Data Loading...
PHENOTYPIC PLASTICITY IN THE LIFE HISTORY TRAITS OF ... Flipbook PDF
419 Ecology, 78(2), 1997, pp. 419–433 q 1997 by the Ecological Society of America PHENOTYPIC PLASTICITY IN THE LIFE HIST
125 Views
35 Downloads
FLIP PDF 134.28KB
Ecology, 78(2), 1997, pp. 419–433 q 1997 by the Ecological Society of America
PHENOTYPIC PLASTICITY IN THE LIFE HISTORY TRAITS OF GUPPIES: RESPONSES TO SOCIAL ENVIRONMENT F. HELEN RODD,1,3 DAVID N. REZNICK,2
AND
MARLA B. SOKOLOWSKI1
1
Department of Biology, York University, 4700 Keele Street, North York, Ontario, M3J 1P3, Canada 2Department of Biology, University of California, Riverside, California 92521 USA
Abstract. Guppies ( Poecilia reticulata ) living in the streams and rivers of Trinidad show significant interpopulation variation in life history traits. Some of the greatest divergence is among guppies that co-occur with different assemblages of predators. Reznick has demonstrated that there is a genetic basis for some, but not all, of these differences. Several features of guppies’ biology co-vary with the predatornities including their social and sexual behaviors and the demography of their populations. Here, we test the hypothesis that some of the observed phenotypic variation in life history traits is induced by differences in the social environment of guppies. We tested this by rearing focal individuals in the laboratory in experimental populations composed of wild-caught guppies. The experimental populations were designed to represent two types of guppy communities: (1) Rivulus communities, where guppies co-occur only with a small killifish, Rivulus hartii, that preys on small guppies, and (2) Crenicichla communities, where they co-occur with several larger species of predator including Crenicichla alta, a cichlid that prefers to prey on large guppies. We manipulated both the demography and the origin (predator community) of the guppies comprising the experimental populations. The focal guppies were offspring of guppies from one of the two predator localities. We found that guppies responded to both the demography and origin of the conspecifics with which they were reared. Therefore, phenotypic plasticity in response to social environment is probably contributing to the variation in life history traits that has been observed in the field. Our discovery that there are differences in the responsiveness of guppies from Rivulus and Crenicichla localities to our manipulations of social environment suggests that the patterns of plasticity have themselves evolved. Key words: age at maturity; behavioral interactions; fecundity; interpopulation variation; life history traits; offspring size; phenotypic plasticity; Poecilia reticulata; population demography; size at maturity.
INTRODUCTION A geographic pattern in the distribution of an organism with a variable trait often provides the first clue that the trait is an adaptation (Endler 1986). Such a pattern may have evolved directly in response to a feature of the environment, i.e., there is a genetic basis for the observed variation (e.g., Endler 1978, 1980). Alternatively, or additionally, such a pattern may represent a plastic response to the environment, such as the induction of defensive morphologies in response to the presence of predators (see Spitze and Sadler 1996). A third possibility is that the pattern is an indirect response to some aspect of the environment. For example, aquatic predators can cause reduced fecundities in insects with terrestrial adults and aquatic larvae; the presence of predators causes the immature stages to reduce feeding rates and this ultimately results in reduced fecundities of the adults (Feltmate and Williams Manuscript received 22 May 1995; revised 22 January 1996; accepted 14 February 1996; final version received 28 May 1996. 3 Present address: Section of Evolution and Ecology, University of California, Davis, California 95616 USA.
1991, Peckarsky et al. 1993). In this paper, we evaluate the indirect effects of predators on the life history traits of a small freshwater fish, the guppy (Poecilia reticulata). Reznick and Endler (1982, Reznick 1989) have shown that there are substantial interpopulation differences in the life history traits of guppies that live with different assemblages of predators in natural populations in Trinidad. Guppies in one type of population co-occur with a small killifish, Rivulus hartii, that preys on small guppies (Seghers 1973, 1974, Liley and Seghers 1975, Mattingly and Butler 1994). Guppies in the other type of population co-occur with a number of predators, the most important of which is a large cichlid, Crenicichla alta, that prefers to prey on larger guppies (Seghers 1973, Liley and Seghers 1975, Mattingly and Butler 1994). Laboratory studies have shown that there is a genetic basis for the interpopulation differences in the life histories of guppies (Reznick 1982). However, the magnitudes of the differences are reduced when the guppies are reared in the laboratory. For example, offspring size of guppies from the two predator localities differs by 10–20% in the laboratory but by 100% in field samples (Reznick 1982). Therefore, some
419
420
F. HELEN RODD ET AL.
of the observed phenotypic variation among wild guppies must be environmentally induced. What characteristic of the environment is contributing to the variation in life history traits? Although guppies living with Rivulus tend to live in smaller, higher elevation streams than those living with Crenicichla, Strauss (1990) showed that 83% of the variation in life history traits was associated with the major species of predator at a site; only an additional 10% could be accounted for by geographic patterns, such as stream size. This suggests that differences in the physical environment make only a minor contribution to observed life history patterns. We propose that variation in the social environment of guppies, including variation in their population demography (Haskins et al. 1961, Seghers 1973, Reznick and Endler 1982, Rodd and Reznick 1997) and social behavior (see Endler 1995 for a review), is contributing to the observed patterns. Total densities of guppies are higher at Rivulus sites than at Crenicichla sites (Reznick and Endler 1982). Populations at Rivulus sites have lower proportions of immature guppies than those at Crenicichla sites (Rodd and Reznick 1997). The sex ratios of adults tend to be more female biased at Rivulus sites than at Crenicichla sites (Haskins et al. 1961, Seghers 1973, Rodd and Reznick 1997). All of these aspects of population demography could influence the number and types of interactions a guppy would experience. Guppies from Rivulus and Crenicichla sites also differ behaviorally. Crenicichla locality guppies school more readily, are less aggressive, and engage more frequently in sexual behavior (Seghers 1973, 1974, Farr 1975, Liley and Seghers 1975, Magurran and Seghers 1990, 1991, 1994; see Endler 1995 for a review). These differences in behavior could also influence the ways in which guppies interact with conspecifics. Previous studies have shown that conspecific interactions can have a profound influence on an individual’s fitness (e.g., Itzkowitz 1990, McLain 1992, Madsen and Shine 1993). The advantages of an ability to assess current and future reproductive prospects, which often depend on conspecific interactions, and to adjust life history traits accordingly have been described (Williams 1966, Schultz and Warner 1989). Indeed, poeciliids, including guppies, are known to vary their life history traits with social conditions (e.g., Borowsky 1973, Warren 1973, Sohn 1977a, b, Dahlgren 1979, 1985). Hence, we hypothesized that some of the observed interpopulation variation in guppy life history traits is a phenotypic response induced by differences in social interactions. To test this, we reared offspring of guppies from both Rivulus and Crenicichla localities in the laboratory in social conditions that were designed to represent those experienced in the field. We reared juvenile guppies either in a demographic environment representative of conditions in a Rivulus locality (high density, femalebiased sex ratio, low proportion of juveniles) or in a
Ecology, Vol. 78, No. 2
Crenicichla locality (low density, even sex ratio, high proportion of juveniles) (Haskins et al. 1961, Seghers 1973, Reznick and Endler 1982, Rodd and Reznick 1997). The juvenile guppies were reared with wildcaught adults and lab-born subadults from Rivulus or Crenicichla localities. Since the behavior of guppies from different localities varies, using guppies from the two localities should manipulate the types of interactions to which the juveniles were exposed. The juveniles were reared to maturity and their life history traits measured. If social environment is indeed responsible for inducing some of the interpopulation variation in life history traits, we should be able to reproduce some of the variation that has been observed in the field. For example, size at maturity is larger and guppy population densities are higher in Rivulus localities. If population density is responsible for some of the variation in size at maturity, guppies reared at high densities should be larger than those reared at lower densities. MATERIALS
AND
METHODS
Methods used to conduct this experiment are fully detailed in Rodd and Sokolowski (1995) and are briefly described here. Both studies consider the response of juvenile guppies to the social environment they encounter while maturing. The previous study considered the effects of social environment on male sexual behavior. This study considers the effects of social environment on life history traits. The females in this study were siblings of the males described in both studies. Juvenile guppies, the test individuals, were reared to sexual maturity under one of four social conditions (Rivulus-type demography 1 Rivulus locality conspecifics; Rivulus-type demography 1 Crenicichla locality conspecifics; Crenicichla-type demography 1 Rivulus locality conspecifics; Crenicichla-type demography 1 Crenicichla locality conspecifics). Test individuals were offspring of guppies from either Rivulus or Crenicichla localities. All combinations of social environment and origin of test individuals were evaluated for a total of eight treatments. Therefore, the three independent variables considered here are: the origin of the test individual’s parents, labeled ‘‘Parents’’; the origin of the guppies with whom the test individuals were reared (‘‘Residents’’); and population demography, the type of demography in which the test individuals were reared (‘‘Demography’’). Parents presumably influenced the genetic background of the guppies. Given the differences in the behavior of guppies from the two predator localities, Residents presumably varied the interactions between the test individuals and the residents. Other characteristics of guppies from the two localities also differ, including male coloration (Endler 1978, 1980) and possibly pheromones (Snyder 1978 cited in Luyten and Liley 1991) (see Endler 1995 for a review), and may also have had an influence on the
March 1997
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
test individual’s development. Demography assessed the effects of density, age structure, and sex ratio. Two replicates of six of the eight treatments were run at a time; 17 sets of replicates were done over 32 mo to accumulate sufficient sample sizes.
Origin of test guppies Test individuals were born in the lab to wild-caught females from either Crenicichla or Rivulus localities in the Northern Range in Trinidad. Dams were isolated before parturition so that we could collect complete broods of known age. The young were reared with no more than twelve (usually 6–8) of their broodmates in a 35.5 3 20.2 cm aquarium until they were introduced into the treatment tanks.
Origin and demography of the residents Populations of guppies from Rivulus or Crenicichla localities were established in large aquaria in the laboratory; these were the Residents. Adult residents were captured, as adults, in Trinidad. They were collected in September 1988, March 1989, February-March 1990, and March-April 1991 from four Rivulus sites and four Crenicichla sites. Subadult residents were labborn offspring of guppies from either Crenicichla or Rivulus localities. Subadults were marked, to distinguish them from test individuals, by the injection of a tiny drop of acrylic paint solution into their caudal peduncle (Rodd and Reznick 1991). Residents were chosen randomly from among the collections from the appropriate predator locality. The sex ratios, densities, and age structures of the populations of residents were designed to represent natural populations of guppies in those localities (Crenicichla-type demography: two adult females : two adult males : two juveniles; Rivulus-type demography: five adult females : two adult males : two juveniles) (Haskins et al. 1961, Seghers 1973, Rodd and Reznick 1997). The treatment tanks were large (31.6 3 76 cm filled to a depth of ø31.5 cm) and the densities of guppies ranged from the maximum observed in the field (25/m2) (Reznick and Endler 1982) to twice that number. However, guppies exhibit schooling behavior and that can increase the density they experience in the field (Seghers 1974, Farr 1975). The demography of the populations was maintained throughout the experiment by replacing the few individuals that died with similarsized animals, by replacing subadult residents as they matured, and by removing newborn individuals.
Test procedures The test individuals were introduced into the treatment tanks when they were 5–6 wk old; at this age they were large enough that they were no longer at risk of being eaten by the residents. Males had not initiated the maturation process (i.e., they showed no development of their anal fin [Reznick 1990]) when they were introduced. One week after being introduced,
421
most individuals were 10–14 mm (standard length) (range: 9.5–15.7 mm). Introductions of test individuals from Rivulus and Crenicichla localities were paired so that a male and/or a female (depending on the size of the brood) were added to each treatment tank within a 1-wk period. We used the presence of melanophores in the region of females’ brood spot to distinguish males from females. All test individuals from one locality in one introduction period were from the same brood. Subsequent introductions were made as previously added test individuals reached maturity. Most test individuals were exposed to the experimental treatments for at least 4 wk before they reached sexual maturity. The fish were fed liver paste or brine shrimp nauplii once or twice a day. The food was dispersed so that all individuals would have access to it. High-density treatment tanks received twice the ration that low-density tanks did so that the availability of food per fish was similar.
Measurement of female life history traits We evaluated the influence of the experimental treatments on age and size at first parturition, number of offspring per brood, size of offspring (total and lean [fat-extracted] mass), interbrood interval, reproductive allocation (percentage of dry mass allocated to their last brood), and growth rate. Females were weighed and their standard lengths measured 1 wk after they were first introduced into the treatment tank. They were then weighed and measured every other week until they were placed in an isolation tank. Females provide no external cues to their reproductive status so we assumed that they would mature at about the same time as their male sibs and give birth 25–30 d later (Reznick 1982). Therefore, females were placed in an isolation tank (4L) 7–14 d after their most rapidly developing male sib attained maturity. They were held there with two mature males, of the same origin as the residents in the treatment tank, until they produced their first brood. Males were added to the isolation tank to ensure that she had an adequate sperm supply. Floating plants were placed in the tank to provide cover for newborn young. Isolation tanks were checked twice daily for the presence of babies. Young were removed immediately, killed with an overdose of anesthetic, and preserved in formalin. Later they were dried overnight at 558C, weighed, then placed in ether to extract their fat until there was no further change in their dry mass, then reweighed. After the females produced their first brood, they were weighed, measured, marked with an injection of latex paint and returned to their treatment tank. After 16 d (minimum length of time between broods), they were returned to the isolation tank until they had a second brood. Interbrood interval was the time duration between broods. We kept females until they produced two broods for replicates nos. 1–6 and three broods for replicates nos. 7–17. Females were euthanized in anesthetic after they produced their last litter, then frozen.
F. HELEN RODD ET AL.
422
Ecology, Vol. 78, No. 2
TABLE 1. Analyses of variance on female life history traits.
Characteristics of the female when she produced her first brood i) Length (mm) Source Covariate† Parents‡ Residents Demography Parents 3 Residents Parents 3 Demography Residents 3 Demography Parents 3 Residents 3 Demography Error Error df R2
MS
12.02§ 11.42 0.80 4.57 20.08 4.02 0.00 7.49
F 1.60 1.52 0.11 0.61 2.68\ 0.54 0.00
80 0.080
ii) Age (d) MS
185.14§ 36.59 21.78 5.36 11.86 145.87 2.81 42.08 88 0.110
Duration of the interbrood intervals (d) iii) First interval (covariate 5 Length 2†)
F
MS
F
4.40* 0.87 0.52 0.13 0.28 3.47\ 0.07
319.47§ 9.44 33.22 8.62 5.17 2.49 1.26 17.59 10.68
29.90*** 0.88 3.11\ 0.81 0.48 0.23 0.12 1.65
66 0.374
Notes: Influences of the three experimental factors: origin of the female’s parents (Parents), origin of the residents (Residents), and demography of the population (Demography) on female traits. Degrees of freedom 5 1 in each case except for the error term; the degrees of freedom for the error term are shown after the error mean square. * P , 0.05, ** P , 0.01, *** P , 0.001. † Covariate used in the ANOVA. Covariates were the length, mass, or age of the female when she produced the indicated brood. All covariates were transformed using natural logarithms. ‡ Parents 5 origin of the test male’s parents (Rivulus or Crenicichla locality). Residents 5 source of the tank residents (Rivulus or Crenicichla locality). Demography 5 population demography. § MS 3 1000. \ 0.05 , P , 0.10.
Later, they were dried and weighed so that reproductive allotment (dry mass of embryos)/(dry mass of embryos 1 somatic dry mass of female) could be calculated (see Reznick and Endler 1982 for details).
Measurement of male life history traits We evaluated the growth and development rates of the test males as well as their final ages and sizes at maturity. Males were weighed and their standard length measured the 1st wk after they were introduced into the treatment tanks, then every other week until the process of maturation was initiated, on the day they initiated the maturation process (when the number of segments on the third ray on the anal fin increased from 9 to 10 [Reznick 1990]), and then weekly until they reached maturity. If males had more than 10 segments when initiation was first observed, we back-calculated the age and size at initiation based on the subsequent relationship between age and segment number (Reznick 1990). During the final phase of maturation, individuals were checked daily to determine the age and size at maturation. Maturation was scored as in Reznick (1990). The duration of the maturation interval was the number of days between initiation and maturation.
Statistical analyses The experiment was designed for a three-way analysis of variance (Sokal and Rohlf 1981, SAS Institute 1989). The three independent variables were parents, residents, and demography. Each independent variable was represented by two levels, either Rivulus or Crenicichla localities. Type III sums of squares were used
since sample sizes were unequal among cells (SAS Institute 1989). Where necessary, data were transformed to meet the assumptions of the analyses (Sokal and Rohlf 1981, Zar 1984). For males, the following variables were transformed using natural logarithms: mass at initiation of maturation, duration of the maturation interval, growth rate during the maturation interval, and length and mass at maturity. Length at initiation was transformed by taking the square root. The reciprocal of ages at initiation and maturation were used. For females, all length and mass measurements, age at maturity, durations of interbrood intervals, growth rate during the first interbrood interval, and numbers of offspring per brood were transformed using natural logarithms. Least squares means by treatment for all traits are provided in Rodd (1994). Several variables (length, mass, and age at the time of the production of current and previous broods) were evaluated as potential covariates for female traits following Reznick (1989). All covariates that satisfied the assumptions of ANCOVA (homogeneity of slopes and variances) and that accounted for significant variation were included in the reported analyses (Table 1). Where none of the main effects nor interactions in a particular ANOVA were, nor approached (0.05 , P , 0.10), statistical significance, the ANOVA is not presented here. This was true for the following variables: female mass for the first brood, female mass and length for the second and third broods, fecundity in the first brood, offspring mass (dry mass and lean dry mass) for the first and second broods, and female growth rates during the first and second interbrood in-
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
March 1997
423
TABLE 1. Continued. Duration of the interbrood intervals (d) iv) Second interval (covariate 5 Length 3) MS
102.42§ 10.55 1.76 29.84 4.75 2.97 11.10 1.47 6.59
Mass of the offspring in the third brood (mg)
Fecundity (number) v) Second brood (covariate 5 Age 1†)
vi) Third brood (covariate 5 Length 3†)
vii) Mean dry mass (covariate 5 Length 1†)
F
MS
F
MS
F
MS
F
15.53*** 1.60 0.27 4.53* 0.72 0.45 1.68 0.22
1.41 0.23 0.10 0.60 0.00 0.02 1.34 1.04 0.24
5.76* 0.93 0.43 2.45 0.01 0.07 5.45* 4.25*
1.10 1.52 1.83 0.78 1.09 0.09 0.06 0.01 0.24
4.66* 6.43** 7.75** 3.29\ 4.61* 0.39 0.24 0.04
29.95 26.87 19.33 5.05 15.68 1.17 0.40 1.60 6.68
4.48* 4.02* 2.89\ 0.76 2.35 0.18 0.06 0.24
34 0.452
72 0.225
38 0.330
tervals. The results of those ANOVAs are presented in Rodd (1994). Interpopulation differences in reaction norms may not be detected if only analyses of variance are used (Sultan 1987). Therefore, we used regression analyses to examine the responses of test males to a range of demographic conditions. It was possible to do these analyses because of short-term deviations from the original experimental design caused by the presence of other test individuals in the tanks. For example, adult female numbers increased when adult test females were still in the treatment tanks, waiting to produce their second and third broods. Adult male numbers increased when older test males were awaiting tests of their sexual behavior. Even with such variations, the total densities in aquaria with a Crenicichla-type demography were always lower than those in Rivulus-type demographies (see Rodd and Sokolowski 1995 for more details). We performed multiple regressions for each male
36 0.246
dependent variable with the numbers of mature females, mature males, and immature animals in the test individual’s tank as the independent variables (PROC GLM, SAS Institute 1989). For each test male, we counted the numbers of guppies in each category in his tank 1 wk before he initiated the maturation process. We used this date as a rough estimate of the time when young males would be sensitive to interactions with conspecifics. The numbers of individuals in each category ranged from: immatures: 2–5, mature males: 2– 7, mature females: 2–7. To ensure that the assumptions of this test were met, we checked residuals and tolerances and verified that correlations among the numbers of guppies in each category were not significant (Sokal and Rohlf 1981, Zar 1984). We also checked for interactions among the independent variables; interactions that were not significant were excluded from the regressions. We further examined the influences of conspecifics on male traits by doing analyses of covariance with the
TABLE 1. Continued. Reproductive allotment Mass of the offspring in the third brood (mg) viii) Mean dry mass (covariate 5 Age 1†)
ix) Mean, lean, dry mass (covariate 5 Age 1†)
Source
MS
F
MS
F
Covariate Parents Residents Demography Parents 3 Residents Parents 3 Demography Residents 3 Demography Parents 3 Residents 3 Demography Error
34.60 9.30 13.55 14.25 44.36 16.59 1.62 1.48 6.40
5.41* 1.45 2.12 2.23 6.93** 2.59 0.25 0.23
14.49 10.61 9.50 5.51 14.69 2.87 1.57 3.16 2.57
5.63* 4.12* 3.69\ 2.14 5.71* 1.12 0.61 1.23
Error df R2
39 0.262
39 0.286
x) Reproductive allotment for females that produced a third brood (covariate 5 Length 3†) MS
4.29§ 1.97 9.08 5.16 2.74 0.03 0.46 0.08 1.22 34 0.362
F 3.50\ 1.61 7.42** 4.22* 2.24 0.02 0.38 0.07
F. HELEN RODD ET AL.
424
Ecology, Vol. 78, No. 2
FIG. 1. Mean number and size (dry mass) (61 SE) of offspring produced in the third brood by females from Rivulus and Crenicichla localities when they were reared with residents from those localities. Demography did not have a significant influence on fecundity (i.e., these results were obtained irrespective of the density at which females were reared). Fecundity data were transformed using natural logarithms.
numbers of immature and/or mature male residents as the covariate (mature females were not used as a covariate since the multiple regressions indicated that they did not contribute to the variation in male traits). Again, we used Type III sums of squares. Where appropriate, data were transformed to comply with the assumptions of the analysis. RESULTS
Females Influence of parental origin.—Daughters of Crenicichla locality guppies were younger at first parturition (least squares mean: 122 d) than those from Rivulus localities (134 d) (Table 1:ii). There was no influence of any of the three main factors (parents, residents, nor demography) on female size at first parturition and they had little influence on offspring size and number in the first two broods (Table 1:i–v). Number of offspring in the second brood was the only exception: there was a significant interaction among all three experimental factors but there was no clear pattern to their effects. For characteristics of the third brood, there were sig-
nificant interactions between the origin of the parents and the origin of conspecific residents. Interlocality variation in plasticity and the influence of the origin of the residents.—Daughters of Rivulus locality guppies had fewer, larger offspring in their third broods when they were reared with guppies from Crenicichla localities than when they were reared with guppies from Rivulus localities (least squares means comparisons: offspring number: P 5 0.009; total offspring mass: P 5 0.023; lean offspring mass: P 5 0.017 [SAS Institute 1989]) (Figs. 1 and 2, Table 1:vi,–ix). Daughters of Crenicichla locality guppies did not respond differently to residents from different localities (offspring number: P 5 0.51; total offspring dry mass: P 5 0.26; lean offspring mass: P 5 0.65). One possible explanation for this response of Rivulus females to the origin of the residents is that the intensity of competition for food was greater for females living with Crenicichla locality residents and hence less food was available to them. A decrease in food availability will cause a decrease in female growth rates and fecundity and an increase in offspring size (Reznick and Yang 1993). To examine this possibility, we looked at
March 1997
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
425
FIG. 2. Natural log of mean mass at sexual maturity (61 SE ) of male Trinidadian guppies reared under different demographic conditions with resident guppies from different predator localities. (A) 5 sons of guppies from Rivulus localities. (B) 5 sons of guppies from Crenicichla localities. The original data for male size (mg) were multiplied by 10 prior to logtransformation.
the daily growth rates of Rivulus locality females. Growth rates were not significantly different between those females when they were reared with Rivulus locality residents vs. with Crenicichla locality residents (least squares means comparisons: growth rate during the first interbrood interval: P 5 0.60; growth rate during the second interbrood interval: P 5 0.92). This suggests that competition for food is not responsible for the differences in offspring size and number. The reproductive allotment (RA) of all guppies, regardless of locality of origin, was greater when they were reared with Rivulus residents (least squares mean 6 1 SE: 11.6 6 0.9%) than when they were reared with Crenicichla residents (7.5 6 1.1%) (Table 1:x). Influence of population demography.—The second interbrood interval (mean 6 1 SE) was significantly shorter (31.1 6 0.7 d) when females were reared in a Crenicichla-type demography (low density) than for those in a Rivulus-type demography (high density) (33.2 6 0.8 d) (Table 1:iv). RA was significantly greater for females reared in a Crenicichla-like demography (11.0 6 0.9%) than for those in a Rivulus-type demography (8.0 6 1.0%) (Table 1:x). These results are consistent with the pattern expected if guppies in the Crenicichla-type demography had more energy at their disposal than those in the Rivulus-like demography (Reznick and Yang 1993). However, growth rates did not differ between females from the two types of demography. We also compared a subset of test females from both demographic treatments so that only those with similar growth rates were considered. In this analysis, the influence of differences in growth rates (energy availability) on the life history traits of interest should be eliminated. There were still differences between females reared in the two demographic conditions (duration of the second interbrood interval: F1,33 5 4.37, P 5 0.044, n 5 42; RA: F1,33 5 4.12, P 5 0.048, n 5 42). Therefore, the differences in interbrood
interval and RA do not seem to be attributable to differences among females in resource acquisition.
Males Influence of parental origin.—Males (mean 6 1 SE) from Crenicichla localities were younger (61.4 6 2.7 d) and smaller (71.53 6 1.88 mg, 14.74 6 0.13 mm) when they initiated the maturation process than males from Rivulus localities (68.6 6 2.9 d, 80.73 6 1.91 mg, 15.38 6 0.13 mm) (Table 2). When all males were considered, the duration of the maturation interval (age at maturity minus age at initiation of maturity) was similar for Crenicichla and Rivulus locality males; when only those individuals with similar growth rates before initiation were considered, the interval was shorter for Crenicichla locality guppies than Rivulus locality guppies (i.e., they developed at a faster rate) (F1,76 5 4.04, n 5 84, P 5 0.048). Crenicichla locality males were younger at maturity (85.6 6 3.0 d) than Rivulus locality males (94.1 6 3.2 d) (Table 2); this result should be considered with caution given that the assumption of homogeneity of variances was not met for this variable, despite the use of several transformations (Sokal and Rohlf 1981, Zar 1984). Therefore, the rate of development was accelerated in Crenicichla locality fish compared with Rivulus locality fish. These results are consistent with earlier studies (Reznick and Endler 1982, Reznick and Bryga 1987, Reznick 1989). For growth rates during the maturation interval, there was an interaction among all three factors examined in this study but the most obvious pattern among them is that males from Crenicichla localities tended to grow faster (1.69 6 0.07 mg/d) than those from Rivulus localities (1.20 6 0.07 mg/d). Finally, for the mass at sexual maturity and the duration of the maturation interval, there was an interaction among all three factors. Interlocality variation in plasticity.—For mass at maturity, only males from Rivulus localities responded
F. HELEN RODD ET AL.
426
Ecology, Vol. 78, No. 2
TABLE 2. Analyses of variance on male life history traits. Maximum growth rate before the initiation of the maturation process (mg/d) Source Parents† Demography Residents Parents 3 Demography Parents 3 Residents Residents 3 Demography Parents 3 Residents 3 Demography Error Error df R2
MS
14.56 166.03 140.55 31.13 0.11 112.18 10.40 28.78 97 0.146
Characteristics at the initiation of maturation Length (mm)
F
MS
F
0.51 5.77* 4.88* 1.08 0.00 3.90* 0.36
2001.98‡ 18.50 0.28 76.98 289.51 475.92 63.12 168.05
11.91*** 0.11 0.00 0.46 1.72 2.83\ 0.38
118 0.127
Note: Influences of the three experimental factors: origin of the male’s parents (Parents), origin of the residents (Residents), and demography of the population (Demography) on male traits. Degrees of freedom 5 1 in each case except for the error term; the degrees of freedom for the error term are shown after the error mean square. * P , 0.05, ** P , 0.01, *** P , 0.001. † Parents 5 origin of the test male’s parents (Rivulus or Crenicichla locality). Residents 5 source of the tank residents (Rivulus or Crenicichla locality). Demography 5 population demography. ‡ MS 3 1000. § MS 3 100 000. \ 0.05 , P , 0.10.
to the treatments (Table 2, Fig. 2A). Residents from Rivulus localities exerted a greater influence on them than residents from Crenicichla localities and Rivulus locality residents were most influential when they were at high densities (least squares means comparisons: P , 0.0077, this is significant when compared with the Bonferroni-corrected P value of 0.0083 [Milliken and Johnson 1984, SAS Institute 1989]). On the other hand, males from Crenicichla localities did not show a significant response to the manipulations (P . 0.014 for all comparisons, this is not significant when compared with the corrected P value of 0.0083; Fig. 2B). The results for the duration of the maturation interval also suggest that males from Rivulus localities were more sensitive to our manipulations than those from Crenicichla localities for this trait. Influences of population demography and origin of the residents.—The duration (mean 6 1 SE) of the maturation interval was significantly shorter for males reared in the Crenicichla locality-type demography (24.4 6 0.6 d) than for those reared in the Rivulus locality-type demography (27.0 6 0.7 d). This difference remained significant even when we only examined individuals with similar growth rates (F1,82 5 5.85, n 5 84, P 5 0.018). For the rest of the traits that were influenced by demography (mass at the initiation of maturation, length at maturation, and the maximum growth rate attained before the initiation of maturation), there was an interaction between demography and the origin of the tank residents. If males were reared with residents from Crenicichla localities, population demography had no influence on these traits (least squares means comparisons: mass at initiation: P 5 0.65 [77.14 6 2.51 mg]; length at maturity: P 5
0.35 [17.05 6 0.19 mm]; maximum growth rate: P 5 0.75 [1.69 6 0.10 mg/d]). However, if males were reared with residents from Rivulus localities, population demography had a significant influence (least squares means comparisons: mass at initiation: P 5 0.0049; length at maturity: P 5 0.0038; maximum growth rate: P 5 0.0037). Males reared at high densities (Rivulus demography) grew more quickly before initiation (2.16 6 0.10 mg/d), were larger at the onset of maturation (80.54 6 2.69 mg), and were longer at the completion of maturation (17.69 6 0.20 mm) than males reared at low densities (Crenicichla demography) (1.69 6 0.12 mg/d, 69.69 6 2.90 mg, 16.84 6 0.21 mm). When we considered a subset of juveniles reared with Rivulus locality residents, those with similar maximum growth rates before initiation, the significance level for mass at initiation fell just below the critical value (F1,35 5 3.53, n 5 37, P 5 0.069) while length at maturity was still significant (F1,36 5 5.00, n 5 38, P 5 0.032). Given that the decline in samples sizes in these analyses is expected to reduce significance levels, they suggest that food consumption did not play a large role in these differences. To determine what aspect of population demography was influencing the size at maturity of males reared with Rivulus locality residents, we first used multiple regression analyses (see Methods) (Table 3) and then analyses of covariance (Table 4). Both analyses indicated that the length at maturity of guppies from both localities increased as the density of mature males increased when they were reared with residents from Rivulus localities; when they were reared with residents from Crenicichla localities, there was no such response. All males, including those reared with Creni-
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
March 1997
427
TABLE 2. Continued. Characteristics of the maturation interval (initiation to completion of maturation)
Characteristics at the initiation of maturation Mass (mg)
Age (d)
MS
F
421.44‡ 125.65 16.16 4.41 52.84 224.04 14.95 38.38
10.98** 3.27\ 0.42 0.11 1.38 5.84* 0.39
118 0.153
Duration (d)
F
MS
F
6.53** 0.00 1.98 1.40 2.06 0.02 0.61
69.56‡ 238.92 31.51 0.18 7.89 9.22 101.05 27.90
2.49 8.56** 1.13 0.01 0.28 0.33 3.62\
MS
0.12‡ 0.00 0.04 0.03 0.04 0.00 0.01 0.02
Growth rate (mg/d)
106 0.098
106 0.143
cichla locality residents, increased their length at maturity in response to an increase in the number of immature males. We could not interpret a similar analysis of male mass at maturity because of interactions among the independent variables. In a multiple regression on the maximum growth rate attained before initiation, for males reared with Rivulus locality residents, growth rates increased with the number of mature male residents. This increase was associated with larger sizes before the initiation of maturation as the density of mature males increased. Mature females and immature individuals had no influence on test males’ growth rates. For males reared with Crenicichla locality residents, the residents had no influence on growth rates. We found no effect of the manipulations of social conditions (demography and residents) on the ages of males at the initiation or completion of maturation.
F
MS
3.66 0.06 0.37 0.01 0.03 0.38 0.87 0.15
24.70*** 0.43 2.47 0.07 0.18 2.55 5.86*
105 0.225
icichla localities. Reznick and Endler found that Rivulus locality males were larger at maturity than Crenicichla locality males but that the difference was greater for guppies collected in the field than for those maintained in the laboratory for two generations (Reznick 1982, Reznick and Endler 1982; D. N. Reznick, unpublished data). We found that the size at maturity of males from Rivulus localities increased with density. Since guppy densities tend to be higher in Rivulus localities than in Crenicichla localities (Reznick and Endler 1982), the genetically based difference in size between guppies from the two localities would be enhanced by the effects of the differences in demography. For two other traits, reproductive allotment (RA) and interbrood interval, the plastic responses to differences in demography would also augment the genetic differences. Previous studies also found greater differences between Rivulus and Crenicichla locality guppies in the field than in the laboratory for female size at maturity and for offspring size (Reznick 1982, Reznick and Endler 1982). We did not detect an influence of our manipulations on female size at maturity. We did detect an effect of conspecifics’ origins on offspring size for daughters of Rivulus locality guppies and on repro-
DISCUSSION
Plasticity as an explanation for phenotypic variation in the field Our results suggest that phenotypic variation induced by population demography contributes to the life history differences between guppies in Rivulus and CrenTABLE 2. Continued.
Characteristics at maturity Length (mm) Source Parents† Demography Residents Parents 3 Demography Parents 3 Residents Residents 3 Demography Parents 3 Residents 3 Demography Error Error df R2
MS
0.74‡ 9.30 3.70 1.42 5.86 32.07 10.04 4.10 120 0.112
Mass (mg)
Age (d)
F
MS
F
0.18 2.27 0.90 0.35 1.43 7.83** 2.45
10.39‡ 170.50 12.02 3.33 75.82 247.32 237.29 36.07
0.29 4.73* 0.33 0.09 2.10 6.86** 6.58**
120 0.147
MS
5.25§ 0.18 0.95 0.85 1.03 0.02 0.10 0.54 111 0.115
F 9.81** 0.34 1.77 1.60 1.93 0.04 0.18
F. HELEN RODD ET AL.
428
Ecology, Vol. 78, No. 2
TABLE 3. Multiple regression analyses with tank residents (by sex and age category) as the independent variables and the life history trait as the dependent variable. Source
df
MS
A) Dependent variable 5 male length at maturity i) Crenicichla locality residents Mature females 1 4.31† Mature males 1 1.30 Immature 1 239.03 Error 66 44.71 R2 5 0.084, intercept 5 5.088 ii) Rivulus locality residents Mature females Mature males Immature Error R2 5 0.286, intercept 5 4.977
1 1 1 54
92.53† 118.52 175.02 29.99
F
P
Slope
0.10 0.03 5.35
0.757 0.865 0.024
20.0023 20.0014 0.0218
3.09 3.95 5.84
0.085 0.052 0.019
0.0094 0.0145 0.0282
B) Dependent variable 5 maximum growth rate before initiation of maturation i) Crenicichla locality residents Mature females 1 0.02† 0.00 0.997 Mature males 1 531.76 0.56 0.459 Immature 1 1410.77 1.47 0.230 Error 54 957.16 R2 5 0.040, intercept 5 2.499 ii) Rivulus locality residents Mature females Mature males Immature Error R2 5 0.226, intercept 5 2.114
1 1 1 43
108.00† 7868.72 1202.39 1176.99
0.09 6.69 1.02
0.763 0.013 0.318
20.0002 0.0326 0.0544
0.0114 0.1325 0.0778
Note: Analyses were run separately for males reared with Crenicichla locality residents and those reared with Rivulus locality residents. † MS 3 10 000.
ductive allotment; however, the effect we observed would appear to reduce the difference between Rivulus and Crenicichla locality guppies for these traits.
Interpopulation variation in plasticity For several traits, guppies from Rivulus localities expressed plasticity in response to our manipulations but those from Crenicichla localities did not. For example, females from Rivulus sites varied the number and size of their offspring in response to the type of residents with which they were reared. Females from Crenicichla sites showed no such response. However, females from Crenicichla localities modify the size of their offspring when reared in isolation on different levels of food availability, so they are capable of varying this trait (Reznick 1983, Reznick and Yang 1993). Males from Rivulus localities showed plasticity in two life history traits, mass at maturity and duration of the maturation interval, for which males from Crenicichla localities did not. This is in contrast to identical patterns of plasticity in size at maturity in males from both localities in response to varying food supply (Reznick 1990; D. N. Reznick, unpublished data). Rodd and Sokolowski (1995) also showed that males from Rivulus localities were more likely to express flexibility in courtship behavior than those from Crenicichla localities.
Why did Rivulus locality guppies show greater flexibility than Crenicichla locality guppies in response to the conditions tested here? We offer several possible explanations. First, environmental heterogeneity can select for phenotypic plasticity (Levins 1968, Via 1993). We have found that there is greater temporal and spatial variation in the sex ratios of adults in Rivulus localities than there is in Crenicichla localities (Rodd and Reznick 1997). Since sex ratio can influence the degree of competition for mates (e.g., Itzkowitz 1990, Madsen and Shine 1993), young males may use their social environment as an indicator of the degree of competition they will face when they mature and hence adjust their reproductive strategy (e.g., adult body size) accordingly (Schultz and Warner 1989). Greater variation in sex ratio among populations of guppies in Rivulus localities may thus select for increased plasticity in response to social environment. Another possibility is that the costs of plasticity (Johnston 1982, Alcock 1989, Lott 1991, Newman 1992) are higher for guppies living in Crenicichla localities. For example, high mortality rates combined with the delay in maturation necessary to attain a larger body size could offset the benefits of a larger size. Male mortality rates are substantially higher in Crenicichla localities than in Rivulus localities (Reznick et al. 1996). A third possibility is that there are reduced benefits of plasticity
March 1997
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
429
TABLE 4. Analyses of covariance to compare the responses of males from different treatments to varying numbers of conspecifics. Source of variation
df
MS
F
P
A) Dependent variable 5 male length at sexual maturity i) Four treatment groups according to source of parents and source of tank residents Immature (I)† 1 1099.64 8.92 0.003 Treatment (T) 3 57.83 0.47 0.704 3 68.40 0.55 0.646 I3T Error 120 123.27 Slope 5 4.20, intercept 5 159.45 Mature males (M)‡ Treatment (T) M3T Error
1 3 3 120
459.06 278.58 389.12 124.01
ii) Separate analyses for different sources of tank residents a) Rivulus locality residents Mature males (M) 1 1229.30 Origin of parents (O)§ 1 9.18 1 5.36 M3O Error 54 99.82 Slope 5 3.72, intercept 5 159.00 b) Crenicichla locality residents Mature males Origin of parents M3O Error Slope 5 20.86, intercept 5 174.32
1 1 1 66
98.27 1.44 0.02 143.80
3.70 2.25 3.14
0.057 0.086 0.028
12.31 0.09 0.05
0.0009 0.763 0.818
0.68 0.01 0.00
0.411 0.921 0.990
B) Dependent variable 5 maximum growth rate before maturation i) Four treatment groups according to source of parents and source of tank residents Mature males 1 408.55 16.30 0.0001 Treatment 3 57.14 2.28 0.084 3 105.81 4.22 0.008 M3T Error 97 25.07 R2 5 0.256 ii) Separate analyses for different sources of tank residents a) Crenicichla locality residents Mature males 1 13.20 Origin of parents 1 6.08 1 12.31 M3O Error 54 20.09 R2 5 0.023 b) Rivulus locality residents Mature males Origin of parents M3O Error R2 5 0.272
1 1 1 43
544.07 63.63 87.08 31.32
0.66 0.30 0.61
0.421 0.585 0.437
17.37 2.03 2.78
0.0001 0.161 0.103
† The number of immature conspecifics present in the male’s tank one week before he initiated the maturation process. ‡ The number of mature male conspecifics present in the male’s tank one week before he initiated the maturation process. § Rivulus or Crenicichla locality.
in Crenicichla localities. Females in Crenicichla localities are less likely to base their mate choice decisions on male size than those in Rivulus localities (Houde 1988, Reynolds and Gross 1992, Endler and Houde 1995) so the advantages of maturing at a larger size are reduced in Crenicichla localities. Also, there may be less male–male competition for mates in Crenicichla localities (Farr 1980). A fourth possibility, at least for male size at maturity, is that plasticity is well developed
in guppies from Rivulus localities because the phenotypic response to increased density, an increased size at maturity, enhances the genetic response to predators, an increased size at maturity. However, in Crenicichla localities, if the risk of mortality via predation increases with size, the genetic and plastic responses would conflict. Therefore, this may be a case where selection for plasticity by one factor may be counteracted by selection on the mean of the trait by a different factor.
430
F. HELEN RODD ET AL.
Plasticity in male life history traits We found that male development and size at maturity were influenced by interactions with conspecifics. These effects were not a result of competition for food because variation in growth rates did not explain the patterns of results. Also, our observations are very different from the effects of food availability on the maturation of males reared in isolation (Reznick 1990). Reznick found that the duration of the maturation interval was fixed, even at a wide range of food availabilities; here, we found that the duration varied with population demography. Previous studies have shown that large male guppies sire more offspring than smaller ones both because females tend to prefer them and because they have an advantage when competing with other males for access to females (Gandolfi 1971, Gorlick 1976, Houde 1988, Magurran and Seghers 1991, Kodric-Brown 1992, Reynolds and Gross 1992). Since male guppies grow little after they mature, body size during their reproductive life-span is essentially their size at maturity (Reynolds et al. 1993, F. H. Rodd, personal observation). In this study, males adjusted their size at maturity in response to the density, reproductive status, and origin of potential competitors. These cues should be reliable indicators of the degree of competition a male can expect to face because there was a short delay (3– 5 wk) between the time when we detected males’ sensitivity to cues from conspecifics (1 wk before the initiation of maturation) and the time when the newly matured males faced those male competitors and choosy females. If it looks as though there will be little competition, they mature early and small; if there will be a lot, they delay maturation and mature at a larger size. This pattern suggests that this plasticity is adaptive. Other poeciliid species also show an effect of agonistic interactions with male conspecifics on male age and size at maturity (Borowsky 1973, 1978, Sohn 1977a, b, Farr 1980, Hughes 1985, Campton and Gall 1988, Campton 1992). The general pattern is that inhibited males delay maturation and hence mature at a larger size than they would have otherwise (Borowsky 1973, 1987, Sohn 1977a, b, Farr 1980, Campton and Gall 1988, Campton 1992). As in this study, Campton (1992) also found increased growth rates of groupreared fish. Hormones, especially androgens, appear to be the mediator between agonistic male–male interactions and delayed sexual development in male poeciliids. Androgens are required to stimulate the initiation of the maturation process (Grobstein 1948) and increasing levels are needed to successfully complete maturation (Turner 1942, Kallman and Schreibman 1973). At higher concentrations, maturation is accelerated (Grobstein 1948). Androgens also retard growth in poeciliids (Pickford and Atz 1957, Clemens et al. 1966). For
Ecology, Vol. 78, No. 2
Xiphophorus helleri, androgen levels are reduced for ø24 h in mature males that lose a bout with another mature male (Hannes et al. 1984). Therefore, after every agonistic interaction, androgen levels are probably depressed in immature males and, as a result, maturation is delayed but growth rates remain high. As density increases, agonistic interactions per individual increase (Magurran and Seghers 1991) and androgen levels are likely reduced for longer and/or to lower levels. In this study, the responses of two traits were consistent with the ‘‘androgen’’ hypothesis: the duration of the maturation interval was longer and growth rates were higher at increased densities. Therefore, these results provide further evidence that hormones may be involved in the life history responses of male guppies to interactions with conspecifics. However, we found no evidence that the initiation of the maturation process was delayed at increased densities. This may have been due to the relatively short duration between the introduction of the test males to the treatments and the initiation of maturation. The size at maturity of focal males increased with the density of resident males when they were reared with Rivulus residents, but not when they were reared with Crenicichla residents. Because aggressive interactions among guppies are reduced in Crenicichla localities (Magurran and Seghers 1991), adult males from these sites may not suppress hormone production in immature males. Alternatively, there may be reduced benefits of an increased size in Crenicichla locality populations (see previous section). Plasticity in female life history traits All females reduced RA in response to being reared with Crenicichla locality residents. Females from Rivulus localities increased offspring size and reduced fecundity in response to Crenicichla locality residents. The observed patterns are similar to the ones expressed by guppies in response to variation in food availability (Reznick and Yang 1993). It is possible that females reared with Crenicichla locality residents acquire less food or expend more energy than those reared with Rivulus locality residents. The increase in interbrood interval and decline in RA in response to Rivulus-type demography were also consistent with variation in energy intake or expenditure (Reznick and Yang 1993). It is possible that energy intake varied with the manipulations because the feeding rates of small guppies were inhibited by larger ones (Nagoshi 1967). It is also possible that energy expenditure increased with density because of an increase in aggressive acts per guppy (Warren 1973, Magurran and Seghers 1991). However, if the plasticity we observed was simply a result of differences in energy budgets, we would expect to find differences in the growth rates of females living with the two types of residents or living in different densities. We found no such differences. This suggests that energy budgets are not involved directly. However, it
March 1997
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
is possible that as resource availability declines, the reduction may first appear as decreased allocation to reproduction (Reznick 1983) and only later as a reduction in growth. Alternatively, the observed patterns of plasticity in female life history traits may represent adaptations. For example, it is possible that females reared with Crenicichla locality residents and those reared at high densities reduce their RA because they are withholding resources for future survival and/or reproduction (Williams 1966, Warner 1984). This did not appear to be the case since we detected no differences in growth rates; however, it is possible that females were storing energy as fat (Reznick 1983), a parameter not measured in this study. Variation in offspring size is another example where plasticity may be an adaptation. Females from Rivulus localities may modify the size of their offspring in response to the behavior of residents when larger offspring would have a significant advantage over smaller ones (e.g., larger offspring may be produced in response to competitive conspecifics) (Brockelman 1975, Lloyd 1987, Hutchings 1991, Gliwicz and Guisande 1992). As in an earlier study (Reznick 1982), we found that daughters of Crenicichla locality guppies matured earlier than those of Rivulus locality guppies. However, we found no evidence of effects of residents nor of demography on female age and size at maturity nor on the first two broods produced by females. This may have been because some females were isolated well in advance of maturation and hence they were not exposed to the experimental manipulations during the period when they were maturing, producing their first brood, and initiating the allocation of resources for their second brood (Reznick and Yang 1993).
Conclusions Our results suggest that the guppies’ phenotypically based responses to population demography are contributing to the observed association between predator community and guppy life history traits. Therefore, the predators are contributing to the patterns in guppy life history traits both indirectly and directly. Predators are indirectly responsible for the patterns because much of the variation in population demography, to which the guppies are responding, is caused by differences in predation pressure among localities (Rodd and Reznick 1997). Predators are also directly responsible for the patterns because some of the life history trait differences have evolved in response to the differences in predation pressure (Reznick 1982, Reznick and Bryga 1987, Reznick et al. 1990). The guppies’ phenotypic responses to differences in population demography would augment the genetic differences between guppies from Rivulus and Crenicichla localities. However, this plasticity may not have evolved directly to enhance fitness in the face of predation pressure. Rather, it has probably evolved to im-
431
prove an individual’s reproductive success under varying social conditions. For example, when competition for mates is low, males mature early at a small size; when competition is stiff, males delay maturity, mature at a larger size, and possibly enhance their ability to procur mates. ACKNOWLEDGMENTS We are grateful to Steve Morey, Jeff Leips, Anne Houde, Pat Ross, Anne Magurran, Kim Hughes, Joel Trexler, Mark Blows, Mark Simpson, Locke Rowe, John Reynolds, Laurence Packer, Don McQueen, Gary Mittelbach, and a reviewer for discussions and/or comments on the manuscript; Heather Bryga and Celia Rodd for advice and field assistance; Kim Hughes, Karen Grant, and Susan Hyndman for assistance with feeding the fish; and Atul Sharma for technical support. Thanks to all of them. This project was supported by NSERC (Canada) grants to M. Sokolowski, NSF (USA) grants to D. Reznick (BSR 88-18071 and DEB 91-19432), the Government of Ontario, York University Faculty of Graduate Studies, and Sigma Xi. LITERATURE CITED Alcock, J. 1989. Animal behavior. An evolutionary approach. Fourth edition. Sinauer, Sunderland, Massachusetts, USA. Borowsky, R. L. 1973. Social control of adult size in males of Xiphophorus variatus. Nature 245:332–335. . 1978. Social inhibition of maturation in natural populations of Xiphophorus variatus (Pisces: Poeciliidae). Science 201:933–935. . 1987. Genetic polymorphism in adult male size in Xiphophorus variatus (Atheriniformes: Poeciliidae). Copeia 1987:782–787. Brockelman, W. Y. 1975. Competition, the fitness of offspring, and optimal clutch size. American Naturalist 109: 677–699. Campton, D. E. 1992. Heritability of body size of green swordtails, Xiphophorus helleri: I. Sib analyses of males reared individually and in groups. Journal of Heredity 83: 43–48. Campton, D. E., and G. A. E. Gall. 1988. Effect of individual and group rearing on age and size at maturity of male mosquitofish, Gambusia affinis. Journal of Fish Biology 33: 203–212. Clemens, H. P., C. McDermitt, and T. Inslee. 1966. The effects of feeding methyl testosterone to guppies for sixty days after birth. Copeia 1966:280–284. Dahlgren, B. T. 1979. The effects of population density on fecundity and fertility in the guppy, Poecilia reticulata (Peters). Journal of Fish Biology 15:71–91. . 1985. Relationships between population density and reproductive success in the female guppy, Poecilia reticulata (Peters). Ekologia Polska 33:677–703. Endler, J. A. 1978. A predator’s view of animal color patterns. Evolutionary Biology 11:319–364. . 1980. Natural selection on color patterns in Poecilia reticulata. Evolution 34:76–91. . 1986. Natural selection in the wild. Princeton University Press, Princeton, New Jersey, USA. . 1995. Multiple-trait coevolution and environmental gradients in guppies. Trends in Ecology and Evolution 10: 22–29. Endler, J. A., and A. E. Houde. 1995. Geographic variation in female preferences for male traits in Poecilia reticulata. Evolution 49:456–468. Farr, J. A. 1975. The role of predation in the evolution of social behavior of natural populations of the guppy Poecilia reticulata (Pisces: Poeciliidae). Evolution 29:151–158.
432
F. HELEN RODD ET AL.
. 1980. The effects of juvenile social interaction on growth rate, size and age at maturity, and adult social behavior in Girardinus metallicus Poey (Pisces: Poeciliidae). Zeitschrift fur Tierpsychologie 52:247–268. Feltmate, B. W., and D. D. Williams. 1991. Evaluation of predator-induced stress on field populations of stoneflies (Plecoptera). Ecology 72:1800–1806. Gandolfi, G. 1971. Sexual selection in relation to the social status of males in Poecilia reticulata (Teleostei: Poeciliidae). Bollettino di Zoologia 38:35–48. Gliwicz, Z. M., and C. Guisande. 1992. Family planning in Daphnia: resistance to starvation in offspring born to mothers grown at different food levels. Oecologia 91:463–467. Gorlick, D. L. 1976. Dominance hierarchies and factors influencing dominance in the guppy Poecilia reticulata (Peters). Animal Behaviour 24:336–346. Grobstein, D. 1948. Optimum gonopodial morphogenesis in Platypoecilus maculatus with constant dosage of methyl testosterone. Journal of Experimental Zoology 109:215– 237. Hannes, R. P., D. Franck, and F. Liemann. 1984. Effects of rank-order fights on whole-body and blood concentrations of androgens and corticosteroids in the male swordtail (Xiphophorus helleri). Zeitschrift fur Tierpsychologie 65: 53–65. Haskins, C. P., E. F. Haskins, J. J. A. McLaughlin, and R. E. Hewitt. 1961. Polymorphism and population structure io Lebistes reticulatus, a population study. Pages 320–395 in W. F. Blair, editor. Vertebrate speciation. University of Texas Press, Austin, Texas, USA. Houde, A. E. 1988. The effects of female choice and male– male competition on the mating success of male guppies. Animal Behaviour 36:888–896. Hughes, A. L. 1985. Male size, mating success, and mating strategy in the mosquitofish Gambusia affinis (Poeciliidae). Behavioral Ecology and Sociobiology 17:271–278. Hutchings, J. A. 1991. Fitness consequences of variation in egg size and food abundance in brook trout Salvelinus fontinalis. Evolution 45:1162–1168. Itzkowitz, M. 1990. Sex-ratio biasing and male reproductive variation in a coral-reef fish. American Naturalist 136:557– 559. Johnston, T. D. 1982. Selective costs and benefits in the evolution of learning. Advances in the Study of Behavior 12: 65–106. Kallman, K. D., and M. P. Schreibman. 1973. A sex-linked gene controlling gonadotrop differentiation and its significance in determining the age of sexual maturation and size of the platyfish, Xiphophorus maculatus. General and Comparative Endocrinology 21:287–304. Kodric-Brown, A. 1992. Male dominance can enhance mating success in guppies. Animal Behaviour 44:165–167. Levins, R. 1968. Evolution in changing environments. Some theoretical explorations. Princeton University Press, Princeton, New Jersey, USA. Liley, N. R., and B. H. Seghers. 1975. Factors affecting the morphology and behaviour of guppies in Trinidad. Pages 92–118 in G. P. Baerends, C. Beer, and A. Manning, editors. Function and evolution in behaviour. Oxford University Press, Oxford, UK. Lloyd, D. G. 1987. Selection of offspring size at independence and other size-versus-number strategies. American Naturalist 129:800–817. Lott, D. F. 1991. Intraspecific variation in the social systems of wild vertebrates. Cambridge University Press, Cambridge, UK. Luyten, P. H., and N. R. Liley. 1991. Sexual selection and competitive mating success of male guppies (Poecilia reticulata) from four Trinidad populations. Behavioral Ecology and Sociobiology 28:329–336.
Ecology, Vol. 78, No. 2
Madsen, T., and R. Shine. 1993. Phenotypic plasticity in body sizes and sexual size dimorphism in European grass snakes. Evolution 47:321–325. Magurran, A. E., and B. H. Seghers. 1990. Risk sensitive courtship in the guppy (Poecilia reticulata). Behaviour 112: 194–210. Magurran, A. E., and B. H. Seghers. 1991. Variation in schooling and aggression amongst guppy (Poecilia reticulata) populations in Trinidad. Behaviour 118:214–234. Magurran, A. E., and B. H. Seghers. 1994. Sexual conflict as a consequence of ecology: evidence from guppy, Poecilia reticulata, populations in Trinidad. Proceedings of the Royal Society of London B 255:31–36. Mattingly, H. T., 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. McLain, D. K. 1992. Population density and the intensity of sexual selection on body length in spatially or temporally restricted natural populations of a seed bug. Behavioral Ecology and Sociobiology 30:347–356. Milliken, G. A., and D. E. Johnson. 1984. Analysis of messy data. Lifetime Learning, Belmont, California, USA. Nagoshi, M. 1967. Experiments on the effects of size hierarchy upon the growth of guppy (Lebistes reticulata). Journal of the Faculty of Fisheries Prefectural University of Mie 7:167–189. Newman, R. A. 1992. Adaptive plasticity in amphibian metamorphosis. BioScience 42:671–678. Peckarsky, B. L., C. A. Cowan, M. A. Penton, and C. Anderson. 1993. Sublethal consequences of stream-dwelling predatory stoneflies on mayfly growth and fecundity. Ecology 74:1836–1846. Pickford, G., and J. W. Atz. 1957. The physiology of the pituitary gland of fishes. New York Zoological Society, New York, New York, USA. Reynolds, J. D., and M. R. Gross. 1992. Female mate preference enhances offspring growth and reproduction in a fish, Poecilia reticulata. Proceedings of the Royal Society of London B 250:57–62. Reynolds, J. D., M. R. Gross, and M. J. Coombs. 1993. Environmental conditions and male morphology determine alternative mating behaviour in Trinidadian guppies. Animal Behaviour 45:145–152. Reznick, D. N. 1982. The impact of predation on life history evolution in Trinidadian guppies: Genetic basis of observed life history patterns. Evolution 36:1236–1250. . 1983. The structure of guppy life histories: the tradeoff between growth and reproduction. Ecology 64:862– 873. . 1989. Life-history evolution in guppies: 2. Repeatability of field observations and the effects of season on life histories. Evolution 43:1285–1297. . 1990. Plasticity in age and size at maturity in male guppies (Poecilia reticulata): an experimental evaluation of alternative models of development. Journal of Evolutionary Biology 3:185–203. Reznick, D. N., and H. Bryga. 1987. Life-history evolution in guppies (Poecilia reticulata): 1. Phenotypic and genetic changes in an introduction experiment. Evolution 41:1370– 1385. 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 IV, F. H. Rodd, and P. N. Ross. 1996. Life history evolution in guppies (Poecilia reticulata): 6. Differential mortality as a mechanism for natural selection. Evolution 50:1651–1660. Reznick, D. N., and J. A. Endler. 1982. The impact of pre-
March 1997
PLASTICITY IN GUPPY LIFE HISTORY TRAITS
dation on life history evolution in Trinidadian guppies (Poecilia reticulata). Evolution 36:160–177. Reznick, D. N., and A. P. Yang. 1993. The influence of fluctuating resources on life history: patterns of allocation and plasticity in female guppies. Ecology 74:2011–2019. Rodd, F. H. 1994. Phenotypic plasticity in the life history traits and sexual behaviour of Trinidadian guppies (Poecilia reticulata) in response to their social environment. Dissertation. York University, Toronto, Canada. Rodd, F. H., and D. N. Reznick. 1991. Life history evolution in guppies: III. The impact of prawn predation on guppy life histories. Oikos 62:13–19. Rodd, F. H., and D. N. Reznick. 1997. Variation in the demography of guppy populations: the importance of predation and life histories. Ecology 78:405–418. Rodd, F. H., and M. B. Sokolowski. 1995. Complex origins of variation in the sexual behaviour of male Trinidadian guppies, Poecilia reticulata: interactions between social environment, heredity, body size and age. Animal Behaviour 49:1139–1159. SAS Institute. 1989. SAS/STAT user’s guide, version 6. Fourth edition. SAS Institute, Cary, North Carolina, USA. Schultz, E. T., and R. R. Warner. 1989. Phenotypic plasticity in life-history traits of female Thalassoma bifasciatum (Pisces: Labridae). 1. Manipulations of social structure in tests for adaptive shifts of life-history allocations. Evolution 43: 1497–1506. Seghers, B. H. 1973. An analysis of geographic variation in the anti predator adaptations of the guppy, Poecilia reticulata. Dissertation. University of British Columbia, Vancouver, Canada. . 1974. Schooling behavior in the guppy (Poecilia reticulata): An evolutionary response to predation. Evolution 28:486–489.
433
Sohn, J. J. 1977a. Socially induced inhibition of genetically determined maturation in the platyfish, Xiphophorus maculatus. Science 195:199–201. . 1977b. The consequences of predation and competition upon the demography of Gambusia manni (Pisces: Poeciliidae). Copeia 1977:224–227. Sokal, R. R., and F. J. Rohlf. 1981. Biometry. Second edition W.H. Freeman, New York, New York, USA. Spitze, K., and T. D. Sadler. 1996. Evolution of a generalist genotype: multivariate analysis of the adaptiveness of phenotypic plasticity. American Naturalist 148:S108–S123. Strauss, R. E. 1990. Predation and life-history variation in Poecilia reticulata (Cyprinodontiformes: Poeciliidae). Environmental Biology of Fishes 27:121–130. Sultan, S. E. 1987. Evolutionary implications of phenotypic plasticity in plants. Evolutionary Biology 21:127–178. Turner, C. L. 1942. A quantitative study of the effects of different concentrations of ethynyl testosterone and methyl testosterone in the production of gonopodia in females of Gambusia affinis. Physiological Zoology 15:263–281. Via, S. 1993. Adaptive phenotypic plasticity: target or byproduct of selection in a variable environment? American Naturalist 142:352–365. Warner, R. R. 1984. Deferred reproduction as a response to sexual selection in a coral reef fish: a test of the life historical consequences. Evolution 38:148–162. Warren, E. W. 1973. The effects of relative density upon some aspects of the behaviour of the guppy—Poecilia reticulata (Peters). Journal of Fish Biology 5:753–765. Williams, G. C. 1966. Adaptation and natural selection. Princeton University Press, Princeton, New Jersey, USA. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, New Jersey, USA.