dispersal in the damselfly Ischnura verticalis (Say). The American Midland Naturalist 68:199-202. Mitchell-Olds, T. and R.G. Shaw. 1987. Regression analysis of natural selection: statisti-cal inference and biological interpretation. Evolution 41:1149-1161. 92 Moore, N.W. 1954. On the dispersal of Odonata. Proceedings of the Bristol Naturalists Society 28:407-417. Otte, D. 1979. Historical development of sexual selection theory. IN Sexual selection and reproductive competition in insects. Blum, M.S. and N.A. Blum (editors.) Academic Press. New York, pp 1-18. Pajunen, V.I. 1962. Studies on the population ecology of Leucorrhinia dubia v.d. Lind. Ann. Zool. Soc. Zool. Bot. Fenn. Vanamo 24:1-79. Parker, G.A. and E.A. Thompson. 1980. Dungfly struggles: a test of the war of attrition. Behavioral Ecology and Sociobiology 7:37^44. Parr, M.J. 1976. Some aspects of the population ecology of the damselfly Enallagma cyathigerum (Charpentier) (Zygoptera: Coenagrionidae). Odonatologica 5:45-57. Parr, M.J. and M. Palmer. 1971. The sex ratios, mating frequencies and mating expectan-cies of three coenagriids in northern England. Entomologica Scandinavica 2:191-204. Pearlstone, P.S.M. 1973. The food of damselfly larvae in Marion Lake, British Columbia. Syesis 6:33-39. Pearson, K. 1903. Mathematical contributions to the theory of evolution. XL On the influence of natural selection on the variability and correlation of organs. Philosophical Transactions of the Royal Society of London A. 200:1-66. Perrins, C M . 1965. Population fluctuations and clutch size in the great tit Parus major, L. Journal of Animal Ecology 34:601-647. Perrins, C M . 1970. The timing of birds' breeding seasons. Ibis 112:242-255. Pickup, J., D.J. Thompson, and J.H. Lawton. 1984. The life history of Lestes sponsa (Hansemann): larval growth (Zygoptera: Lestidae). Odonatologica 13:451-459. Pierce, C.L., P.H. Crowley, and D.M. Johnson. 1985. Behavior and ecological interactions of Odonata. Ecology 66:1504-1512. Price, T., M. Kirkpatrick, and S.J. Arnold. 1988. Directional selection and the evolution of breeding date in birds. Science 240:798-799. Prout, T. 1986. The delayed effect on adult fertility of preadult competition: two-species population dynamics. The American Naturalist 127:809-818. Prout, T. and F. McChesney. 1985. Competition among immatures affects their adult fertility: population dynamics. The American Naturalist 126:521-558. Rasmussen, J.B. 1983. An experimental study of competition and predation and their effects on growth and coexistence of chironomid larvae in a small pond. Ph.D. disser-tation. University of Calgary, Calgary, Canada. 93 Riechert, S.E. 1988. The energetic costs of fighting. American Zoologist 28:877-884. Rowe, R.J. 1980. Territorial behaviour of a larval dragonfly Xanthocnemis zealandica (MacLachlan) (Zygoptera:Coenagrionidae). Odonatologica 9:285-292. Schluter, D. 1988. Estimating the form of natural selection on a quantitative trait. Evo-lution 42:849-861. Schluter, D. and P.R. Grant. 1984. Determinants of morphological patterns in communi-ties of Darwin's finches. The American Naturalist 123:175-196. Schluter, D. and J.N.M. Smith. 1986. Natural selection on beak and body size in the song sparrow. Evolution 40:221-231. Schoener, T.W. 1973. Population growth regulated by intraspecific competition for energy or time: some simple representations. Theoretical Population Biology 4:56-84. Schoener, T.W. 1976. Alternatives to Lotka-Volterra competition: models of intermediate complexity. Theoretical Population Biology 10:309-333. Schoener, T.W. 1978. Effects of density restricted encounter on some single-level compe-tition models. Theoretical Population Biology 13:365-381. Searcy, W.A. and J.C. Wingfield. 1980. The effects of androgen and antiandrogen on dom-inance and aggressiveness in male red-winged blackbirds. Hormones and Behaviour 14:126-135. Semlitsch, R.D., D.E. Scott, and J.H.K. Pechmann 1988. Time and size at metamorphosis related to adult fitness in Ambystoma talpoideurn. Ecology 69:184-192. Sih, A. 1980. Optimal behavior: can foragers balance two conflicting demands? Science 210:1041-1043. Singer, M.C. 1982. Sexual selection for small size in male butterflies. The American Naturalist 119:440-443. Smith, D.C. 1983. Factors controlling tadpole populations of the chorus frog (Pseudacris triseriata) on Isle Royale, Michigan. Ecology 64:501-510. Smith, D.C. 1987. Adult recruitment in chorus frogs: effects of size and date at metamor-phosis. Ecology 68:344-350. Thompson, D.J. 1978. The natural prey of larvae of the damselfly Ischnura elegans (Odonata:Zygoptera). Freshwater Biology 8:377-384. Thompson, D.J. 1982. Prey density and survival in damselfly larvae; field and laboratory studies. Advances in Odonatology 1:267-280. 94 Thompson, D.J. 1987. Regulation of damselfly populations: the effects of weed density on larval mortality due to predation. Freshwater Biology 17:367-371. Tilman, D. 1987. The importance of the mechanisms of interspecific competition. The American Naturalist 129:769-774. Ubukata, H. 1974. Relative abundance and phenology of adult dragonflies at a dystrophic pond in Usubetsu, near Sapporo. Journal of the Faculty of Science, Hokkaido Univer-sity. Series 6 Zoology 19:812-833. Uttley, M.G. 1980. A laboratory study of mutual interference between freshwater inverte-brate predators. D. Phil dissertation, University of York. Van Buskirk, J. 1986. Establishment and organization of territories in the dragonfly Sympetrum rubicundulum (Odonata: Libellulidae). Animal Behaviour 34:1781-1790. Van Buskirk, J. 1987. Influence of size and date of emergence on male survival and mating success in a dragonfly, Sympetrum rubicundulum. The American Midland Naturalist 118:169-176. Weis-Fogh, T. 1977. Dimensional analysis of hovering flight. IN: Scale Effects in Animal Locomotion. T.J. Pedley (editor) Academic Press. New York. pp.405-420. Wilbur, H.M. 1980. Complex life cycles. Annual Review of Ecology and Systematics 11:67-93. Wilbur, H.M., P.J. Morin, R.J. Harris. 1983. Salamander predation and the structure of experimental communities: anuran responses. Ecology 64:1423-1429. Williams, G.C. 1966. Adaptation and Natural Selection. Princeton University Press. Princeton. Ydenberg, R.C. and A.I. Houston. 1986. Optimal trade-offs between competing be-havioural demands in the great tit. Animal Behaviour 34:1041-1050. 95 Appendix This appendix contains data on the mean mass (±s.d.) of male and female odonates from six families. The data are from three sources. Animals collected at the study pond are given in live mass. Data from the Spencer Entomological Museum are dry weights. Specimens were chosen so that they came from the same location at the same time. Data from Jenkins (1981) are all dry weights. teneral mass species M F Zygoptera Coenagrionidae Amphiagrion abbreviation Argia emma Coenagrion interrogatum Coenagrion resolutum Enallagma carunculatum Enallagma cyathigerum Enallagma ebrium Ischnura cervula Ischnura damula Lestidae Lestes congener Lestes disjunctus F mature mass M M F 7.9±1.5 n=3 13.9±2.6 n=5 6.0±0.5 n=5 7.8±1.3 n=6 8.9±1.0 n=3 11.8±1.0 n=13 8.7±0.5 n=3 6.8±0.5 n=2 6.1±1.0 n=3 14.5±2.3 n=3 26.1±4.0 n=5 9.1±0.8 n=5 10.5±1.3 n=5 14.6±1.4 n=3 17.0±4.6 n=13 15.6±2.5 n=3 10.2±1.3 n=4 6.4±2.1 n=4 M source 1.84 Spencer 1.88 Spencer 1.52 Spencer 1.35 Spencer 1.64 Spencer 1.44 Spencer 1.79 Spencer 1.50 Spencer 1.05 Spencer 12.6±1.4 18.9±2.8 1.50 Spencer n=8 n=8 15.1±2.7 18.8±6.8 1.91 Pond n=15 n=ll 96 teneral mass F mature mass F species M F M M F M source Anisoptera Aeshnidae Aeshna 146.2±6.4 145.1±54.4 0.99 Spencer californica n=4 - n=3 Aeshna 228.7±34.2 257.5±3.0 1.13 Spencer canadensis n—11 n=2 Aeshna 247.4±13.6 413.0±17.7 1.67 Spencer cyanea n=3 n=3 Aeshna 231.2±49.5 264.8±46.8 1.15 Spencer eremita n=4 n=7 Aeshna 225.8±35.5 247.5±47.7 1.10 Spencer interrupta n=15 n=ll Aeshna 204.7±3.2 231.8±5.6 1.13 Spencer juncea n=4 n=3 Aeshna 220.0±6.1 285.0±2.4 1.30 Spencer multicolor n—3 n=3 Aeshna 219.9±8.2 278.6±15.1 1.27 Spencer palmata n=3 n=3 Aeshna 171.2±9.5 217.5±46.9 1.27 Spencer septentrionalis n=4 n=4 Aeshna 127.2±13.5 193.1±17.2 1.52 Spencer sitchensis n=4 n=3 Aeshna 181.4±49.4 256.0±15.5 1.41 Spencer subarctica n=4 n=7 Anax 200± 31 208±37 1.04 368±58 331±28 0.90 Jenkins Junius n= 4 n— 3 n=18 n=2 Corduliidae Cordulia 68.0±4.3 72.7±16.8 1.07 Spencer shurtleffi n=4 n=3 Epitheca 96.3±17.8 109.4±6.2 1.14 Spencer canis n=2 n=2 Epitheca 74.2±19.9 114.2±40.4 1.54 Spencer spinigera n=3 n=3 Somatochlora 116.3±4.8 180.0±14.6 1.55 Spencer albicincta n=3 n=2 Somatochlora 143.3±15.9 189.0±22.6 1.32 Spencer metallica n=4 n=2 Somatochlora 84.3±3.8 107.6±2.1 1.28 Spencer whitehousei n=5 n=2 97 teneral mass F mature mass F species M F M M F M source Gomphidae Gomphus 3 3 ± 4 35±6 1.06 53±5 66±6 1.25 Jenkins exilis n= 9 n=3 n=7 n=2 Gomphus 56± 11 53±9 0.95 94±22 129±24 1.37 Jenkins graslinellus n= 3 n=5 n=8 n=5 Libellulidae Erythemis 37± 9 32±5 0.86 70±11 83±12 1.19 Jenkins simplicicoUis n= 9 n=10 n=29 n=4 Leucorhinia 75.5±8.3 90.6±13.2 1.20 Spencer borealis n=4 n=4 Leucorhinia 52.5±0.6 57.1±9.0 1.09 Spencer glacialis n=3 n=3 Leucorhinia 34.3±0.8 41.4±2.4 1.21 Spencer hudsonica n=4 n=4 Libellula 28± 7 34±3 1.21 92±10 109±19 1.18 Jenkins cyanea n= 8 n=5 n=28 n=3 Libellula 136.9±5.6 174.4±6.6 1.27 Spencer forensis n=3 n=2 Libellula 116.2±0.4 151.6±14.7 1.30 Spencer julia n=2 n=2 Libellula 41± 7 37±8 0.90 115±16 129±17 1.12 Jenkins luciuosa n= 14 n=7 n=52 n=8 Pachydiplax 19± 5 17±7 0.89 51±13 67±14 1.31 Jenkins longipennis n= 9 n=ll n=72 n=8 Plathemis 55± 4 51±5 0.93 125±19 105±36 0.84 Jenkins lydia n= 9 n=10 n=29 n=4 Sympetrum 29.8±8.7 36.8±16.1 1.23 Spencer danae n=4 n=3 Sympetrum 35.0±2.6 46.2±5.8 1.32 Spencer internum n=3 n=3 Sympetrum 70.8±3.3 79.2±6.4 1.12 Spencer madidum n=6 n=2 98 

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.831.1-0098136/manifest

Comment

Related Items