The cost of plasticity in Daphnia pulex.
One reason why an individual would fail to change its phenotype so as to always match the optimum is the costs of changing (Bradshaw 1965; DeWitt 1995; DeWitt et al. 1998). DeWitt et al. (1998) delineate five types of costs: maintenance costs, production costs, information acquisition costs, developmental stability costs, and genetic costs. In this paper we focus on the first two, maintenance and production costs. Maintenance costs are those associated with the costs of sensory and regulatory mechanisms producing plastic changes (DeWitt et al. 1998). To be a cost of plasticity, however, they must be costs incurred above and beyond those incurred by individuals with fixed phenotypes. They involve costs to the organism to retain the ability to change (to be plastic) even if no change occurs. As such, they are constant and borne by all plastic individuals regardless of the phenotype expressed in any particular environment.
Production costs are those associated with changing an organism's phenotype or producing structures normally not expressed. These costs are not simply those involved in producing a particular structure or morphology, however. They are costs that a plastic individual incurs that are not incurred by an individual with a fixed phenotype producing an identical structure. Past work has not distinguished between direct costs, those costs incurred in the formation of a structure or morphology whether by a fixed or plastic pathway, and plasticity costs, those costs incurred only by the plastic pathway.
Optimality models of plasticity evolution indicate that the existence of production and maintenance costs can change the outcome of evolution. An optimization model with production costs showed that a plastic strategy is favored at intermediate cost levels when cues indicating a change in the environment are reliable (Lively 1986). Very high or very low costs result in fixed strategies. A second optimality analysis suggests that maintenance costs may have little influence on the evolution of plasticity (Padilla and Adolph 1996). In contrast, a quantitative genetic model found that maintenance costs can result in nonoptimal reaction norms being favored (Van Tienderen 1991).
Natural selection should act to minimize costs, if possible. Direct costs may not be reducible, however. For example, a morphological structure will always require a minimal amount of materials for its construction. In contrast, plasticity costs may or may not be reducible. They might be, if they involved nothing more than alternative genes that turn on in specific circumstances. On the other hand, they might not be reducible if such costs involve the maintenance of receptor systems and internal signaling capabilities. Yet, if such systems had multiple functions within the organism, the costs attributable to plasticity would be minimized.
Direct costs and plasticity maintenance and production costs can be separately assessed through a series of correlation and regression analyses. We describe first the experimental system that we used to look for costs, then the general statistical models that we used to estimate those costs.
We measured costs using plasticity of the freshwater crustacean Daphnia pulex (Cladocera: Crustacea) in response to the presence of chemical signals from a predator, the insect Chaoborus americanus. Daphnia responds to a chemical stimulus released by larval Chaoborus, a common natural predator, by changes in the morphology of juveniles, the life stage most vulnerable to predation. In D. pulex these changes most noticeably involve the creation of bumps at the back of the neck (e.g., Dodson 1989; Schwartz 1991; Tollrian 1993), an increase in tailspine length, and a general increase in juvenile body size (Spitze and Sadler 1996). These changes function to lower the vulnerability of Daphnia by preventing Chaoborus from effectively handling the animal and increasing the probability of escape (Krueger and Dodson 1981). Spitze and Sadler (1996) demonstrated that plasticity in necktooth expression, fornix width, body length, and tailspine length were significantly correlated with decreased predation by Chaoborus in the laboratory. Induced Daphnia are less vulnerable and better able to break free of predatory strikes and grasps once attacked (Havel and Dodson 1984). Such effects should lead induced Daphnia to show a higher probability of surviving a predatory attack. Inducible clones are more likely to be found in ponds with Chaoborus (Parejko and Dodson 1991).
Previous studies suggest that this plasticity in Daphnia has a direct cost. Induced individuals showed an increased time to maturity (e.g., Havel and Dodson 1987; Vuorinen et al. 1989; Walls and Ketola 1989; Walls et al. 1991; Luning 1992; Black 1993; Tollrian 1995), a reduction in growth rate (Riessen and Sprules 1990), or a reduction in fecundity (Ketola and Vuorinen 1989; Walls and Ketola 1989; Walls et al. 1991; Luning 1992; see summary in Tollrian 1995). These effects manifest more at low food levels (Riessen and Sprules 1990; Walls et al. 1991; Tollrian 1995). The previous studies are inconsistent as to which traits demonstrated these costs, however, and some studies failed to find any costs (e.g., Spitze 1992). Finally, all of these studies only measured direct costs, rather than plasticity costs.
In our study, we looked at three traits of second-instar juveniles: body length, body depth (front to back distance), and tailspine length. Daphnia are most vulnerable to Chaoborus predation while small as juveniles, and changes in morphology are greatest in second-instar juveniles. We chose these traits, rather than neckteeth, because they involve much larger structures and so were more likely to show a cost. Because we were interested in demonstrating plasticity costs, rather than effects of Chaoborus predation, focusing on these body-size traits is reasonable.
Measuring direct costs is straightforward. Previous measures of these costs in Daphnia have mainly been done using single clones. The standard protocol is to raise multiple replicates of a single clone in the presence and absence on the inducing stimulus, and measure the difference between treatments in life history trait values. In contrast, our protocol is based on a within-treatment measure. Because our experiment involved multiple clones, we had to take two genetic factors into account. First, clones differed in juvenile morphology. Second, clones potentially differed in life history traits due to reasons other than juvenile morphology. Thus, we estimated the correlation between juvenile morphology and life history traits using a method that removed any genetic (among-clone) effects. In other words, we estimated the average within-clone correlation, the correlation that would be measured in a one-clone experiment, averaged over all the clones in the study.
Measuring plasticity costs is a bit more complicated. based on a model by Van Tienderen (1991), DeWitt (1998; DeWitt et al. 1998) recently proposed a conceptual framework and statistical model for such an analysis. The following description assumes that plasticity and its costs are assessed by raising replicates of multiple genotypes (clones, siblings, etc.) in two environments. All analyses are done using genotypic mean values. Plasticity is measured as the absolute value of the difference between environments in genotype mean values. Maintenance and production costs of plasticity are those costs that a plastic genotype incurs above and beyond the costs of a fixed genotype that produces exactly the same phenotype. These additional costs can be measured by a multiple regression using the following model:
W = X + [X.sup.2] + plX + X p [multiplied by] plX + [X.sup.2] - plX, (1)
where W is the measure of fitness in one of the environments, X is the trait value in that environment, and plX is the plasticity of the trait. The terms X and [X.sup.2] measure direct selection on the trait and account for the linear and nonlinear components of selection, respectively (Lande and Arnold 1983). The term plX measures plasticity costs, the effects on fitness once the direct effects are taken into account. A cost of plasticity appears as a negative regression coefficient. That is, more plastic genotypes will, on average, have lower fitness values. This term combines both maintenance and production costs. The interaction terms (X [multiplied by] plX and [X.sup.2] [multiplied by] plX) measure additional production costs. For example, a positive X [multiplied by] plX term would indicate that plasticity costs increase with the value of the trait. With respect to Daphnia, a significant interaction term might indicate that an additional cost is incurred by a plastic individual when producing a long tailspine, but no additional cost is incurred when producing a short tailspine. See Phillips and Arnold (1989) for a discussion of interaction terms in selection models. This regression is repeated for each environment. Differences in the cost of plasticity measured in the two environments might provide clues to the mechanistic basis for those costs. (See A Historical Note section in the Discussion for more details on the development of this model and an alternative conceptualization.)
As noted above, the plX term measures both maintenance and production costs. These costs can be separated only under special circumstances, when a trait is manifest or is variable in only one environment. For example, this can be done in systems in which phenotypic plasticity involves the turning on of an otherwise quiescent developmental system in response to a specific signal with the two environments being the presence and absence of the signal. Maintenance costs should exist in all environments, while production costs would exist only when the development system is cued to produce the new structure or morphology. So, the plX regression coefficient in the absence of the signal would measure maintenance costs, while the difference in the regression coefficient in the two environments would measure production costs.
MATERIALS AND METHODS
Daphnia pulex is a small (ca. 1-2 mm) freshwater herbivore. Reproduction occurs by one of two modes, direct developing young or diapausing eggs. Direct developing individuals are always produced parthenogenetically, one brood during each adult molt. The cycle is as follows. Shortly after the adult molts, she releases ova into her ovaries. After completing initial development, the eggs are deposited in the brood chamber on the back of the adult following her next molt. The eggs then develop into juveniles in the brood chamber and are released during the subsequent molt. So, at any one time the adult has offspring at three stages of development, unreleased ova, developing eggs in the ovaries, and developing juveniles in the brood chamber. Following release, the juveniles grow for three to five instars, depending on conditions, before becoming adults. The first adult instar is defined as the instar in which the first brood is deposited in the brood chamber. Adults may live up to 30 instars, but typically no more than four to six instars in the field. In the current experiment all reproduction was by direct development.
The 50 clones used in our experiment came from 22 populations in northern Illinois and Indiana. Our goal was to maximize genetic variation within our sample. Half of the clones came from 21 Illinois populations. From these populations only one or two clones were taken; if more than one was taken, they were determined to be electrophoretically unique. Previous work by others (Hebert et al. 1988; Lynch et al. 1989) and us (Scheiner and Yampolsky, in press) showed that populations in this region reproduce almost exclusively by parthenogenesis. The remaining clones were from a single, cyclically parthenogenetic population in northwestern Indiana, PA of Lynch et al. (1989). Again, each clone was electrophoretically unique. Clones were maintained individually in the laboratory in 200 mL of artificial pond water mixed with filtered, autoclaved pond water, and fed excess amounts of Scenedesmus acutus from cultures in exponential growth (Lynch et al. 1986). Densities of the stock cultures were not controlled.
Effects of Chaoborus Extract
We reared the 50 clones with and without Chaoborus extract. Each combination of clone and treatment was replicated twice. The Chaoborus extract was prepared from 200 g of commercially obtained frozen samples of C. americanus. The samples were boiled in 400 mL of water, filtered, and frozen in 30-mL lots, so that a single preparation was used during the entire experiment (Hebert and Grewe 1985). For the induced treatment, extract was added to the food at a rate of 0.05 mL/L. Water was changed and food was supplied daily at 0.50 [[micro]gram]C/mL (5 x [10.sup.4] cells/ml) of S. acutus for both treatments. This food level was low (ca. 20% of "well fed" levels) and chosen because costs are more likely to appear when an individual is energy limited. Cell density was determined with a spectrophotometer and supplied from a single daily mass sample to ensure uniformity across treatments. Temperature was 22 [degrees] C, and light regimes were 14:10 L:D, with light supplied by "daylight" fluorescent tubes.
To control for maternal effects, measurements were done on individuals whose mothers were raised in the same treatment. For each clone, from the stock culture we took four neonates and randomly assigned them to one of the two treatments. Usually the neonates came from a single clutch, although if individuals died, they were replaced from other clutches. These animals were raised individually in a shell vial with ca. 25 mL of water in the absence or presence of extract. From each of these individuals one offspring from her second clutch was used for the experiment. These individuals were raised under the same conditions as the previous generation and measured for the following traits: length at birth, length at second juvenile instar, tailspine length at second juvenile instar, body depth at second juvenile instar, number of hours to maturity, length at maturity, and for each of the first four clutches, time of birth, length of mother, clutch size, and length at birth of three offspring. The second juvenile instar was measured because previous studies showed that induction responses are greatest at this instar (Tollrian 1993, 1995). Maturation time and interclutch intervals were determined to the nearest six hours by monitoring individuals daily and correcting molt time using the development stage of the individuals in the brood chamber (Lynch et al. 1989). Length measurements were done with a LASICO ocular filar on a Wild stereomicroscope and an S-4A Auto-processor. Lengths were converted to biomass using the following, empirically derived, relationship based on individuals raised at high food levels: [log.sub.10](mass, [[micro]gram]) = 0.788 ([+ or -] 0.124) + 2.481 ([+ or -] 0.484) [log.sub.10](length, mm) (n = 39, [r.sup.2] = 0.40). Values in parentheses are 1 SE. Mass gain of the mother during each adult instar was determined as the difference in estimated masses based on her length following release of the clutch into the brood pouch and her length for the previous instar. Clutch mass was estimated using the mean offspring length, converted to biomass, times the clutch size. Reproductive effort was calculated as clutch mass divided by clutch mass plus growth mass increment. The intrinsic rate of increase (r) was calculated using the stable-age (Euler's) equation and information from the first four clutches; estimates of r for D. pulex asymptote after about four adult instars (Riessen and Sprules 1990). For daphnids living in seasonal ponds, the intrinsic rate of increase is the best composite measure of fitness because these populations are mostly expanding. Our results, however, hold for all measured fitness traits.
We estimated routine metabolic rate of individual adult Daphnia using a polaragraphic oxygen sensor (Strathkelvin Instruments, Glasgow, Scotland). Our apparatus consisted of an oxygen sensor (SI 1302) attached to an oxygen meter (OM 780) and a small, water-jacketed, respirometry chamber (model RC 300). Percent oxygen depletion and temperature were recorded continuously using a computerized data acquisition system (Sable Systems, Salt Lake City, UT). Water at a constant temperature (22 [+ or -] 0.01 [degrees] C) was flushed through the jacket of the respirometry chamber. Oxygen-free and saturated solutions of fresh water for calibration were obtained by bubbling nitrogen or room air through the water, and the apparatus was calibrated several times a day. All measurements were made at 22 [degrees] C. For each measurement, an individual was placed in the respirometry chamber together with 0.40-0.75 mL of water containing algae at the same concentration used to rear the female. Oxygen depletion rates were recorded for 10-15 minutes, and then the individual was removed for length measurements. To correct for background oxygen consumption, presumably due to algal and bacterial respiration, baseline recordings of oxygen consumption were recorded before each measurement. Individuals were given a five-minute equilibration period prior to measurement and pilot tests indicate that somewhat longer equilibration did not dramatically alter our estimates of metabolic rate. Much longer equilibration is impossible because of the small size of the cuvette. Body mass data were obtained by measuring body length immediately following the oxygen consumption measurements and using the length-mass conversion formula given above. The resulting measurements gave values similar to those reported previously for D. pulex (e.g., Richman 1958). Most individuals were measured following the release of their fourth clutch, although a few individuals were measured following release of the second clutch. The individuals measured were a subsample of those measured for morphological and life-history traits. Our measurements represent an estimate of the overall rate of energy expenditure of an actively reproducing individual swimming in its normal environment.
Of the original 2 x 50 experimental design, three clones failed to produce second-generation offspring or died before maturation. Analyses of all traits were done as 2 x 47 two-way ANOVAs or ANCOVAs with SAS procedure GLM (SAS Institute 1989). Treatment (environment) was treated as a fixed effect and clone (genotype) was treated as a random effect. Type III sums-of-squares were calculated; F-tests were done using the "Test" option, which computes a Satterthwaite correction for unbalanced designs. For some traits, ANCOVAs were used to correct for size effects. Length at maturity was corrected for length at birth and adult growth and offspring number; length and mass were corrected for the mother's length or mass at maturity. Interaction terms with the covariates were not significant for adult growth, offspring number, and offspring length. They were significant for length at maturity and clutch mass. However, the only change in conclusions was to indicate statistical significance for the clone term for length at maturity and the covariate term for clutch mass. The qualitative results were unchanged. Therefore, for simplicity, we present only the analyses without the interaction terms.
Direct costs were estimated as among-individual, within-clone correlations using procedure NESTED. The model used was: [X.sub.ij] [Y.sub.ij] = [[Lambda].sub.i] + [e.sub.ij], where [X.sub.ij] and [Y.sub.ij] are the trait values of the jth replicate of the ith clone, [[Lambda].sub.i] is effect of the ith clone, and [e.sub.ij] is the deviation of the jth replicate within the ith clone. The correlations of the within-clone deviations ([e.sub.ij]) measure direct costs averaged across clones. The correlations were estimated separately for each treatment. Direct costs are indicated as negative correlations between juvenile morphology and adult life-history traits. We performed separate analyses of the PA population and the parthenogenetic populations. [TABULAR DATA FOR TABLE 1 OMITTED] As results did not differ qualitatively with the analyses of the pooled populations, only the latter are presented.
Plasticity costs were estimated using the multiple regression procedure described in the Introduction using SYSTAT (vers. 6.1 for Windows, SPSS, Inc., Chicago, Illinois). Plasticity of juvenile body length, juvenile body depth, and juvenile tailspine length were calculated as the absolute difference in clone mean values in the presence and absence of the extract. Absolute values were used because, with regard to costs, the sign of the reaction norm slope should not matter. Conclusions did not differ, however, if signed values were used (results not shown). Because preliminary analyses failed to find evidence of nonlinear selection on the traits (nonsignificant [X.sup.2] terms), analyses were done using a simplified model: W = X + plX + X [multiplied by] plX. By excluding the nonsignificant nonlinear terms, we increased the power of the analyses to detect any plasticity costs. To be as liberal as possible in our search for costs, we did not correct for multiple tests. Again, results did not differ qualitatively between the overall analyses and those of the PA and parthenogenetic populations, so only the former are presented.
Juvenile morphology changed in response to the Chaoborus extract. Individuals grown in the presence of the extract were smaller as juveniles in both length and depth, but had longer tailspines, than individuals grown in its absence (Tables 1, 2). In the presence of the extract, tailspines were 29% of body length, while they were only 27% in its absence. For juvenile body length, variation was found among treatments and for the treatment-clone (G x E) interaction. For juvenile body depth, there was a significant treatment-clone interaction. Plasticity was measured as the difference in treatment [TABULAR DATA FOR TABLE 2 OMITTED] means for each clone. Plasticity of body length and depth were genetically correlated (among-clone mean correlation r = 0.87, P = 0.0001, N = 42). For tailspine length, variation was found among clones (genetic); the treatment-clone interaction was marginally nonsignificant (P = 0.08). Because we want to be as liberal as possible in our search for plasticity costs, we continued to analyze this trait despite the low amount of G x E variation. Plasticity of tailspine length and juvenile body length were genetically correlated (r = 0.68, P = 0.0001, N = 42), as were tailspine length and body depth (r = 0.70, P = 0.0001, N = 42). With regard to overall growth pattern, in the presence of the extract individuals were smaller up to maturity. However, adults in the extract treatment converged in size with those in the no-extract treatment over the next four instars [ILLUSTRATION FOR FIGURE 1 OMITTED]. All other morphological and life-history traits exhibited some combination of treatment, clone, and treatment-clone interaction effects. In both treatments fecundity was low to very low, indicating that the animals were stressed, making it more likely that we would be able to detect costs if they were present.
We found no evidence for direct costs of juvenile structures in any treatment. Individuals raised in the presence of the extract had higher fecundities and greater intrinsic rates of increase than those raised without extract (Table 1). However, these differences across treatments are potentially confounded with genetic and genotype-environment interaction effects. We controlled for these confounding effects by calculating average within-clone correlations between life-history traits and juvenile body length, juvenile body depth, or juvenile tailspine length (Table 3). For all three traits, there was a significant negative correlation with time to maturity in the absence of the extract, opposite that expected if there were direct production costs. That is, individuals with longer tailspines or larger bodies matured faster. For tailspine length, there was a negative correlation between tailspine length and reproductive effort, again in the absence of the extract. This negative correlation might be indicative of a cost; individuals with longer tailspines devoted fewer resources to reproduction. For the other two traits, correlations with reproductive effort were of the same sign as for tailspine length, but were not statistically significant.
TABLE 3. Measurement of direct costs for juvenile traits. For each treatment, the average within-clone correlations of life history traits with juvenile body length, juvenile body depth, or juvenile tailspine length are shown. Significant (P [less than] 0.05) effects are shown in bold. Chaoborus extract Trait Absent Present Juvenile body length Time to maturity -0.52 0.29 Mean clutch size 0.11 -0.04 Reproductive effort -0.17 0.19 Intrinsic rate of increase 0.28 0.09 Juvenile body depth Time to maturity -0.47 0.34 Mean clutch size 0.02 -0.15 Reproductive effort -0.23 0.15 Intrinsic rate of increase 0.26 0.13 Juvenile tailspine length Time to maturity -0.34 -0.01 Mean clutch size -0.09 0.29 Reproductive effort -0.36 0.24 Intrinsic rate of increase -0.08 0.29
In the presence of the extract, time to maturity was positively correlated with both body length and depth, possibly indicating a cost. These correlations were not statistically significant, however. The minimal detectable correlation for these traits was 0.39. If a one-tailed test were used, then the minimal detectable correlation was 0.33. Similarly, if the sample size were increased to three replicates per treatment, these correlations would be statistically significant. So, under the most liberal statistical criteria - no correction for multiple tests and a one-tailed test - there is limited evidence for a cost to time to maturity. More importantly, however, this effect on time to maturity did not result in an effect on the intrinsic rate of increase.
No correlation was found in either treatment for any trait with the intrinsic rate of increase. Five of the six correlations were positive, opposite that expected if there were a cost. The single negative correlation (tailspine length in the absence of the extract) was small.
Costs of Plasticity
We found scant evidence for plasticity costs either in the presence or absence of the Chaoborus extract, despite using a liberal statistical criterion of no correction for multiple tests. Maintenance and production costs were measured as the regression of life-history traits against trait plasticity. For both tailspine length and juvenile body length, none of the regression coefficients were statistically significant (Table 4). For the main effect of plasticity, the only regression coefficients that approached statistical significance were in the no-extract treatment for juvenile body length against mean clutch size and reproductive effort (P = 0.08). Most of the other coefficients did not even approach statistical significance. For juvenile body depth against reproductive effort, the regression coefficients did differ significantly from zero. However, the probability values were relatively large (P = 0.03 and P = 0.02) and any correction for multiple tests would declare these not significant. Most importantly, again the intrinsic rate of increase was not statistically significant for any of the traits in either treatment (P [greater than] 0.11) and five of the six coefficients were positive, opposite that expected if a cost existed. Nor were there significant separate production costs as measured by the trait-plasticity interaction term of the regression. The lack of significant costs was not due to a lack of genetic variation for either trait plasticity (G x E variation) or the life-history traits (G or G x E variation) (Table 2). Nor was the lack of costs due to low power; even an infinite sample size would have still resulted in nonsignificant conclusions for all of these traits.
[TABULAR DATA FOR TABLE 4 OMITTED]
Patterns of Reproduction
Several of the treatment effects were unexpected. Clutch sizes were twice as large in the presence of the extract than in its absence (Table 2). The puzzle to be solved is why fecundity increased in the presence of the extract. Typically in daphnids, fecundity increases with body size (Lynch 1989; Lampert 1993). Yet, animals in the no-extract treatment, with the lowest fecundities, were larger than those in the extract treatment (Table 1). Within treatments, we found a positive correlation between body size and fecundity for individuals in the extract treatment, and no correlation in the no extract treatment [ILLUSTRATION FOR FIGURE 2 OMITTED]. Adult mass gain (growth) was just slightly larger in the no-extract treatment compared to the extract treatment. Given the large differences in clutch mass, these results indicate that individuals in the extract treatment were able to put additional mass into reproduction without sacrificing growth. Where did these extra resources come from? The lack of direct costs was somewhat surprising given previous reports for their existence. One possible explanation is that differences in metabolic rates between treatments may have masked direct costs of juvenile structures.
The three legs of the resource allocation tripod are reproduction, growth, and maintenance. Maintenance resource costs can be measured by examining metabolic rates. We measured metabolic rates to try to obtain clues about the change in allocation strategy that allowed individuals reared in the presence of Chaoborus extract to increase reproductive output at equal body sizes.
Metabolic rate scaled isometrically with body mass in both treatments, slopes of log-log regressions of metabolic rate on mass were not significantly different (P [greater than] 0.05) from one. Therefore we analyzed mass-specific metabolic rate ([[micro]liter]/mg/h). Oxygen consumption rates decreased significantly in the presence of the extract (Tables 1, 2). This result suggests that changes in allocation patterns were responsible for the increased reproductive output. Mass specific metabolic rate was uncorrelated with any other morphological or life-history trait (results not shown).
Costs of Plasticity
We found scant evidence for either production or maintenance costs of plasticity for juvenile body length, juvenile body depth, or juvenile tailspine length. The lack of evidence for costs was not due to a lack of genetic variation, which would preclude their measurement. We also deliberately used low food levels to increase the chance of manifesting costs in energy limited individuals. Finally, we were statistically liberal in our search by not correcting for multiple tests and using tailspine length despite the lack of statistically significant treatment-clone interaction variation. Thus, our inability to find costs is likely robust. Our use of animals from multiple ponds does not effect our conclusion, as we are looking for an underlying cost that is independent of local adaptation. Rather, our use of multiple sources likely increased genetic and genetic-environment interaction variation in our sample, maximizing our ability to measure costs if they existed.
This study is one of only two attempts, that we are aware of, to measure maintenance or production costs of morphological plasticity. The other study (DeWitt 1995, in press) examined costs of plasticity in shell morphology of the snail Physa in response to the presence of either fish or crayfish predators. DeWitt also failed to find any evidence of plasticity costs. In contrast, a recent study of plastic responses to high temperature found significant maintenance costs. In a study of Drosophila melanogaster, Krebs and Feder (1997) measured isofemale lines for larval-to-adult survivorship at 25 [degrees] C and the magnitude of the induction of a heat shock protein (Hsp70) at 41 [degrees] C that increases high temperature survival. They report a negative correlation among lines between these two traits indicating that lines with a greater plastic response to an environmental stress suffered a cost even in the absence of protein expression.
A Historical Note
Here we detail the development of our multiple regression model for measuring plasticity costs so that proper credit can be given. As noted in the Introduction, the original concept derives from a model of Van Tienderen (1991). based on those ideas, DeWitt (1998; DeWitt et al. 1998) conceived of this analysis as a two-step process [ILLUSTRATION FOR FIGURE 3 OMITTED]. Again, analyses are done using genotypic mean values and plasticity is measured as the absolute value of the difference between environments in mean values.
For one environment, regress a measure of fitness against the trait value. Save the residuals from this analysis. Then regress those residuals against trait plasticity. A cost of plasticity appears as a negative regression. One of us (SMS) realized that the two-step process could be recast as a multiple regression. In addition, a multiple regression procedure would have an advantage over the two-step procedure because it permits the inclusion of interaction terms that could measure additional maintenance costs. These ideas concerning the multiple regression model were further refined based on extensive discussions with DeWitt. The final ideas presented in this paper derive directly from his model.
We found little evidence for direct costs of juvenile morphological structures. The lack of direct costs contradicts the findings of other studies. Almost all other studies found an increased time to maturity in the presence of Chaoborus extract, especially at low food levels (Havel and Dodson 1987; Ketola and Vuorinen 1989; Vuorinen et al. 1989; Walls and Ketola 1989; Black and Dodson 1990; Riessen and Sprules 1990; Walls et al. 1991; Luning 1992; Black 1993; Tollrian 1995). Age at maturity increased in the presence of extract in our study as well, however, the effect was not statistically significant due to the large treatment-clone interaction effect.
Effects on clutch size varied among previous studies. Some studies, like ours, found an increase in fecundity in the presence of extract (Black 1993; Tollrian 1995). Other studies found either no differences or decreased fecundities (Havel and Dodson 1987; Ketola and Vuorinen 1989; Walls and Ketola 1989; Black and Dodson 1990; Riessen and Sprules 1990; Walls et al. 1991; Luning 1992). A decrease in the intrinsic rate of increase was found in most studies, especially at low food levels, the exceptions being Black (1993) and Tollrian (1995). An increased length at birth was found in one study (Luning 1992), while no effect was found in others (Ketola and Vuorinen 1989; Vuorinen et al. 1989; Walls and Ketola 1989; Riessen and Sprules 1990; Tollrian 1995). One study found no change in tailspine length in the presence of Chaoborus extract (Black 1993), contrary to our findings. One study found an increase in body depth in the presence of Chaoborus extract (Tollrian 1995), contrary to our findings. However, all of these studies differed from ours in the use of only a single clone.
This is an important caveat because almost all the traits we measured showed significant treatment x clone interactions (Table 2). Differing results among past studies may be due to clonal variation. In contrast with the above studies, two studies have measured multiple clones. Spitze (1992) did a population survey including the same PA population used in our experiment while his other clones were from the same region as ours. In agreement with our study, in one experiment Spitze found that the presence of Chaoborus extract increased fecundity and the intrinsic rate of increase, although in a second experiment he found a decrease in the intrinsic rate of increase (cf. Table 1 with tables C1, D1 of Spitze 1992). Age at maturity did not change in one experiment, but decreased in the other. The former result is in agreement with our experiment. He found no effect on size at maturity, contradicting our findings, but did find a decrease in size at birth, in agreement with us. In all cases, even when his results were in the same direction as ours, the magnitudes of the effects were much greater in our experiment. The key difference between his experiments and ours was that our food level was just 17% of the level that he used. As a result of these husbandry differences, our animals were ca. 10-20% smaller at maturity, with clutch sizes ca. 60-90% smaller. Our lower food levels might explain the differences in results. Spitze did not test the effects of food level, but these differences in life-history traits across food levels are consistent with other studies (Lynch 1989; Riessen and Sprules 1990; Tollrian 1995).
Walls et al. (1997) used clones collected from 12 different ponds, using one per pond. They varied both food level and the presence of water conditioned by Chaoborus. Their low food level was comparable to our experiment. In the presence of the predator cue, at both high and low food levels, individuals matured at a larger size, opposite our results. There was no effect of the cue on the intrinsic rate of increase, again contradicting our findings. At low food levels, individuals had larger clutches in the presence of the cue, in agreement with our results. These individuals matured earlier, contradicting our results. In general, the effects of the Chaoborus cue was weak. They found genetic variation for most traits and some evidence for genetic variation in plasticity.
The totality of these studies reveal mixed evidence for direct costs. We agree with Spitze (1992) that there is a large amount of genetic variation for life-history traits in response to the presence of a predator in this species. In particular, there is a large amount of genotype-environment interaction variation. Such variation means that comparisons among experiments are problematical because differences in husbandry conditions can easily result in different outcomes. Attempts to compare single clones with each other or with population samples is even more problematic. Single clones are often chosen because they do well under laboratory conditions, particularly in the absence of the predator. Given genotype-environment variation, selection of a single clone might result in one that does well under standard laboratory conditions, but does poorly when induced in the presence of a predator. Publication bias, the failure to publish single-clone studies that fail to find a cost, would exacerbate this problem. In our experiment, for each trait at least one clone exhibited each of the responses to Chaoborus found by previous investigators.
We conclude, based on the totality of the evidence, that no overwhelming direct costs exist for D. pulex in response to the presence of Chaoborus. While some clones may show a cost, others do not. Again, the lack of evidence for costs was not due to a lack of genetic variation. Given that predation is less on juveniles with larger bodies (Havel and Dodson 1984; Spitze 1991), the question remains, given the lack of costs for a larger juvenile body, why are they actually smaller in the presence of the extract? We have two possible explanations: (1) a cost exists, but it was too small to be measured in our experiment; or (2) the cost is expressed in some fashion other than those measured in this experiment. Resolution of these possibilities requires additional study.
Predicted Responses to the Extract
The changes in juvenile morphology and adult life-history traits due to the presence of the Chaoborus extract that we observed match almost no previous predictions. A model of resource allocation in Daphnia in the presence of Chaoborus (Taylor and Gabriel 1992) predicts larger offspring, delayed maturity, and, under low food levels, a smaller size at maturity. In a quasi-natural selection experiment of Chaoborus predation on D. pulex (Spitze 1991), Chaoborus selected for individuals that were larger at birth, grew faster, matured earlier, and had greater fecundities. In our experiment, offspring were smaller in the presence of the Chaoborus extract, contradicting both the theoretical and experimental predictions. The theoretical and experimental studies were in conflict over whether later or earlier maturation is favored, while we found no difference in this trait. The only agreement between these previous predictions and our data was with the higher fecundities found in the selection experiment.
One reason that direct or plasticity costs were not found might be that changes in metabolic rate masked them. That is, costs would have existed for individuals at equal metabolic rates. Individuals in both treatments had approximately equal preadult growth rates as indicated by equal ratios between treatments of lengths at the second juveniles instar and maturity (Table 1). The extract treatment resulted in decreased adult growth rates, decreased size at maturity, and lower oxygen consumption rates. Individuals in the extract treatment also had higher fecundities (Table 1), especially those with larger body sizes [ILLUSTRATION FOR FIGURE 3 OMITTED]. Reduced metabolic rates without reduced reproductive rates could be achieved by decreasing activity or by increasing assimilation efficiency and changing resource allocation from maintenance to growth and reproduction.
Changes in growth efficiency can evolve; for example Neat et al. (1995) report increased growth efficiency as a consequence of laboratory natural selection at low temperature. Alternatively, decreasing activity could reduce metabolic rates. Daphnia change their behavior in response to Chaoborus extract (Dodson 1988). In particular, they decrease their movement rate (Pijanowska and Kowalczewski 1997). A decrease in movement by Daphnia in the presence of a predator might be adaptive because Chaoborus detect their prey using hydrodynamic cues. Reduced activity would also result in a decrease in energy expenditures permitting a change in the allocation of resources and life-history traits. In the presence of Chaoborus, the optimal strategy for Daphnia is to increase the allocation to reproduction (Taylor and Gabriel 1992), a prediction consistent with our results (Table 1).
These arguments illustrate how measurements of routine metabolic rate can help generate hypotheses about the mechanisms involved in plastic life-history responses to environmental cues. Our rapid assays of metabolic rate allowed us to characterize metabolic rate of a large number of clones. Determining the relative importance of changes in activity levels versus changes in assimilation efficiency will require the construction of detailed energy budgets for clones differing in the degree of plasticity (Lampert 1986; Glazier 1992). Of particular concern is the presence of embryonic tissue in the adults that we measured. Glazier (1991) estimates that the metabolic rates of eggs and early embryos are about one-third that of adult tissue in D. magna. Given that Chaoborus extract dramatically alters allocation to growth and reproduction, it will be important to control for the presence of embryonic tissue in more detailed studies of the mechanisms responsible for changes in allocation to growth and reproduction in the presence of predators by Daphnia.
The cost of being plastic is potentially an important factor in the evolutionary dynamics of phenotypically plastic traits. We found little evidence for either production or maintenance costs, as did the one other study that measured such costs on a morphological trait (DeWitt 1995, 1998). Nevertheless, the method presented here for quantifying the magnitude of plasticity costs and distinguishing between production and maintenance costs of plasticity represents a significant advance over past methods for studying costs of plasticity (DeWitt et al. 1998).
A priori, we expected that plasticity costs would be small. That expectation was met for the two experiments looking at plasticity of morphological traits in response to predation risk. In contrast, the study of increased protein production in response to temperature found significant costs (Krebs and Feder 1997). The critical distinction between these studies is that plastic responses to increased predation risk are much more complicated, involving behavior, morphology, and physiology. Being more complex traits, there are more opportunities for costs to be amortized over the entirety of the organism. We hypothesize that when plasticity costs are found, they will involve the plasticity of simple changes in gene expression or biochemical actions. Such systems are likely to have irreducible pleiotropic costs.
Measurement of the costs of plasticity where an entire life-history strategy is altered will require understanding interactions between suites of traits: behavior, morphology, and physiology. For example, the lack of measurable plasticity costs in the present study might have been due to differences in routine metabolic rates among treatments. Behavioral changes in filtration rates or activity patterns could be responsible for these differences. Thus, physiology and behavior could both be involved in plastic changes in life-history patterns.
Our study highlights the importance, when attempting to measure plasticity costs, of including both substantial genetic variation and a wide range of environmental conditions. These types of experiments present formidable logistic challenges, particularly if detailed energy budgets are required for multiple clones in multiple environments. However, the few theoretical models that have included costs found substantial effects on the outcome of evolution (Lively 1986; Van Tienderen 1991; Padilla and Adolph 1996). Thus measurements of the costs of plasticity are essential to understanding its evolution. Future efforts in this area should focus on two goals: (1) determine the empirical scope of plasticity costs across different types of traits, especially costs relative to fitness benefits; and (2) include costs in more types of evolutionary models.
We thank K. Spitze for supplying the PA clones, for several illuminating discussions on the evolutionary ecology of Daphnia, and especially for suggesting looking at tailspine length and body depth. L. Yampolsky was instrumental in setting up the lab for Daphnia work. We thank L. Baker, T. Kehoe, and C. Yanke for assistance in the lab, and R. Huey for the loan of the oxygen sensor. Comments on previous versions of the manuscript were provided by K. Hughes, L. Yampolsky, and several reviewers. The impetus to finally analyse this data, plus many stimulating ideas, came from the participation of SMS in a workshop on phenotypic plasticity at the University of Kentucky; we thank the organizers and other participants. In particular, T. DeWitt provided many insights into the issue of plasticity costs, access to unpublished manuscripts, and useful comments on earlier drafts of this manuscript. This project was supported by National Science Foundation Grants DEB-9221027 to SMS and DEB-9303164, INT-99424091 and USDA-96-35302-3739 to DB.
BLACK, A. R. 1993. Predator-induced phenotypic plasticity in Daphnia pulex: life history and morphological responses to Notonecta and Chaoborus. Limnol. Oceanogr. 38:986-996.
BLACK, A. R., AND S. I. DODSON. 1990. Demographic costs of Chaoborus-induced phenotypic plasticity in Daphnia pulex. Oecologia 83:117-122.
BRADSHAW, A. D. 1965. Evolutionary significance of phenotypic plasticity in plants. Adv. Genet. 13:115-155.
DEWITT, T. J. 1995. Functional trade-offs and phenotypic plasticity in the freshwater snail Physa. Ph.D. diss., State University of New York, Binghamton.
-----. 1998. The costs of phenotypic plasticity: tests with predator-induced morphology and life history in a freshwater snail. J. Evol. Biol. 11:465-481.
DEWITT, T. J., A. SIH, AND D. S. WILSON. 1998. Costs and limits to benefits as constraints on the evolution of phenotypic plasticity. Trends Ecol. Evol. 13:77-81.
DODSON, S. I. 1988. The ecological role of chemical stimuli for the zooplankton: predator-avoidance behavior in Daphnia. Limnol. Oceanogr. 33:1431-1439.
-----. 1989. The ecological role of chemical stimuli for the zooplankton: predator-induced morphology in Daphnia. Oecologia 78:361-367.
GLAZIER, D. S. 1991. Separating the respiration rates of embryos and brooding females of Daphnia magna: implications for the cost of brooding and the allometry of metabolic rate. Limnol. Oceanogr. 36:354-362.
-----. 1992. Effects of food, genotype, and maternal size and age on offspring investment in Daphnia magna Ecology 73: 910-926.
HAVEL, J. E., AND S. I. DODSON. 1984. Chaoborus predation on typical and spined morphs of Daphnia pulex: behavioral observations. Limnol. Oceanogr. 29:487-494.
-----. 1987. Reproductive costs in Chaoborus-induced polymorphism in Daphnia pulex. Hydrobiologia 150:273-281.
HEBERT, P. D. N., AND P. M. GREWE. 1985. Chaoborus-induced shifts in the morphology of Daphnia ambigua. Limnol. Oceanogr. 30:1291-1297.
HEBERT, P. D. N., R. D. WARD, AND L. J. WELDER. 1988. Clonal-diversity patterns and breeding-system variation in Daphnia pulex, an asexual-sexual complex. Evolution 42:147-159.
KETOLA, M., AND I. VUORINEN. 1989. Modification of life history parameters of Daphnia pulex Leydig and Daphnia magna Straus by the presence of Chaoborus sp. Hydrobiologia 179:149-155.
KREBS, R. A., AND M. E. FEDER. 1997. Natural variation in hsp70 expression, basal thermotolerance, and inducible thermotolerance in larval Drosophila melanogaster. Evolution 51:173-179.
KRUEGER, D. A., AND S. I. DODSON. 1981. Embryological induction and predation ecology in Daphnia pulex. Limnol. Oceanogr. 26: 219-223.
LAMPERT, W. 1986. Response of the respiratory rate of Daphnia magna to changing food conditions. Oecologia 70:495-501.
-----. 1993. Phenotypic plasticity of the size at first reproduction in Daphnia: the importance of maternal size. Ecology 74:1455-1466.
LANDE, R., AND S. J. ARNOLD. 1983. The measurement of selection on correlated characters. Evolution 37:1210-1226.
LIVELY, C. M. 1986. Canalization versus developmental conversion in a spatially variable environment. Am. Nat. 128:561-572.
LONING, J. 1992. Phenotypic plasticity of Daphnia pulex in the presence of invertebrate predators: morphological and life history responses. Oecologia 92:383-390.
LYNCH, M. 1989. The life history consequences of resource depression in Daphnia pulex. Ecology 70:246-256.
LYNCH, M., L. J. WEIDER, AND W. LAMPERT. 1986. Measurement of carbon balance in Daphnia. Limnol. Oceanogr. 31:17-33.
LYNCH, M., K. SPITZE, AND T. CREASE. 1989. The distribution of life-history variation in the Daphnia pulex complex. Evolution 43:1724-1736.
NEAT, F., K. FOWLER, V. FRENCH, AND L. PARTRIDGE. 1995. Thermal evolution of growth efficiency in Drosophila melanogaster. Philo. Trans. R. Soc. Lond. B Biol. Sci. 332:25-30.
PADILLA, D. K., AND S. C. ADOLPH. 1996. Plastic inducible morphologies are not always adaptive: the importance of time delays in a stochastic environment. Evol. Ecol. 10:105-117.
PAREJKO, K., AND S. I. DODSON. 1991. The evolutionary ecology of an antipredator reaction norm: Daphnia pulex and Chaoborus americanus. Evolution 45:1665-1674.
PHILLIPS, P. C., AND S. J. ARNOLD. 1989. Visualizing multivariate selection. Evolution 43:1209-1222.
PIJANOWSKA, J., AND A. KOWALCZEWSKI. 1997. Predators can induce swarming behaviour and locomotory responses in Daphnia. Freshwater Biol. 37:649-656.
RICHMAN, S. 1958. The transformation of energy by Daphnia pulex. Ecol. Monogr. 28:273-291.
RIESSEN, H. P., AND W. G. SPRULES. 1990. Demographic costs of antipredator defenses in Daphnia pulex. Ecology 71:1536-1546.
SAS INSTITUTE. 1989. SAS/STAT user's guide. Vers. 6. 4th ed., Vol. 2. SAS Institute Inc., Cary, NC.
SCHEINER, S. M., AND L. Y. YAMPOLSKY. In press. The evolution of Daphnia pulex in a temporally varying environment. Genet. Res.
SCHWARTZ, S. S. 1991. Predator-induced alterations in Daphnia morphology. J. Plankton Res. 13:1151-1161.
SPITZE, K. 1991. Chaoborus predation and life-history evolution in Daphnia pulex: temporal pattern of population diversity, fitness, and mean life history. Evolution 45:82-92.
-----. 1992. Predator-mediated plasticity of prey life history and morphology: Chaoborus americanus predation on Daphnia pulex. Am. Nat. 139:229-247.
SPITZE, K., AND T. D. SADLER. 1996. Evolution of a generalist genotype: multivariate analysis of the adaptiveness of phenotypic plasticity. Am. Nat. 148:S108-S123.
TAYLOR, B. E., AND W. GABRIEL. 1992. To grow or not to grow: optimal resource allocation in Daphnia. Am. Nat. 139:248-266.
TOLLRIAN, R. 1993. Neckteeth formation in Daphnia pulex as an example of continuous phenotypic plasticity: morphological effects of Chaoborus kairomone concentration and their quantification. J. Plankton Res. 15:1309-1318.
-----. 1995. Predator-induced defenses: costs, life history shifts, and maternal effects in Daphnia pulex. Ecology 76:1691-1705.
VAN TIENDEREN, P. H. 1991. Evolution of generalists and specialists in spatially heterogeneous environments. Evolution 45:1317-1331.
VUORINEN, I., M. KETOLA, AND M. WALLS. 1989. Defensive spine formation in Daphnia pulex Leydig and induction by Chaoborus crystallinus De Geer. Limnol. Oceanogr. 34:245-248.
WALLS, M., AND M. KETOLA. 1989. Effects of predator-induced spines on individual fitness in Daphnia pulex. Limnol. Oceanogr. 34:390-396.
WALLS, M., H. CASWELL, AND M. KETOLA. 1991. Demographic costs of Chaoborus-induced defenses in Daphnia pulex: a sensitivity analysis. Oecologia 87:43-50.
WALLS, M., C. LAUREN-MAATTA, M. KETOLA, P. OHRA-AHO, M. REINIKAINEN, AND S. REPKA. 1997. Phenotypic plasticity of Daphnia life history traits: the roles of predation, food level and toxic cyanobacteria. Freshwater Biol. 38:353-364.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||The Genetics of Phenotypic Plasticity, part 8|
|Author:||Scheiner, Samuel M.; Berrigan, David|
|Date:||Apr 1, 1998|
|Previous Article:||The relative importance of historical events and gene flow on the population structure of a Mediterranean ragwort, Senecio gallicus (Asteraceae).|
|Next Article:||Population structure of a clonal gorgonian coral: the interplay between clonal reproduction and disturbance.|