Printer Friendly

THE INFLUENCE OF AGE-SPECIFIC SURVIVORSHIP ON PUMPKINSEED SUNFISH LIFE HISTORIES.

INTRODUCTION

The questions of when to mature and how much effort to put into reproduction are central to life history theory. In considering when an organism should mature, it is assumed that, once a physiological minimum size of reproduction is attained, reproduction at the earliest opportunity would be favored by natural selection if reproductive costs were nonexistent (Partridge and Sibly 1991). Yet, many organisms do not mature at the earliest opportunity, which suggests there is a trade-off between current reproduction, future survival and future reproduction (Bell 1980, Roff 1984). In the case of an organism with indeterminate growth and size-dependent fecundity, delaying maturity leads to higher initial fecundity through larger body size (Charlesworth 1980, Stearns 1992), but this delay is also associated with an elevated risk of mortality prior to first reproduction because of a lengthened juvenile period (Williams 1966, Bell 1980, Stearns 1992). Delaying maturity can also increase the fitness of the off-spring produced by the parent if the delay results in the creation of higher quality offspring, or the enhancement of their survival potential through improved parental care by. the larger, older individual (Michod 1979, Stearns 1992).

Once an organism matures, there is the question of how much effort to put into reproduction in a given year. The trade-offs associated with reproductive effort are similar to those associated with age at maturity. For an organism living in an environment with limited food resources, a large amount of effort expended in current reproduction requires energy that might otherwise be available for growth or maintenance (Gadgil and Bossert 1970, Bell 1980, Kozlowski and Uchmanski 1987). If high reproductive effort in a given year is done at the expense of growth and/or maintenance, the potential for future reproduction could be reduced because of increased risk of mortality, decreased fecundity associated with a diminished growth rate, or both of these factors (Law 1979, Shine and Schwarzkopf 1992).

The potential importance of the age-specific mortality rate as a factor affecting the evolution of life histories has been emphasized in a number of theoretical models (reviewed in Stearns 1976). Gadgil and Bossert (1970) and Charnov and Schaffer (1973) emphasized the need to separate juvenile and adult mortality in determining optimal life histories. While several theorists focused on the relative survival rate of adults and juveniles as a factor in the evolution of iteroparity (e.g., Charnov and Schaffer 1973, Charlesworth 1980), this factor has also been used to predict age at maturity and reproductive effort. By modeling the trade-offs associated with reproduction, Gadgil and Bossert (1970) and Schaffer (1974) predicted that decreased reproductive effort and delayed maturity would occur when adult survival is increased relative to juvenile survival.

There have been few formal tests of the adult:juvenile survival predictions in natural systems, and tests of these predictions have been inconclusive. Hutchings (1993) tested these predictions with data from three Newfoundland brook trout (Salvelinus fontinalis) populations, and found, in accordance with theory, that reproductive effort declined with an increase in the adult-to-juvenile survival ratio and also that the population with the highest adult-to-juvenile survival ratio had the latest mean reproductive age. However, the prediction of an overall increase in mean reproductive age with the adult-to-juvenile survival ratio was not supported, and no statistical tests of the age-specific survival predictions were made with data from only three populations. Fox (1994), using field data from 27 pumpkinseed sunfish (Lepomis gibbosus) populations in Ontario, found that the 90th percentile age in his samples (an indicator of life-span and the adult survival rate), was more highly correlated with reproductive traits than any measure of growth rate. However, adult mortality was not measured directly in this study, and the adult :juvenile survival predictions of life history theory could not be tested in the absence of data on juvenile mortality.

Reznick et al. (1990) manipulated the survivorship rates in guppy (Poecilia reticulata) populations by the transplantation of guppies into and out of environments containing the pike cichlid (Crenicichla alta); a species that was believed to prey preferentially on large, mature guppies. Changes in age at maturity and annual reproductive effort in subsequent generations were consistent with the age-specific mortality hypothesis. However, subsequent studies showed that the pike cichlid was not very size selective on guppies (Mattingly and Butler 1994) and that there was little difference in the relative rate of adult and juvenile mortality between sites where this cichlid was present or absent (Reznick et al. 1996). These subsequent findings cast some doubt on the original interpretation of the study results as being a consequence of changes in the relative mortality of adults and juveniles (Reznick et al. 1996).

The first objective of our study was to test the predictions of life history theory that, relative to juveniles, increased adult survival favors delayed maturity and reduced annual reproductive effort. The species used in our study was the pumpkinseed sunfish, a common species found in lakes, streams, and ponds in eastern North America (Scott and Crossman 1973). Pumpkinseeds are iteroparous, and their populations exhibit a high degree of variation in reproductive life history traits, even in adjacent water bodies (Deacon and Keast 1987, Fox and Keast 1991). Our test of the adult:juvenile survival predictions was based upon multiyear field data generated on populations with very different life histories.
TABLE 1. Physical characteristics of the five study lakes.

                         Surface area           Mean depth
Lake                          (ha)                  (m)

Little Round (LR)              7.5                  11.0
Warrens (WA)                  34.0                   2.6
Beloporine (BP)                7.2                   2.6
Black (BL)                    39.6                   7.8
Vance (VN)                     8.0                   3.6


The second objective of our study was to estimate the survival cost of reproduction in pumpkinseed populations using an optimal life history modeling approach, and to determine whether these costs are related to the life history characteristics of the populations. Although survival costs of reproduction have been demonstrated in many organisms (reviewed in Reznick 1985), few studies have compared these costs among populations with different life history characteristics. Previous studies have shown that pumpkinseeds exhibit a high degree of interpopulation variation in life history characteristics, and that populations with slow growth and stunted adult body size exist that can mature early and have high reproductive allocations in contradiction to the predictions made by several life history models (Danylchuk and Fox 1994a, Fox 1994). Such life history traits are of particular interest because the combination of poor growth, early maturity, and high reproductive allocation would be predicted to incur unusually high survival costs.

METHODS

Study populations

Five pumpkinseed populations were selected from a database of 27 eastern and central Ontario pumpkinseed populations used in a previous life history study (Fox 1994). Criteria used to select these populations were suitability of their water bodies for estimating age-specific mortality rates, environmental stability, diversity in their life history traits, and minimal human influence. The populations selected for this study inhabit closed or nearly closed systems (i.e., lakes with no major inlets or outlets) that are small enough to make age-class population estimates feasible (Table 1); thereby making it possible to estimate the survival probability of cohorts. The lakes containing these populations were observed to have stable water levels, and none had a known history of winterkill (which would result in large annual fluctuations in mortality and age at maturity; Fox and Keast 1991). Little Round Lake (LR) and Warrens Lake (WA) populations exhibit early maturity, a high gonad-to-body mass ratio (gonadosomatic index, or GSI) and stunted adult body size (Danylchuk and Fox 1994a, Fox 1994). The Vance Lake (VN) population exhibits later maturity, lower GSI, and much larger adult body size; characteristics more typical of populations in eastern and central Ontario lakes (Fox 1994). The Beloporine (BP) and Black Lake (BL) pumpkinseeds have life history traits that are intermediate to those of the other three populations. The lakes are found within 20[prime] of latitude and 150 km of one another [ILLUSTRATION FOR FIGURE 1 OMITTED].

Human activity is minimal on all of the lakes except BL. Domestic animal grazing is minimal or absent, and only a few rural residences are located along the shorelines of BP, LR, VN, and WA. Sharbot Lake Provincial Park encompasses nearly half of the shoreline of BL, with a dozen cottages lining the opposite shore. Fishing mortality on pumpkinseeds was considered to be negligible in all of the lakes, as the species is not considered desirable by anglers in Canada (Scott and Crossman 1973), and pumpkinseed was not observed to be the target species of anglers in any of the lakes during the period of study. Therefore, it is reasonable to assume that the mortality observed in the study populations was almost entirely due to natural causes, and that the life histories of these populations are not a short-term response to variable fishing pressure (see Trippel 1995).

In addition to pumpkinseeds, the study lakes are populated by several cyprinid species and yellow perch (Perca flavescens), a potential predator that has been observed to prey upon juvenile pumpkinseeds (M. G. Fox, unpublished data). Predation pressure on adult pumpkinseeds in BP and LR is presumed to be relatively low, as no largemouth bass (Micropterus salmoides), northern pike (Esox lucius), or walleye (Stizostedion vitreum) are present. Other potential predators of yearling and older pumpkinseeds are present in VN (largemouth bass), WA (northern pike), and BL (largemouth bass, smallmouth bass (M. dolomieui), northern pike, and walleye). Other centrarchids inhabiting these water bodies are bluegill sunfish (Lepomis macrochirus), rock bass (Ambloplites rupestris), and black crappie (Poxomis nigromaculatus) in BL, and rock bass in VN.

Determination of the life history characteristics

For each annual assessment, [approximately] 100 individuals per lake were collected in late May or early June, just prior to or at the beginning of the reproductive season. Fish were collected with funnel traps and beach seines in the shallow (0.5-2 m depth) littoral zones of the lakes where pumpkinseeds are most commonly found (Keast and Harker 1977, Werner et al. 1977). Individuals were sacrificed in an ice slurry in the field, and frozen for subsequent analysis.

Pumpkinseeds were measured for total length (TL, in mm) and weighed to the nearest 0.1 g. Gonads were inspected to determine the sex and sexual status of each individual. Only females were used to assess reproductive traits because, unlike males, they allocate the majority of seasonal reproductive energy to the production of gonads (Miller 1963, Gross 1979). A female was considered to be mature if yolked eggs were present. Gonads of mature and maturing individuals were dissected out and weighed (nearest 0.01 g). Age was interpreted from acetate impressions of scales, with annuli identified using the criteria of Regier (1962). To evaluate the precision of age interpretation, a random sample of 10 scales from each population was blindcoded. The same person interpreted each scale on six separate occasions [approximately] 6 mo after the initial reading. The precision of age interpretation was very good (after Campana and Casselman 1993), as the coefficient of variation (Chang 1982) was low in each population (BP = 3.4%; BL = 3.4%; LR = 2.7%; VN = 2.2%; WA = 0%).

Mean age at maturity ([Alpha]) for females was calculated for each population using an equation adapted from DeMaster (1978):

[Alpha] = [summation of] x ([f.sub.x] - [f.sub.x-1]) [where] x = 0 to [Omega] (1)

where x is the age in years, [f.sub.x] is the proportion of females mature at age x, and [Omega] is equal to the maximum age in the sample. The mean length at maturity was calculated in a similar fashion after Trippel and Harvey (1987), where the proportion of mature females was separated into 10 mm size classes.

Gonadosomatic Index (GSI; 100 [center dot] [gonad mass] [center dot] [[total mass].sup.-1]) of mature females was used to represent the relative annual reproductive effort in each population (Bell 1980, Hutchings 1993, Fox 1994). GSI is not a function of pumpkinseed body mass or age within populations (see Deacon and Keast 1987, Fox and Keast 1991), so relationships examined with this parameter would not be biased by among-population differences in size or age structure. Since pumpkinseeds are multiple spawners (Crivelli and Mestre 1988, Fox and Crivelli 1998) and fish were collected only at the beginning of the reproductive period of their population, we used the 90th percentile of the GSI distribution of mature females in the population (GSI90) as a secondary indicator of annual reproductive effort. GSI90 had been previously used as an indicator of the maximum annual reproductive allocation of pumpkinseed populations, and was shown to relate closely to other indicators of reproductive allocation (see Danylchuk and Fox 1994a). The advantage of GSI90 over mean GSI is that the former excludes individuals that were mature but had not yet fully developed their egg mass; however the disadvantage of GSI90 is that it may also exclude fully ripe females that aren't among the top 10% in fecundity at the time the fish were sampled.

Life history characteristics of each population were determined by averaging annual estimates of age at maturity, mean GSI, and GSI90 collected over the period from 1990-1994. The number of years of data range from three for VN to five for BP.

Determination of adult and juvenile survival probabilities

Annual mark-recapture surveys were used to estimate adult abundance and age-class structure in each lake. Surveys were conducted on each population over the period from 1992-1994, except for VN, which was only assessed in 1993 and 1994. Surveys of four of the populations were conducted from late May to early June, prior to the reproductive season. Because of logistical constraints, the VN population was surveyed after the reproductive season, during the first 2 wk of August in both of the 2 yr. Although the post-reproductive assessment would attribute some of the annual mortality in a given year to the year prior, we believe this would have a minimal effect on our overall analysis because most of the mortality in pumpkinseeds after the first year of life would be expected to occur over winter, when feeding stops and lipid reserves are critical (see Hutchings 1994, Justus and Fox 1994).

Pumpkinseeds were collected from each lake with 1 m long x 40 cm diameter wire funnel traps. The funnel traps have a 10-cm opening, and have been adequate in retaining individuals age 1 and older, roughly corresponding to pumpkinseeds 55 mm total length (TL) and larger (Fox 1994). Depending upon the lake, 8-16 funnel traps were randomly set throughout the littoral zone and remained in the water for 12-24 h periods, after which they were emptied. Captured pumpkinseeds were measured (TL), marked by clipping the upper lobe of the caudal fin, and released. Previously marked individuals that were recaptured were also measured and released. After the release of all fish, traps were moved to a new location to reduce the likelihood of immediately recapturing a marked individual. This process continued for a period of 5-14 d until the ratio of recaptured individuals to total number of individuals caught in successive samples was nearly constant.

Abundance estimates were derived separately for each age class using the Schnabel method for multiple mark-recapture in a closed population (Seber 1982); all populations were assumed to have constant sex ratios. The determination of length distributions for the various age classes was based upon the length-at-age relationships generated from the sample of individuals taken from each population for the assessment of life history characteristics. In the case of VN, the length-at-age relationships were generated from a subsample of 100 individuals taken in August at the time the mark-recapture surveys were performed.

The Schnabel census has three main assumptions: (1) population size is constant without recruitment or losses; (2) sampling is random; (3) all individuals have the same chance of capture in any given sample (Seber 1982). These assumptions must not be violated to ensure accurate population estimates. To assess the validity of these assumptions, a regression plot of the proportion of marked individuals in a catch ([m.sub.i]/[n.sub.i]) on the number of previously marked individuals ([M.sub.i]) was generated for each census. Slopes and intercepts were then tested for deviation from 1/N and 0, respectively (Seber 1982).

The proportion of individuals alive at age x that survive to age x + 1 ([p.sub.x]) in successive years was estimated for each age class, based upon the estimated number of individuals alive in each age class ([N.sub.x]). The mean adult survival probability ([S.sub.A]) was weighted for the abundance of each age class across all years sampled using the equation adapted from Stearns (1976):

[Mathematical Expression Omitted] (2)

with age 3 as the initial age class for the adult life history stage and [Omega] equal to the oldest age class observed. Each population was tested for a stable age distribution across all years sampled using a [x.sup.2] contingency table (Seber 1982); percent frequency of [N.sub.x] estimates were arcsine-transformed prior to [X.sup.2] tests (Steel and Torrie 1980). Most natural populations are rarely in stable age distributions, but moderate deviations from stable distributions do not often change predictions about the demography of the population, as it is assumed that these are the patterns towards which the population is converging (Stearns 1992).

Adult survival curves were generated for each population, based upon the probability of surviving from age 3 to age 5. Each point on the adult survival curve was generated from cohort analysis of mark-recapture estimates of age-class abundance. When two or more survival estimates for an age class were available, these were averaged.

Juvenile survival probabilities ([S.sub.j]) were based upon the estimated number of young-of-year (YOY) individuals recruited into each population during 1994 (No) relative to the estimated mean number of age-3 individuals ([N.sub.3]) determined from the mark-recapture studies in 1992, 1993, and 1994. In each lake, the total number of nests producing larvae was determined by direct counts while snorkeling in the littoral zone. The presence of larvae was chosen for the commencement of the juvenile life history stage. Each lake was visited every 7-10 d throughout the reproductive period from early June to mid-August. Nests were marked with colored stones to avoid double counting. During the middle of the reproductive season in each population, all larvae were removed from a subsample of 10 nests with a turkey baster, sacrificed and preserved to be counted later in the lab. These counts were used to generate an estimate for the mean number of larvae produced per nest in each population. The estimated number of YOY recruited in 1994 was calculated by multiplying the mean number of larvae per nest by the total number of nests producing larvae. The generation of the juvenile survival probabilities was based upon virtual population analyses, and it was assumed that the age-0 recruitment in 1994 is representative of the long-term relative recruitment rates of the populations.

To test the hypothesis that populations with a higher ratio of adult-to-juvenile survival probabilities have later ages at maturity and decreased annual reproductive effort, Pearson correlations were calculated between the ratio of adult-to-juvenile survival probability and age at maturity, mean GSI, and GSI90. Also, juvenile and adult survival probabilities for each population were individually correlated with the same reproductive life history traits.

Adult survival estimates calculated from field data are likely to include a mortality component that has been incurred as a result of reproduction, and may therefore provide a biased estimator of the selection pressure on reproductive traits (Partridge and Harvey 1988, Stearns 1992). In order to ensure that the results of our hypothesis tests were not driven by this potential source of bias, we adjusted adult survival probabilities (and thus, adult-to-juvenile survival ratios) upward by removing the mortality component attributable to the survival cost of reproduction, which was generated from our fitness model. The main hypotheses were tested again with the adjusted survival ratios.

Description of fitness model used to estimate the survival cost of reproduction

The survival cost of reproduction, defined here as the percentage decline in the annual survival rate of reproducing females relative to their theoretical survival rate had they not reproduced, was estimated for each population through the use of an optimal fitness model. The model, which was used to calculate the instantaneous rate of population increase, r, under different ages of first maturity, was derived from the discrete version of the Euler-Lotka equation:

1 = [summation of] [e.sup.-rx][l.sub.x][m.sub.x] [where] x = [Alpha] to [Omega] (3)

where x is the age in years, [Alpha] is the age at maturity, to is the maximum age in the sample, [l.sub.x] is the probability of surviving from birth to the beginning of age x, and [m.sup.x] is the expected number of offspring per female of age x (Stearns 1992).

The probability of surviving to age x ([l.sub.x]) was computed using the equation

[l.sub.x] = [N.sub.x]/[N.sub.0] (4)

where [N.sub.x] is the mean abundance estimated for each age, averaged over 3 yr (2 in the case of VN), and [N.sub.0] is the estimated number of YOY recruited in 1994. A cohort analysis was not possible since too few annual abundance estimates were conducted to permit an entire year class to be assessed from age 0 to the oldest observed age class (age 6).

The expected number of female offspring produced per female ([m.sub.x]) was generated using the allometric relationship between body length and fecundity and the probability of surviving the larval stage. Mean total length at age x (T[L.sub.x]) was used to determine the growth trajectories for each population, and was subsequently used to determine [m.sub.x]. To remove among-population bias due to sampling dates, T[L.sub.x] was determined by back calculating the total length, estimated from interpretations of acetate impressions of scales, to the beginning of the current age class using the Fraser-Lee method (Bagenal and Tesch 1978) and a standard body length-scale length intercept (as per Carlander 1982) generated from 27 populations surveyed in eastern and central Ontario (Fox 1994).

Pumpkinseeds are multiple spawners, so fecundity of a female cannot be determined by counting eggs in the ovary at a given point in time. However, Fox and Crivelli (1998) developed a method of assessing number of spawning cycles and total reproductive allocation in pumpkinseed populations from a sample of females reared artificially with males over the spawning season under a feeding regime that would maintain body mass in the absence of spawning. The method involves weekly assessment of readiness to spawn and mass changes in each female, with reproductive allocation determined by summing mass losses in the weeks that individual females were deemed to have spawned. Reproductive allocation (expressed in percentage of prespawning body mass) determined by this method was shown to be comparable to estimates made from a bioenergetic model, and from batch fecundity (based on egg size distribution) of wild females. One of the study populations (BP) was used in these reproductive allocation experiments, resulting in the following relationship between the fecundity and prespawning body length:

[E.sub.x] = [e.sup.(3.39+0.043T[L.sub.x]]) (5)

where [E.sub.x] is the estimated number of eggs per female of age x, and T[L.sub.x] is the prespawning body length of the female (r = 0.62, df = 17). We used this relationship to calculate the fecundity of Beloporine lake females in our optimal fitness calculations. We assumed that 50% of the eggs that hatched would become females in calculating the number of female young produced by each female.

Length-fecundity relationships for the other four populations were based upon the relationship determined for Beloporine Lake females. Although length-fecundity relationships have been shown to differ substantially among populations of brook trout within the same geographic area (see Hutchings 1993), we believe this relationship would not be very different in our pumpkinseed populations because unlike brook trout, there was no significant difference in mean egg diameter among ages sampled within the Beloporine Lake population (P [greater than] 0.1), and there was no significant difference in mean egg diameter among four local populations (including those of Beloporine, Little Round, and Warrens lakes; M. G. Fox, unpublished data). However, we did assume that among-population differences in mean GSI are reflective of differences in fecundity, and we therefore adjusted our fecundity equation for each population by multiplying by the ratio of its mean GSI to that of BP (7.37). Furthermore, we examined the sensitivity of our model outputs to the fecundity model used (see Methods: Estimation of the survival cost of reproduction of the populations).

A growth cost of reproduction was also introduced into the model to reflect the potential loss in fecundity through the reduction in body mass resulting from advancing maturation, or the potential increase in body mass and the associated gain in fecundity achieved by delaying maturity (Roff 1983, Hutchings 1993, Fox 1994). In the reproductive allocation experiment used to generate Eq. 5, the total allocation of Beloporine Lake females to reproduction averaged 1.7 times mean GSI (Fox and Crivelli 1998). Using a length-mass relationship generated for each population, the growth cost was calculated by adding 1.7 times mean GSI to the estimated mass at age for a delay in maturity, and subtracting 1.7 times mean GSI from the estimated mass at age for a decrease in age at maturity (adapted from Fox 1994). The resulting mass from the change in initial age at maturity was converted to length and substituted into Eq. 5 for the fecundity calculation.

Larval survival (L) was calculated by relating the estimated number of larvae produced in 1994 ([N.sub.0]) to the estimated number of eggs produced by mature females in each population using the equation

[Mathematical Expression Omitted] (6)

where [E.sub.x] is the number of eggs produced by a female at age x, [N.sub.x] is the mean number of females alive at age x, and [f.sub.x] is the proportion of mature females at age x. The larval phase lasts up to 2 wk, a point event in time when compared to the rest of an individual's life-span (Roff 1984), and was considered to be constant as it is not dependent upon the age of the female parent. Thus, the expected number of female offspring per female ([m.sup.x]) was assessed as

[m.sub.x] = [0.5e.sup.(3.39+0.043[TL.sup.x])] GSI/7.37 [multiplied by] l. (7)

By substituting the equations for [l.sub.x] and [m.sub.x] into the Euler-Lotka equation,

[Mathematical Expression Omitted] (8)

the observed fitness levels (r) were solved numerically for each population.

Estimation of the survival cost of reproduction of the populations

To estimate the survival cost of reproduction, we used an iterative approach, in which costs were input into the model, and varied until the observed age at maturity in the population was also the optimal age at maturity (i.e., this age had the highest fitness level relative to other ages). The implicit assumption with this approach is that the observed age of maturity is adaptive.

In model runs, fitness was determined for a given age at maturity by assuming no individuals would be mature prior to the age tested, and all individuals would be mature at this and all subsequent ages. The ages tested in the model were discrete, as there is only one time of year when reproduction occurs in pumpkinseeds, and maturity occurs just prior to reproduction. None of the mean ages of maturity calculated in our populations were integers because some individuals mature at a different age than others, and the formula of Demaster (1978) takes this into account. However, in three of the five populations, the observed mean age of maturity calculated from this formula (range: 2.89-3.12) was at or very close to a discrete age (3). For these populations, the survival cost of reproduction was determined to be the level at which maturity at age 3 [TABULAR DATA FOR TABLE 2 OMITTED] produced a higher fitness than maturity at age 2. In the other two populations, the observed mean age of maturity was close to midway between two actual ages (i.e., 2.38 and 3.40). For these populations, the survival cost of reproduction was determined to be the level at which maturity at the age above and the age below the mean age of maturity produced equivalent fitness.

Survival costs were input into the model as a percentage reduction in age-specific survival probability after the initial age at maturity when the age at maturity was reduced, or a percentage increase in age-specific survival probability prior to the initial age at maturity when maturity was delayed. For example, when maturity was delayed to an age greater than the observed age (such as from age 3 to 4), the observed probability of survival was increased in age-3 fish by the percentage being modeled. Since age-3 females would not be reproducing if maturity was delayed to age 4, the model assumes that they would experience an increase in survival equivalent to the survival cost of reproduction, as more energy could then be allocated to growth and/or maintenance, including lipid storage going into the winter (Justus and Fox 1994). The new [l.sub.x] values were then substituted into Eq. 8 to solve for the fitness parameter.

In order to determine the sensitivity of our model output to the fecundity model used in our base runs, we did a second set of runs, altering the length-fecundity relationship (Eq. 5) by increasing or decreasing the exponent (i.e., the expression 3.39 + 0.043T[L.sub.x]) by 25%. The effect of increasing the exponent by 25% is to greatly increase the fecundity of older pumpkinseeds relative to younger ones, whereas reducing this exponent by 25% serves to greatly increase the fecundity of younger pumpkinseeds relative to older ones. The 25% variation in the exponential function used in this analysis was arbitrary, but the extreme differences it produces in the relative fecundity of small and large individuals should more than encompass the among-population variation in wild pumpkinseeds. The survival cost of reproduction determined with these alternate formulations of the length-fecundity relationship were compared with the survival cost of reproduction determined from the base model to indicate the degree to which this cost is affected by the fecundity equation used.

RESULTS

Test of adult :juvenile survival predictions

Summary data for the female reproductive and body size characteristics are displayed in Table 2, and were generally consistent with the life history differences found among these populations in earlier studies (Danylchuk and Fox 1994a, Fox 1994). Although the two populations exhibiting stunted growth (LR and WA) displayed the earliest average ages at maturity and the highest mean GSIs, there was no significant correlation between these two life history traits at the population level (r = -0.23, P = 0.71). The LR and WA populations demonstrated moderate decreases in their growth trajectories coincident with their mean age at maturity, whereas the decreases found for BP and VN were minimal [ILLUSTRATION FOR FIGURE 2 OMITTED]. The BL population exhibited [TABULAR DATA FOR TABLE 3 OMITTED] no detectable change in growth pattern around the mean age at maturity.

Adult population estimates and 95% confidence intervals generated from the mark-recapture studies are shown in Table 3. Mean adult population densities ranged from 83 pumpkinseeds/ha in BL, to [greater than] 3200/ha in BP. No individuals estimated to be older than age 6 were captured in any of the populations, and none estimated to be older than age 5 were captured in LR.

All Schnabel population estimates were found to be valid, as none of the assumptions were shown to be violated. For each census, the plot of the number of marked individuals per catch ([m.sub.i]/[n.sub.i]) against the number of previously marked individuals ([M.sub.i]) was found to have a slope not significantly different from 1/N and an intercept not significantly different from 0 (Seber 1982). None of the regression plots were found to violate the assumptions of linear regression.

Stable age distributions were found to occur in three of the populations across all years sampled (BL: [x.sup.2] = 8.59, P = 0.21; LR: [x.sup.2] = 7.97, P = 0.09; WA: [x.sup.2] = 5.36, P = 0.50). The BP ([x.sup.2] = 19.47) and VN ([x.sup.2] = 16.97) populations were not found to have stable age distributions (P [less than] 0.005). The BP abundance estimate for age 6 in 1992 was extremely low relative to 1993 and 1994, thereby causing the age-3 frequency to be overinflated and accounted for the lack of a stable age distribution [ILLUSTRATION FOR FIGURE 3 OMITTED]. The age-6 estimate in 1992 had no effect on subsequent analyses, in that a survival estimate for this age class could not be computed in the absence of age-7 individuals. With respect to the VN population, the estimate for age 3 in 1994 was 60% less than that for 1993. With only 2 yr of data, it is impossible to determine whether the low age-3 estimate was the result of a poor 1991 year class.

[TABULAR DATA FOR TABLE 4 OMITTED]

Table 4 displays the data used to estimate the number of age 0 (YOY) individuals recruited into the study populations. The estimated number of YOY varied by an order of magnitude among the populations, with VN producing [less than] [10.sup.5] individuals and WA producing just over [10.sup.6]. The two stunted populations (LR and WA) produced the highest number of nests over a longer time period, but had the lowest number of YOY per nest. The VN population had the fewest nests and the shortest recruitment season.

Adult and juvenile survival probability estimates are shown in Table 5. Adult-to-juvenile survival probability ratios ranged from 10.6 to 116.8, and were [log.sub.e] transformed and correlated with mean reproductive traits [ILLUSTRATION FOR FIGURE 4 OMITTED]. The populations with higher ratios of adult-to-juvenile survival tended to exhibit delayed maturity and lower annual reproductive effort, as would be predicted by life history theory. Mean GSI showed a significant negative correlation and GSI90 showed a marginally significant negative correlation (P [less than] 0.10) with the survival ratio (Table 6). However, there was no significant correlation between mean age at maturity and the adult-to-juvenile survival ratio.

Comparisons of populations displaying similar survivorship in one life stage (juvenile or adult) but not the other are instructive, and four such comparisons could be made from our data (Table 7). The prediction of life history theory that an increase in adult survival relative to juvenile survival favors a reduction in annual reproductive effort, was supported by all four comparisons. However, the adult:juvenile survival prediction related to age at maturity was supported in only two of the four comparisons.
TABLE 5. Adult and juvenile survival probability estimates for the
five study populations.

                                                          Adult:
                          Adult         Juvenile        juvenile
                        survival        survival        survival
Lake                  probability     probability     probabilities

Little Round (LR)         0.22           0.008             29.0
Warrens (WA)              0.19           0.018             10.6
Beloporine (BP)           0.67           0.016             41.9
Black (BL)                0.47           0.004            116.8
Vance (VN)                0.48           0.009             51.3


Juvenile survival probability showed no correlation with mean age at maturity, but populations with higher juvenile survival tended to have greater reproductive allocations (Table 6). Adult survival probability showed a significant negative correlation with GSI90, and a positive but insignificant correlation with mean age at maturity.

Adult survivorship ([a.sub.x]) curves for ages 3-5 are illustrated in Fig. 5. The LR and WA populations experienced very low survivorship to age 4 relative to the other populations, and had highly concave adult survivorship curves as a result. In contrast, the adult survivorship curve of BP was convex as a result of high survivorship to age 4 and low survivorship between [TABULAR DATA FOR TABLE 6 OMITTED] ages 4 and 5. BL exhibited a nearly linear survivorship curve from ages 3 to 5.

Survival costs of reproduction

The results of fitness model runs using the four alternative growth and survival cost assumptions are illustrated in Fig. 6. For all populations, the assumption of a growth cost but no survival cost of reproduction was insufficient to make a delay in maturity beyond age 2 an optimal reproductive strategy. Growth plus a 5-10% survival cost of reproduction produced optimal ages at first reproduction that conformed with field data on mean age at maturity in the three populations with normal growth patterns (BL, BP, VN). The two stunted populations required the assumption of much higher survival costs of reproduction in the model to produce optimal ages at first reproduction equivalent to the observed mean ages of maturity. The survival costs of reproduction estimated from the model were 39 and 52% for LR and WA, respectively.

The sensitivity of survival cost of reproduction estimates to variations in the length-fecundity relationship is shown in Table 8. The table shows that a survival cost of reproduction must be assumed in order for the observed age of the maturity to be optimal in each of the populations, even when the length-fecundity exponent is increased by as much as 25%. Furthermore, the estimated survival cost of reproduction in the two stunted, early maturing populations was higher than that of the other three populations within the range of functions modeled, even when the length-fecundity model used was different in the various populations.

[TABULAR DATA FOR TABLE 7 OMITTED]

Test of predictions with adult survival rates adjusted by removing the survival cost of reproduction

The adjustment of field-determined adult survival rates to reflect theoretical adult survival rates if reproduction had not occurred had the greatest effect on the LR and WA populations, due to their high estimated survival costs of reproduction. When the main hypotheses were tested again with the adjusted survival ratios, all of the correlations weakened slightly (Table 9). However, there was still a strong, significant correlation between the adult-to-juvenile survival ratio and mean GSI, and no significant correlation between this ratio and mean age at maturity. There were also no significant correlations between the adjusted adult survival rate and any of the life history variables (P [greater than] 0.16 in all cases).

DISCUSSION

Relationship between age-specific survivorship and life history traits

The prediction that populations with higher ratios of adult-to-juvenile survival would mature later and display lower annual reproductive effort was only partly supported by our results. Relationships between survivorship and life history traits were in the direction predicted by life history theory. The strong and significant negative correlation between mean GSI of the population and its adult-to-juvenile survival ratio provides support for the age-specific survival hypothesis. However, the weaker correlation between this ratio and mean age at maturity and its lack of significance are not supportive of the age-specific survival hypothesis, or at least that part of it that deals with the timing of first maturity.

Our results suggest that reproductive effort is affected by age-specific survivorship. One of the potential costs of high reproductive effort is a decrease in adult survivorship (Gadgil and Bossert 1970, Bell 1980), and such an effect could lead to a spurious correlation between reproductive effort and the adult-to-juvenile survival ratio (Partridge and Harvey 1988). However, a strong correlation between this ratio and GSI remained even when the adult survival rate was adjusted by removing the component attributable to reproduction. Furthermore, neither juvenile nor adult survivorship alone was sufficient to explain the variation in reproductive effort among study populations. The relationship between reproductive effort and survivorship was significant only when adult survival was considered relative to juvenile survival. This supports the contention of several life history theorists that what matters in the evolution of life history traits is the relative rate of adult and juvenile survival, not the overall rate of mortality (Gadgil and Bossert 1970, Partridge and Harvey 1988).

The significant relationship between age-specific survivorship and reproductive effort and the lack of such a relationship between age-specific survivorship and age at maturity is not unprecedented. Although Hutchings (1993) did not do a statistical test of the age-specific hypothesis predictions, his data [ILLUSTRATION FOR FIGURE 2 OMITTED] exhibited a strong negative relationship between the adult-to-juvenile survival ratio of a brook trout population and its mean GSI, but no evident relationship between the survival ratio and the mean age of mature individuals. This outcome was not noted in Hutchings' (1993) study; perhaps because the collective results, dealing with both growth and survival predictions, were strongly supportive of life history theory.

The concordance of our results with those of Hutchings (1993) suggests that age-specific survivorship may exert a stronger selective pressure on reproductive effort than it does on age at maturity. One reason why this might be so is that age at maturity is not always synonymous with age at first reproduction. Several studies on fishes have shown that mature individuals do not necessarily spawn (Raffetto et al. 1990; Ridgway et al. 1991; Baylis et al. 1993; Danylchuk and Fox 1994b). Danylchuk and Fox (1994b) found that the proportion of age-2 female pumpkinseeds observed spawning in LR was much lower than their proportion of the mature females in the population, but this underrepresentation of the youngest mature age class was not detected in BP, where the individuals were larger in size. Early maturation without assurance of a spawning opportunity could be a bet-hedging strategy with a minimal energetic cost, since many fishes, including pumpkinseeds (Crivelli and Mestre 1988), are capable of resorbing unspawned eggs. If mean age at maturity of female pumpkinseeds underestimates their mean age at first reproduction in populations with small, early maturing individuals, this might explain why the age at maturity prediction in our study did not conform to theory. However, this possibility cannot explain the lack of a relationship between age-specific survivorship and age at maturity in Hutchings' (1993) study, because increasing the relative age at maturity of the population with the smallest, earliest maturing individuals would actually worsen this relationship in his data set. Thus, a more parsimonious explanation for the absence of a significant relationship between the adult :juvenile survival ratio and mean age at maturity is that some other factor related to fitness exerts a stronger influence than age-specific survivorship on the timing of first maturity.

Other factors potentially affecting mean age at maturity

The potential influence of growth rate on the evolution and phenotypic expression of life history traits has been recognized in the development of many life history models. Two such models (Roff 1984, Stearns and Koella 1986) utilize somatic growth functions to represent the gain in fecundity associated with growth, but neither separates juvenile from adult growth. A third model (Hutchings 1993) is based upon the assumption that the survival rate is directly related to the growth rate. The latter model does separate juvenile from adult growth, and was used to graphically assess the prediction that age at maturity is positively related to the ratio of adult-to-juvenile growth (Hutchings 1993; [ILLUSTRATION FOR FIGURE 3 OMITTED]).

Fox (1994) examined the relationship between growth and mean age at maturity in 27 pumpkinseed sunfish populations, including those utilized in the present study. He tested growth indicators that reflect the potential influence of phenotypic plasticity (length at age 2) and evolutionary trade-offs (adult-to-juvenile growth ratio, adjusted for reproductive output), and found that both were significantly correlated with mean age at maturity. The prediction that a high adult-to-juvenile growth ratio leads to delayed maturity is also strongly supported with data from the five pumpkinseed populations used in our present study (r = 0.98, P = 0.003), and the prediction that fast juvenile growth leads to early maturity is supported with a weaker, marginally significant correlation (r = -0.82, P = 0.09). Hutchings (1993) did not address the phenotypic growth question, but his data show a marginally significant correlation between adult-to-juvenile growth ratio and mean reproductive age (r = -0.82, P = 0.09; calculated from data in Hutchings' Fig. 3). Hutchings (1996) later showed that brook trout genotypes which modify their age at maturity in accordance with their growth rate will be fitter than those which do not display this phenotypic plasticity. The degree of concordance among these results indicates that growth rate (juvenile and/or age-specific) can strongly influence age at maturity, and its influence on maturity can be stronger than that of age-specific survivorship in some species.
TABLE 8. Sensitivity of survival cost of reproduction estimates (in
percentages) to variations in the length-fecundity relationship
assumed in our optimal fitness model.

                           Base         0.75x base     1.25x base
Lake                     equation        exponent       exponent

Little Round (LR)           39              29             50
Warrens (WA)                52              35             68
Beloporine (BP)              8               2             20
Black (BL)                   5               2             13
Vance (VN)                  10               2             22


Intraspecific and interspecific competitive interactions have also been hypothesized to affect life history traits (Pianka 1970). Fox (1994) tested the predictions of the r- and K-selection model, that populations with high density will display late maturity and low reproductive investment. He found significant correlations between catch per unit effort (an indicator of population density) and both mean age at maturity and GSI, but these correlations were in the opposite direction of those predicted by the model. In the present study, population density was not significantly correlated with either mean age at maturity or GSI (P [greater than] 0.90 in both cases), but neither study contained estimates of juvenile density. If juvenile density affects age at maturity, the likely mechanism would be through juvenile growth rate, which has already been examined.

Fox (1994) also considered the effects of interspecific competition on life history traits, and found that pumpkinseed populations living sympatrically with bluegill sunfish (Lepomis macrochirus, a juvenile competitor) matured significantly later with a significantly lower GSI than pumpkinseed populations inhabiting water bodies without bluegills. Only one of the five lakes used in the present study (BL) contained a bluegill population. Since BL was not an outlier in any of our tests, it is unlikely that the presence of bluegills in this lake influenced our results.

Survival cost of reproduction

The results of our study indicate that reproduction in pumpkinseeds carries a survival cost, and that this cost is highest in populations that reproduce early and at a small size. The fitness model simulations demonstrated that all populations should mature at age 2 rather than at the observed ages, unless a survival cost of reproduction is assumed. To predict an optimal age at maturity equal to the observed age in the two stunted, early maturing populations, a survival cost of reproduction of at least 39% had to be introduced into the base model. The estimated survival cost of reproduction in these two populations was at least 2.5 times as high as those of the three "normal" populations under the assumption of a single length - fecundity relationship for all populations. Results of a tethering experiment showed that the mortality risk from predation was virtually nil to adult pumpkinseeds in LR, the earliest maturing of the study populations (T. C. Pratt and M. G. Fox, manuscript in preparation). High adult mortality in the LR population would therefore be intrinsic, and is undoubtedly related to early maturity and high reproductive allocation. This evidence is consistent with the results of our model output, and provides support for the prediction of life history theory that increased reproductive effort results in reduced adult survival (Williams 1966).

The LR and WA populations experienced greater reductions in length increment than the other three populations, coincident with the onset of maturity [ILLUSTRATION FOR FIGURE 2 OMITTED]. The three later maturing populations demonstrated maximized fitness at the observed reproductive strategies with only 5-10% survival costs of reproduction assumed in the base fitness model. Given that the LR and WA populations mature early at a small body size and have high reproductive allocation relative to the other populations, one would expect reproduction to have a greater effect on post-reproductive survival in these two populations. Our modeling results are consistent with this supposition.

Reduced survival in reproductive individuals has been shown to be either a direct or indirect consequence of reproduction in grasses (Law 1979), fish (Pressley 1981, Dufrense et al. 1990, Hutchings 1993), snakes and lizards (Shine 1980, Shine and Schwarzkopf 1992), birds (Reid 1987, Nur 1988), and small mammals (Millar 1994). Survival costs of reproduction in an early maturing brook trout population were more than twice those of two later maturing populations, and reproductive activity increased the age-specific risk of over-winter mortality in these populations by 17-89% (Hutchings 1994). Mature age-2 female pumpkinseeds were shown to have proportionately less lipid present in their body tissue than immature age-2 females, and it was suggested the mature individuals would experience reduced over-winter survival (Justus and Fox 1994). Several studies have shown individuals with greater proportionate lipid reserves to have a consequent reduction in over-winter mortality of smallmouth bass (Oliver et al. 1979, Shuter et al. 1980) and yellow perch (Post and Evans 1989).
TABLE 9. Pearson correlations between the adult-to-juvenile survival
ratio, adjusted to remove the effect of survival cost of
reproduction, and the reproductive life history characteristics for
the five study populations.

                              Correlations with adult : juvenile
Life history                             survival ratio
characteristic                        r              P

Mean age at maturity                 0.35           0.56
Mean GSI                            -0.90           0.036
90th-percentile GSI                 -0.70           0.19

Notes: Adult : juvenile survival probabilities were
[log.sub.[e.sup.-]] transformed prior to analysis. GSI =
gonadosomatic index.


Much attention has been paid to the theoretical implications of reproductive costs on the evolution of life histories, particularly survival costs (reviewed in Reznick 1985, Roff 1992, Stearns 1992). Comparatively few studies have assessed survival costs directly, as most evidence found to support reduced post-reproductive survival has been circumstantial in nature (Roff 1984).

Although a direct assessment of the survival cost of reproduction was not made in this study, the results of our study suggest that these costs exist in pumpkinseeds, and are greatest in populations that mature early and at a small size. Furthermore, our results suggest that the selective pressure of a low adult-to-juvenile survival ratio is strong enough to favor the evolution of increased reproductive effort in pumpkinseeds, even when this results in the decline of an already low adult survival rate. Research is in progress to determine intrinsic and extrinsic mortality rates in these populations, as the source of the mortality may influence its effect on life history traits.

ACKNOWLEDGMENTS

Financial support for this research was provided by the Natural Sciences and Engineering Research Council of Canada (NSERC) in the form of a Research Grant and a Trent-NSERC award to MGE We thank Charlotte Breadnet, Ian McDonald, Dwayne Peltzer, Trevor Pratt, Stacey Robb, and Vicki Stevenson for their assistance in the field. We also thank Ed and Florence Warren for use of their cabin, and Sharbot Lake Provincial Park staff for the use of a campsite in the park. Finally we thank Tim Ehlinger, Carol Folt, Jeff Hutchings, Erica Nol, Brian Shuter, and Jim Sutcliffe for their helpful comments on earlier versions of this manuscript.

LITERATURE CITED

Bagenal, T. B., and F. W. Tesch. 1978. Age and growth. Pages 101-136 in T B. Bagenal, editor. Methods for assessment of fish production in fresh waters. Third edition. IBP Handbook Number 3. Blackwell Scientific, Oxford, UK.

Baylis, J. R., D. D. Wiegmann, and M. H. Hoff. 1993. Alternating life histories of smallmouth bass. Transactions of the American Fisheries Society 122:500-510.

Bell, G. 1980. The costs of reproduction and their consequences. American Naturalist 116:45-76.

Campana, S. E., and J. M. Casselman. 1993. Stock discrimination using otolith shape analysis. Canadian Journal of Fisheries and Aquatic Sciences 50:1062-1083.

Carlander, K. D. 1982. Standard intercepts for calculating lengths from scale measurements for some centrarchid and percid fishes. Transactions of the American Fisheries Society 111:332-336.

Chang, W. Y. B. 1982. A statistical method for evaluating the reproducibility of age determination. Canadian Journal of Fisheries and Aquatic Sciences 39:1208-1210.

Charlesworth, B. 1980. Evolution in age-structured populations. Cambridge University Press, Cambridge, UK.

Charnov, E. L., and W. M. Schaffer. 1973. Life history consequences of natural selection: Cole's result revisited. American Naturalist 107:791-793.

Crivelli, A. J., and D. Mestre. 1988. Life history traits of pumpkinseed, Lepomis gibbosus, introduced into the Camargue, a mediterranean wetland. Archiv fur Hydrobiologie 111:449-466.

Danylchuk, A. J., and M. G. Fox. 1994a. Seasonal reproductive patterns of pumpkinseed (Lepomis gibbosus) populations with varying body size characteristics. Canadian Journal of Fisheries and Aquatic Sciences 51:490-500.

Danylchuk, A. J., and M. G. Fox. 1994b. Age and size-dependent variation in the seasonal timing and probability of reproduction among mature female pumpkinseed, Lepomis gibbosus. Environmental Biology of Fishes 39:119-127.

Deacon, L. I., and J. A. Keast. 1987. Patterns of reproduction in two populations of pumpkinseed sunfish, Lepomis gibbosus, with differing food resources. Environmental Biology of Fishes 19:281-296.

DeMaster, D. P. 1978. Calculation of the average age of sexual maturity in marine mammals. Journal of the Fisheries Research Board of Canada 35:912-915.

Dufrense, F., G. J. Fitzgerald, and S. Lachance. 1990. Intermale competition and reproductive costs in threespine stickleback. Behaviourial Ecology 1:140-147.

Fox, M. G. 1994. Growth, density, and interspecific influences on pumpkinseed sunfish life histories. Ecology 75: 1157-1171.

Fox, M. G., and A. J. Crivelli. 1998. Body size and reproductive allocation in a multiple spawning centrarchid. Canadian Journal of Fisheries and Aquatic Sciences 55:737-748.

Fox, M. G., and A. Keast. 1991. Effect of overwinter mortality on reproductive life history traits of pumpkinseed (Lepomis gibbosus) populations. Canadian Journal of Fisheries and Aquatic Sciences 49:1792-1799.

Gadgil, M., and W. H. Bossert. 1970. Life historical consequences of natural selection. American Naturalist 104: 1-22.

Gross, M. R. 1979. Cuckoldry in sunfishes (Lepomis: Centrarchidae). Canadian Journal of Zoology 57:1507-1509.

Hutchings, J. A. 1993. Adaptive life histories effected by age-specific survival and growth rate. Ecology 74:673-684.

-----. 1994. Age- and size-specific costs of reproduction within populations of brook trout, Salvelinus fontinalis. Oikos 70:12-20.

-----. 1996. Adaptive phenotypic plasticity in brook trout, Salvelinus fontinalis, life histories. Ecoscience 3:25-32.

Justus, J. A., and M. G. Fox. 1994. The cost of early maturation on growth, body condition and somatic lipid content in a lake pumpkinseed (Lepomis gibbosus) population. Ecology of Freshwater Fish 3:9-17.

Keast, A., and J. Hacker. 1977. Fish distribution and benthic invertebrate biomass relative to depth in an Ontario lake. Environmental Biology of Fishes 2:235-240.

Kozlowski, J., and J. Uchmanski. 1987. Optimal individual growth and reproduction in perennial species with indeterminate growth. Evolutionary Ecology 1:214-230.

Law, R. 1979. The cost of reproduction in annual meadow grass. American Naturalist 113:3-16.

Mattingly, H., and M. J. Butler IV. 1994. Laboratory predation on the Trinidadian guppy: implications for the size-selective predation hypothesis and guppy life history evolution. Oikos 69:54-64.

Michod, R. E. 1979. Evolution of life histories in response to age-specific mortality factors. American Naturalist 113: 531-550.

Millar, J. S. 1994. Senescence in a population of small mammals? Ecoscience 1:317-321.

Miller, H. C. 1963. The behaviour of the pumpkinseed sunfish, Lepomis gibbosus (Linnaeus), with notes on the behaviour of other species of Lepomis and the pygmy sunfish, Elassoma evergladei. Behaviour 22:88-151.

Nur, N. 1988. The consequences of brood size for breeding blue tits. III. Measuring the cost of reproduction: Survival, future fecundity, and differential dispersal. Evolution 42: 351-362.

Oliver, J. D., G. F. Holeton, and K. E. Chua. 1979. Overwinter mortality of fingerling smallmouth bass in relation to size, relative energy stores, and environmental temperature. Transactions of the American Fisheries Society 108: 130-136.

Partridge, L., and P. H. Harvey. 1988. The ecological context of life history evolution. Science 241:1449-1454.

Partridge, L., and R. Sibly. 1991. Constraints in the evolution of life histories. Philosophical Transactions of the Royal Society of London, Series B 332:3-13.

Pianka, E. R. 1970. On "r" and "K" selection. American Naturalist 104:592-597.

Post, J. R., and D. O. Evans. 1989. Size-dependent overwinter mortality of young-of-the-year yellow perch (Perca flavescens): laboratory, In Situ, and field experiments. Canadian Journal of Fisheries and Aquatic Sciences 46:1958-1968.

Pressley, P. H. 1981. Parental effort and the evolution of nest-guarding tactics in the threespine stickleback, Gasterosteus aculeatus. L. Evolution 35:282-295.

Raffetto, N. S., J. R. Baylis, and S. L. Serns. 1990. Complete estimates of reproductive success in a closed population of smallmouth bass (Micropterus dolomieui). Ecology 71: 1523-1535.

Regier, H. A. 1962. Validation of the scale method for estimating age and growth of bluegills. Transactions of the American Fisheries Society 91:362-374.

Reid, W. V. 1987. The cost of reproduction in the glaucous-winged gull. Oecologia 74:458-467.

Reznick, D. N. 1985. Cost of reproduction: an evaluation of the empirical evidence. Oikos 44:257-267.

Reznick, D. N., H. Bryga, and J. A. Endler. 1990. Experimentally-induced life history evolution in a natural population. Nature 346:357-359.

Reznick, D. N., M. J. Butler, F. H. Rodd, and P. Ross. 1996. Life-history evolution in guppies (Poecilia reticulata). 6. Differential mortality as a mechanism for natural selection. Evolution 50:1651-1660.

Ridgway, M. S., B. J. Shuter, and E. E. Post. 1991. The relative influence of body size and territorial behaviour on nesting asynchrony in male smallmouth bass, Micropterus dolomieui (Pisces: Centrarchidae). Journal of Animal Ecology 60:665-681.

Roff, D. A. 1983. An allocation model of growth and reproduction in fish. Canadian Journal of Fisheries and Aquatic Sciences 40:1395-1404.

-----. 1984. The evolution of life history parameters in teleosts. Canadian Journal of Fisheries and Aquatic Sciences 41:989-1000.

-----. 1992. The evolution of life histories: theory and analysis. Chapman and Hall, New York, New York, USA.

Schaffer, W. M. 1974. Selection for optimal life histories: the effects of age structure. Ecology 55:291-303.

Scott, W. B., and E. J. Crossman. 1973. Freshwater fishes of Canada. Bulletin 184, Fisheries Research Board of Canada. Supply and Services Canada, Ottawa, Ontario, Canada.

Seber, G. A. F. 1982. The estimation of animal abundance. Hafner Press, New York, New York, USA.

Shine, R. 1980. "Costs" of reproduction in reptiles. Oecologia 46:92-100.

Shine, R., and L. Schwarzkopf. 1992. The evolution of reproductive effort in lizards and snakes. Evolution 46:62-75.

Shuter, B. J., J. A. McLean, F. E. J. Fry, and H. A. Regier. 1980. Stochastic simulation of temperature effects on first year survival of smallmouth bass. Transactions of the American Fisheries Society 109:1-34.

Stearns, S. C. 1976. Life history tactics: a review of the ideas. Quarterly Review of Biology 51:3-47.

-----. 1992. The evolution of life histories. Oxford University Press, Oxford, UK.

Stearns, S. C., and J. C. Koella. 1986. The evolution of phenotypic plasticity in life-history traits: predictions of reaction norms for age and size at maturity. Evolution 40: 893-913.

Steel, R. G. D., and J. H. Torrie. 1980. Principles and procedures of statistics: a biometrical approach. Second edition. McGraw-Hill Book Company, New York, New York, USA.

Trippel, E. A. 1995. Age at maturity as a stress indicator in fisheries. BioScience 45:759-771.

Trippel, E. A., and H. H. Harvey. 1987. Reproductive responses of five white sucker (Catostomus commersoni) populations in relation to lake acidity. Canadian Journal of Fisheries and Aquatic Sciences 44:1018-1023.

Werner, E. E., D. J. Hall, D. R. Laughlin, D. J. Wagner, L. A. Wilsmann, and F. C. Funk. 1977. Habitat partitioning in a freshwater fish community. Journal of the Fisheries Research Board of Canada 34:360-370.

Williams, G. C. 1966. Natural selection, the costs of reproduction and a refinement of Lack's principle. American Naturalist 100:687-690.
COPYRIGHT 1999 Ecological Society of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Bertschy, Kirk A.; Fox, Michael G.
Publication:Ecology
Geographic Code:1USA
Date:Oct 1, 1999
Words:9962
Previous Article:SPATIAL HETEROGENEITY OF STREAM WATER NUTRIENT CONCENTRATIONS OVER SUCCESSIONAL TIME.
Next Article:THE EVOLUTIONARY ECOLOGY OF LIFE HISTORY VARIATION IN THE GARTER SNAKE THAMNOPHIS ELEGANS.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters