First grow, then breed and finally get fat: hierarchical. allocation to life-history traits in a lizard with invariant clutch size

Functional Ecology 2009, 23, 595 601 doi: 10.1111/j.1365-2435.2008.01518.x First grow, then breed and finally get fat: hierarchical Blackwell Publishing Ltd allocation to life-history traits in a lizard with invariant clutch size Lukás KubiCka and Lukás Kratochvíl* Faculty of Science, Department of Ecology, Charles University in Prague, Viniçná 7, 128 44, Praha 2, Czech Republic Summary 1. Organisms frequently encounter environments with different productivity. Fitness of an individual then depends on decisions made concerning energy allocation to particular life-history traits in a given environment. In reptiles and other ectotherms, individuals on a poor diet commonly reach smaller size and invest less in reproduction, but they often produce larger eggs than well-fed individuals. 2. Several lineages of reptiles including geckos have evolved an invariant clutch size as a derived mode of reproduction. Gecko females produce maximally two large eggs per clutch, but clutches are unusually frequent. Therefore, geckos serve as an interesting group for studying the generality of nutrition-dependent plasticity in life-history. 3. In a laboratory experiment, we manipulated diet in adult females of the Madagascar ground gecko Paroedura picta. Both food-limited and well-fed females followed the same growth trajectory in body and head length. In contrast, allocation to reproduction was highly nutrition-dependent. Although females in both treatment groups reproduced, food-limited females compromised both quantity and quality of their progeny: they laid clutches of smaller eggs in longer intervals. Fat storage was formed only in well-fed females. 4. We propose that the results are best explained by the consecutive hierarchical allocation of resources to growth, reproduction and storage, and discuss the consequences for investigation of life-history trade-offs. Key-words: canalization, egg size, nutrition, reproduction, geckos, phenotypic plasticity Introduction Life-history traits such as body size, egg number, egg size, and storage formation are interrelated characteristics tightly connected to individual fitness. The limited amount of energy an animal can obtain from its environment and convert into somatic or reproductive tissues inevitably leads to the existence of trade-offs between particular life-history traits (Roff 2002). In consequence, selection cannot optimize values of all the traits in question at the same time but only leads to optimized solutions of mutually interlinked trade-offs. However, organisms often encounter environments that are temporary and spatially diverse in productivity. Therefore, a solution of trade-offs optimal in certain conditions could be below the optimum levels under different circumstances. The fitness of an individual facing a given environment then depends on environment-matching decisions concerning energy *Corresponding author. E-mail: lukkrat@email.cz allocation to particular life-history traits. Alternative decisions, that is, individual nutrition-dependent strategies, were probably tested and optimized by natural selection during the evolutionary history of a lineage. Based on the encountered environments and on fitness resulting from past strategies, some life-history traits could be either canalized to values optimal within a wide range of environments, or phenotypically plastic (Stearns & Kawecki 1994). In respect of nutrition provisioning, most life-history traits in reptiles and other ectotherms have been proved to be plastic. Individuals on poor diet commonly start to reproduce at a smaller size, and reach smaller asymptotic body size (e.g. Madsen & Shine 1993; Atkinson & Sibley 1996, 1997). Experimental studies on ectotherms kept in laboratory conditions under different food regimes showed that food-limited females lay clutches less frequently and/or produce less offspring during a single reproductive bout than well-fed females (in squamate reptiles e.g. Seigel & Ford 1991; Lourdais et al. 2003; Du 2006; Warner et al. 2007). On the other hand, egg size has been 2008 The Authors. Journal compilation 2008 British Ecological Society

596 L. Kubiçka & L. Kratochvíl shown to be canalized to a single size or food-limited mothers have been even shown to produce larger eggs (e.g. Seigel & Ford 1991; Gregory & Skebo 1998; Du 2006; Warner et al. 2007 in squamate reptiles; Stelzer 2001 in a rotifer; Bashey 2006 in a fish). The counterintuitive production of larger offspring by mothers facing energetically harsh environment was viewed as an adaptive strategy, as larger offspring could have selective advantages in a highly competitive environment with scarce resources (Reznick et al. 1996; Bashey 2006). Nutrition-dependency of life-history traits in reptiles has been usually studied in species with relatively large but infrequent clutches, where clutch size is influenced not only by availability of resources but also by female size (e.g. Madsen & Shine 1992; Olsson & Shine 1997; Ji & Wang 2005). This strategy is typical for a wide taxonomical range of reptiles. Several reptile clades have independently evolved a derived mode of reproduction, whereby females lay relatively frequent clutches with low (one or two eggs) and rather constant number of eggs per clutch (Shine & Greer 1991). The evolution of such invariant clutch size (ICS) in the ancestors of two widely radiated groups of lizards probably reflected opposite selective forces: selection for reduction of total clutch mass in anoles, whereas selection for enlarged offspring size in geckos (Kratochvíl & Kubicka 2007). Taxa with ICS provide an opportunity to test the generality of nutrition-dependent plasticity/canalization in resource allocation in a new context. ICS provides a situation where the decision on clutch size is highly constrained. Since the clutch size is restricted to one or two eggs, females can adjust mainly the clutch frequency and the egg size. Also, the estimation of nutrition-dependent allocation to particular life-history traits in reptiles with the ancestral mode of reproduction is usually being complicated by the low frequency of clutches. In ICS lineages with frequent clutches, we can observe repeated individual decisions about resource investment in a relatively short period of time. This study aims to test the impact of different nutritional levels on resource allocation in a species with ICS. We manipulated diet in adult but not fully grown females of the Madagascar ground gecko, Paroedura picta (Peters, 1854), and recorded their growth, condition and reproduction parameters. Based on the literature data in other reptiles, we predicted that growth, body size and total reproductive investment would exhibit strong plasticity with respect to the amount of energy available, that is, that females on poor diet would have retarded growth, smaller asymptotic size and less frequent clutches. On the other hand, owing to the importance of egg size increase in the evolution of ICS in geckos (Kratochvíl & Kubicka 2007), we expected that egg size should not be negatively affected by limited food resources. Material and methods STUDIED ORGANISM The Madagascar ground gecko belongs among the largest species of the Malagasy-Comoran genus Paroedura (Dixon & Kroll 1974; Jackman et al. 2008). This terrestrial species inhabits dry forests and savannas of central and southern Madagascar (Dixon & Kroll 1974; Henkel et al. 2000). Although P. picta is very popular among hobbyist and is suitable as a laboratory experimental animal (Blumberg et al. 2002; Kratochvíl et al. 2006, 2008; this study), information on its ecology in the field is almost completely lacking. In nature, females were reported to lay eggs seasonally (September May) (Henkel et al. 2000), but they are able to breed continuously in captivity (personal observation). Females start reproduction well before reaching maximum body size. As a typical member of the family Gekkonidae (e.g. Kratochvíl & Frynta 2006), females P. picta lay one or two hardshelled eggs per clutch. EXPERIMENTAL SETUP For the manipulation of food intake, we chose 16 females of the Madagascar ground gecko P. picta from our outbred laboratory populations (altogether c. 200 individuals). Based on our experience from the preliminary experiment (see below), the expected differences between treatment groups were significant enough to be detectable even in a small number of animals. All experimental females were young adult virgins of the approximately same age and body size, which had not reached the final body size yet. Since hatching, they were kept in isolation at a constant temperature (26 ºC). During the experiment, they were housed individually in glass cages (30 30 20 cm) with dry sandpeaty substrate and a thermal gradient of 26 40 ºC provided by heating cables. A cardboard shelter covered the whole thermal gradient in every cage. Water and calcium powder in small dishes were provided ad libitum. Experimental females were randomly allocated to one of two feeding regimes. For seven consecutive months, food-intake of females in the first group (hereafter called well-fed ) was set to 0 5 g of crickets per day, whereas females in the second treatment group ( low-fed ) obtained only two-thirds of this amount. These feeding rates were calculated during the preliminary experiment, where we tested different food regimes over three months. We chose amounts of food that were consumable during a single night and allowed successful breeding in low-fed females. Live crickets of mass as close as possible to the set feeding dose (weighed with 0 01 g precision) were dusted with vitamins and minerals (Roboran H, Univit, Czech Republic) and provided to the animals twice a week. To avoid confounding effect of environmental gradient, the placement of the cages within the experimental room was counterbalanced with respect to the membership to treatment groups. The females were weighed and measured (snout-to-vent length, SVL, and head length as the closest distance between tip of snout and ear opening) every 2 weeks. Each female was allowed to mate with her assigned male every 2 weeks. Although viable sperm can be stored within female reproductive tract for at least 6 months (personal observation), we chose a shorter interval for re-mating to control the possible effect of sperm depletion or aging. We checked for the presence of laid eggs daily during the first 18 weeks of the experiment and twice a week before feeding thereafter. When only a single egg was laid and the second followed within 4 days, we considered these two eggs a single double-egg clutch. Four days represents a half of the shortest interval between two consecutive double-egg clutches observed in our P. picta breeding stock and could thus be considered a reliable rule for the assignment of single- vs. double-egg clutches. We recorded mass of a mother and her eggs at the day of their finding (the day of oviposition during the first 18 weeks). Eggs were placed individually into small plastic boxes and incubated at constant 30 C, a temperature well-suited for embryonic development in the Madagascar ground gecko (Blumberg et al. 2002). The incubator was monitored daily and hatchlings mass and SVL were recorded immediately on the day of hatching.

Hierarchical allocation to life-history traits 597 experimental groups: (i) interval between the first mating and the first clutch; (ii) laying period (time between the first and the last clutch during the experiment); (iii) total number of eggs; (iv) mean interclutch interval (defined as laying period divided by reproductive events minus one); (v) total number of clutches; (vi) ratio of singleto double-egg clutches, (vii) total reproductive effort (total egg-mass produced during the whole experiment). We tested the deviations from normality in the distribution of body dimensions and reproductive characteristics by Kolmogorov Smirnov test (all n.s.) and compared differences between experimental groups using t-tests. More conservative nonparametric tests do not change the significance of any of the results (not shown). The nested anova tests were used for testing the dependency of egg-mass and Olsson s (1994) condition index of hatchlings on the female-identity and the treatment group. The dependence of hatching success on egg size was tested by logistic regression. Pearson s correlation coefficient was used to check the dependency of hatchling mass on egg mass. Results Fig. 1. Growth rate of Paroedura picta. (a) Well- and low-fed females followed the same growth trajectory in body length. Solid circles represent low-fed females. (b) Since approximately the 70 day of the experiment, well-fed females (open circles) exhibit better body condition than low-fed females. In low-fed females, body condition index decreased, especially in the second half of the experiment. STATISTICAL ANALYSES The growth pattern of both treatment groups is evidently nonlinear (Fig. 1a). Therefore, to compare individual growth rates between groups we applied the asymptotic growth curve, specifically the logistic-bylength model (Schoener & Schoener 1978; Powell & Russell 1985) SVL = (a 3 /(1 + be r t ) 1/3, where e is the base of the natural logarithm, t is age (in days), a is the estimated asymptotic SVL (in mm), r is the characteristic growth rate, and b is the function of the length at time 0. We estimated three parameters of the curve using Levenberg-Marquart algorithm implemented in nonlinear regression module of Statistica (StatSoft 2001). To evaluate the temporal changes in female condition during the experiment, we calculated Olsson s (1994) condition index, that is, the ratio of the cube root of mass to SVL (mass l/3 /SVL). This procedure normalizes the discrepancy between the increase in the three-dimensional parameter (body mass) and in the linear, one-dimensional parameter (SVL), and it is appropriate to compare condition between groups differing in body size (in contrast to the more often used residuals from common regression). Expanding on the methods of Kearney & Shine (2004, 2005), we compared the following reproductive characteristics between the On average, each low-fed female was provided with 70 6 ± 0 2 g, whereas each well-fed female was provided with 105 5 ± 0 1 g of live crickets during the experiment, which is close to the amount set before the start of the experiment (70 and 105 g, respectively). After the termination of the experiment, we decided to exclude the data obtained from two females (one from each treatment group) from the results and all analyses. The first one, a member of the low-fed group, did not reproduce at all during the whole experiment (but she started to breed spontaneously several months after its termination). The second excluded female, this time from the well-fed group, reproduced very poorly with an enormous number of single-egg clutches (15 out of 16 clutches), and her offspring exhibited a high number of malformations (crinkly tail, disproportional head shape). Soon after the beginning of the experiment, her abdomen swelled and she died shortly after the termination of the experiment. However, the exclusion of this female did not effect the significance of the results and their interpretations (with the exception of the marginally non-significant result concerning total number of eggs laid, see below). Throughout the remainder of the text, we report results obtained only in 14 experimental animals (seven in each treatment group). FEMALE SOMATIC CHARACTERISTICS At the beginning of the experiment, all females were of approximately the same age (329 ± 8 days, mean ± SE are provided throughout the text) and body dimensions (SVL, 74 84 ± 0 44 mm; head length, 22 20 ± 0 11 mm; mass, 12 187 ± 0 237 g; condition index, 0 0307 ± 0 0002 g 1/3 mm 1 ). The low- vs. well-fed females did not differ in any mentioned characteristic at the beginning of the experiment (one-way anova; all P > 0 11). Females grew approximately 4 89 ± 1 37 mm during the experiment. Surprisingly, final structural body dimensions (head length, SVL) did not differ between treatment groups

598 L. Kubiçka & L. Kratochvíl Table 1. Summary of the parameters monitored during the experiment. The results for both treatment groups are given as means ± SE. Statistically significant differences among the experimental groups are highlighted with asterisks Treatment group (mean ± SE) Parameter Low-fed females Well-fed females F P Growth rate 0 0564 ± 0 0101 0 0407 ± 0 0065 1 70 0 216 Final SVL (mm) 79 46 ± 0 85 79 99 ± 0 96 0 17 0 690 Final head length (mm) 23 49 ± 0 20 23 94 ± 0 12 3 63 0 081 Final mass (g) 13 065 ± 0 630 16 258 ± 0 397 18 38 0 001* Final condition index (g 1/3 mm 1 ) 0 0296 ± 0 0005 0 0317 ± 0 0005 7 57 0 018* Interval between first mating and first clutch (days) 19 0 ± 4 0 17 0 ± 1 9 0 21 0 658 Laying period (days) 188 0 ± 4 0 191 9 ± 1 5 0 83 0 379 Total number of eggs 37 3 ± 1 9 46 1 ± 1 9 10 76 0 007* Interclutch interval (days) 9 2 ± 0 3 7 6 ± 0 3 15 35 0 002* Total number of clutches 19 6 ± 0 9 24 1 ± 0 8 14 03 0 003* Percentage of single-egg clutches (%) 9 63 ± 3 12 8 96 ± 3 86 0 02 0 894 Egg mass (g) 0 856 ± 0 004 0 955 ±0 004 610 33 <<0 001* Total reproductive effort (g) 31 808 ± 1 629 43 788 ± 2 234 18 77 <0 001* Condition index of hatchlings (g 1/3 mm 1 ) 0 0296 ± 0 0001 0 0304 ± 0 0001 53 92 <<0 001* (Table 1). Individual logistic-by-length model of asymptotic equation explained on average 92 5% (range 81 4 97 1) of the variability in growth. It is evident that females in both groups followed the same growth trajectory (Fig. 1a), and the members of both groups did not differ in the grow rates estimated by fitting of the asymptotic growth curves (Table 1). On the other hand, the experimental groups strongly differed in body mass and correspondingly also in the condition index at the end of the experiment (Table 1; Fig. 1b). During the experiment, the low-fed females progressively reduced their condition index, while the well-fed females increased in body condition (Fig. 1b). FEMALE FECUNDITY AND EGG CHARACTERISTICS Females in both treatment groups started to lay eggs at the same time. There was no difference between treatment groups in the interval between the first mating and the first clutch (Table 1). Neither group differed in the duration of the total laying period (Table 1). Thus, we can conclude that females in both groups laid eggs for the whole 7 months of the experiment. Altogether, females laid 584 eggs during the experiment. The low-fed females laid significantly less eggs than the wellfed females (Table 1; P = 0 054 when the outlier female from well-fed group is not excluded), which is also reflected in the differences in the mean interclutch interval and the total number of clutches (Table 1). Nonetheless, there was no significant difference in the ratio of single- to double-egg clutches between groups (Table 1). Only two females, one from each treatment group, always laid double-egg clutches. Individual females differed in egg size even within treatment groups (nested anova; female identity: F = 37 20, P << 0 001). However, differences between treatment groups were more pronounced and explained a much higher portion of the total variability in egg size (Table 1). Egg mass in low-fed females was on average more than 10% lower than in well-fed females. Mean mass of eggs from the double-egg clutches was smaller than in eggs from the single-egg clutches in five out of 12 females with both single- and double-egg clutches, the opposite was true in the other seven females. Females from both treatments differed in the total reproductive effort, which was about 27% lower in the low-fed females (Table 1). HATCHING SUCCESS AND OFFSPRING CHARACTERISTICS Out of 584 eggs, 17 were broken during manipulation, these eggs were omitted in the subsequent analyses. 486 juveniles successfully hatched from the remaining 567 eggs. Hatching success was inversely dependent on the original egg mass (logistic regression: Wald statistics = 9 31; P = 0 002). The egg mass and the hatchling mass were positively correlated (r = 0 81, P << 0 001; Fig. 2). In some cases, the juveniles did not absorb the yolk-sack before hatching. Such individuals occurred in both low-fed and well-fed mothers and formed evident outliers from egg mass-hatchling mass relationship (cf. Fig. 2). Hatchlings condition index varied among individual mothers (nested anova; female identity: F = 4 49, P << 0 001), but similarly to the situation in egg size, the differences between treatment groups were more pronounced and explained a much higher portion of the total variability (Table 1). Discussion Manipulation of nutrition in reproducing gecko females had obvious effects on several life-history traits. The experiment demonstrated strong plasticity of maternal and juvenile body condition and maternal reproductive characteristics. On the other hand, contrary to our expectations, females on poor diet did not exhibit retarded growth and did not reach smaller

Hierarchical allocation to life-history traits 599 Fig. 2. Hatchlings mass increased with egg size. Well-fed mothers produced larger eggs (marked as open circles) than low-fed mothers. A group of outliers with considerably smaller mass of juvenile with respect to egg mass represent offspring hatched without internalized yolk sac. Fig. 3. Schematic depicting of the comparison of resource investment between well- and low-fed females. Both treatment groups allocated equally to growth, low-fed females then allocated less to reproduction. Storage was formed only in well-fed mothers. asymptotic body size. Structural body size (here represented by SVL and head length) in P. picta thus seems to be canalized with respect to nutritional status, at least within nutrition levels provided in our experiment. As characteristic for reptiles (Shine & Charnov 1992), Madagascar ground gecko females start to reproduce before attaining their maximal body size. Although reptiles are usually viewed as a group of animals with indetermined growth (but see e.g. Congdon et al. 2001), the data obtained in our laboratory showed that the growth of P. picta of both sexes is minimal or non-existent at higher ages (H. JirkU & L. Kratochvíl, unpublished data). We observed considerable increase in body mass only in the well-fed females. The low-fed females did not change mass during the experiment, which means that they progressively decreased their body condition (Fig. 1b) as they depleted the fat storage. At the end of the experiment, the low-fed females were slender, especially in comparison with the well-fed females with their visible abdominal fat bodies and wide tails (Fig. 3). Allocation to reproduction was highly nutrition-dependent. Although females in both treatment groups reproduced for the whole duration of the experiment, the food-limited females laid clutches over longer intervals, producing fewer eggs and smaller eggs compared with the well-fed females. Smaller eggs had lower hatching success and the hatchling size strongly reflected the egg size. Although fitness is difficult to estimate in captivity, it appears that the food-limited females compromised both quantity and quality of their progeny. The nutrition-dependent strategy in P. picta is different from that described in vertebrates with variable clutch size, where food-limited females decrease offspring number but do not reduce egg size (e.g. Bashey 2006; Du 2006; Warner et al. 2007). Female geckos (the whole clade Gekkota) are generally constrained to lay just one or two eggs per clutch. Obligatory single-egg clutches occur mostly in small species, but females of larger species occasionally lay clutches of a single egg as well (Doughty 1996; Doughty & Thompson 1998). Colli et al. (2003) reported single-egg clutches in small females, while double-egg clutches in large females of the gecko Gymnodactylus geckoides amarali. As reptile females on poor diet generally lay smaller clutches (e.g. Seigel & Ford 1991; Olsson & Shine

600 L. Kubiçka & L. Kratochvíl 1997), we expected that food-limited females of P. picta would lay a larger proportion of single-egg clutches. Additionally, according to the small clutch size model (e.g. Charnov et al. 1995; Guinnee et al. 2004), females in organisms with small clutches should not be able to optimally divide resources among their progeny as the total amount of resources devoted to a clutch must be divided by a small integer number (number of eggs in the clutch). As eggs of small, suboptimal size are expected to exhibit low fitness, mothers are predicted to invest more resources to individual eggs from smaller clutches in comparison to those from larger clutches. Female P. picta should then lay larger eggs in single-egg clutches in comparison to double-egg clutches. Our data support neither of these predictions. The low-fed females laid similar proportion of single-egg clutches as the well-fed females, and single laid eggs were of comparable size to those laid in doubles. Single-egg clutches in P. picta thus seem to be a result of a reproductive failure leading to unilateral follicular ablation rather than adaptive manipulation with clutch size or offspring size. Although we used only two levels of resources, taken together, we suggest that the decisions on resource allocation in gecko females probably follow a simple hierarchical allocation rule schematically depicted in Fig. 3. All females allocated the same amounts of energy to structural size. Such canalization of a life-history trait could be caused by stabilizing natural selection towards its optimal value and it suggests considerable importance of this trait for individual fitness (Stearns & Kawecki 1994). Following the fulfilment of growth energetic requirements, the gecko females allocated energy to reproduction. The low-fed females were not able to reach values observed in the well-fed females in parameters measuring quantity and quality of their juveniles and they did not accumulate energy in the form of fat storage. Formation of fat storage thus has the lowest priority in resource allocation and it is only allowed after reaching maximal reproductive effort in the well-fed females. It would be interesting to explore the mechanisms preventing the well-fed females from producing even larger eggs in even shorter intervals. The egg size reached in well-fed females could be the result of a limited gap of pelvic opening (concerns egg width), and the limit for egg length may be a form of penalty relating to the amount of calcium needed for eggshell in elongated eggs (Kratochvíl & Frynta 2006). Alternatively, the well-fed females could form eggs of optimal size, whose further increase would not bring any fitness advantage and it is thus better to store the energy in the form of fat storage for the next clutch. Reznick et al. (1996) compared the nutrition-dependent reproductive strategies in females of three viviparous species of the poeciliid fishes. Two lecithotrophic species responded to low-food environment by producing larger young with greater fat reserves, while the only examined matrotrophic species (Heterandria formosa) responded to low-food environment by the production of smaller juveniles. The authors proposed that matrotrophy might represent a constraint preventing an adaptive increase of offspring size observed in other ectotherms in response to a nutritionally harsh environment. We observed the same pattern in P. picta, which is neither viviparous nor matrotrophic. Both P. picta and H. formosa share conspicuously shortened interclutch/birth intervals in comparison with their relatives (Reznick et al. 1996; this study). We therefore suggest that it is not matrotrophy, but selection for production of offspring in high frequency that can prevent these species from producing offspring of optimal or enlarged size at low-food levels. Geckos are iteroparous, relatively long-lived animals. Paroedura picta can live in captivity for more than 6 years (personal observation). Therefore, for females, it is probably advantageous to invest in somatic growth even in harsh environments, and then to reproduce as frequently as possible and to modulate investment into each clutch in response to the immediate condition status/nutritional environment of the mother. Females store surplus energy only after the maximal reproductive effort is reached. When facing nutritional limitation, they increase interclutch intervals, but at the same time they produce suboptimal eggs rather than wait for the accumulation of resources needed for vitellogenesis of optimally sized eggs. The Madagascar ground gecko thus seems to adopt income breeding strategy (Stearns 1992). It strongly contrasts with the strategy of capital breeders typical for other squamate reptiles such as some viperid snakes, that do not start reproduction until the resources accumulated in the form of fat storage and later used during vitellogenesis get over a certain threshold (Naulleau & Bonnet 1996; see Lourdais et al. 2003 for evidence that neither these snakes are pure capital breeders). In summary, our study demonstrated that lineages with ICS can have different nutrition-dependent reproductive strategies to other reptiles and that the reproductive strategies of reptiles are more diverse than expected. We encourage further research of the nutrition-dependent reproductive strategies in reptiles, especially in the earlier neglected lineages with highly derived modes of reproduction such as ICS. The existence of hierarchical allocation suggested by our results could have a significant impact on the detection of trade-offs between particular life-history traits (Worley et al. 2003). The existence of trade-offs between two life-history traits is often tested by searching for the negative correlation between measures of these traits. However, a positive correlation could be found between traits among individuals in a single population differing in the resource acquisition (van Noordwijk & de Jong 1986; Reznick et al. 2000; Roff 2002) even where a trade-off between these two traits actually exists. In the case of hierarchical allocation, we can expect lack of correlation between traits at different hierarchical levels sharing common pool of resources. Although trade-offs among all three traits (structural growth, reproduction and formation of storage) studied in our experiment are likely to be unavoidable, the hierarchical allocation rules resulted in the same amount of energy invested in growth in both the experimental groups, and higher investment in both reproduction and storage in the well-fed group. We thus cannot for example detect a negative correlation between structural growth and reproduction, as the investment in growth is fixed at our experimental food

Hierarchical allocation to life-history traits 601 levels. The recent paper by Knops et al. (2007) documented that a negative correlation between two life-history traits (growth and reproduction) does not automatically imply a trade-off between them. The hierarchical allocations can further clarify why even the implication in the opposite direction cannot always be true; that is, why the existence of a trade-off does not always imply a negative correlation between traits. Acknowledgements The authors thank Alison Golinski, David Horák, Zuzana Starostová, Paul Doughty and an anonymous referee for their comments on the earlier version of the manuscript. The research was supported by the MSMT (project No. 0021620828) and the Grant Agency of the Czech Academy of Sciences (project No. KJB601110706). The experiment was held under approval and supervision of the Ethical Committee of our institution. References Atkinson, D. & Sibly, R.M. (1996) On the solution to a major life history puzzle. Oikos, 77, 359 364. Atkinson, D. & Sibly, R.M. (1997) Why are organisms usually bigger in colder environments? Making sense of a life history puzzle. Trends in Ecology and Evolution, 12, 235 239. Bashey, F. (2006) Cross-generational environmental effects and the evolution of offspring size in the Trinidadian guppy Poecilia reticulata. Evolution, 60, 348 361. Blumberg, M.S., Lewis, S.J. & Sokoloff, G. (2002) Incubation temperature modulates post-hatching thermoregulatory behavior in the Madagascar ground gecko, Paroedura pictus. Journal of Experimental Biology, 205, 2777 2784. Charnov, E.L., Downhower, J.F. & Brown, L.P. (1995) Optimal offspring sizes in small litters. Evolutionary Ecology, 9, 57 63. Colli, G.R., Mesquita, D.O., Rodrigues, P.V.V. & Kitayama, K. (2003) Ecology of the gecko Gymnodactylus geckoides amarali in a neotropical savanna. Journal of Herpetology, 37, 694 706. Congdon, J.D., Nagle, R.D., Kinney, O.M. & van Loben Sels, R.C. (2001) Hypotheses of aging in a long-lived vertebrate, landing s turtle (Emydoidea blandingii). Experimental Gerontology, 36, 813 827. Dixon, J.R. & Kroll, J.C. (1974) Resurrection of the generic name Paroedura for the Phyllodactyline geckos of Madagascar, and description of a new species. Copeia, 1974, 24 30. Doughty, P. (1996) Allometry of reproduction in two species of gekkonid lizards (Gehyra): effects of body size miniaturization on clutch and egg sizes. Journal of Zoology, 240, 703 715. Doughty, P. & Thompson, M.B. (1998) Unusual reproductive patterns in the Australian marbled gecko (Phyllodactylus marmoratus). Copeia, 1998, 747 752. Du, W.G. (2006) Phenotypic plasticity in reproductive traits induced by food availability in a lacertid lizard, Takydromus septentrionalis. Oikos, 112, 363 369. Gregory, P.T. & Skebo, K.M. (1998) Trade-offs between reproductive traits and the influence of food intake during pregnancy in the garter snake, Thamnophis elegans. The American Naturalist, 151, 477 486. Guinnee, M.A., West, S.A. & Little, T.J. (2004) Testing small clutch size models with Daphnia. The American Naturalist, 163, 880 887. Henkel, F.W., Schmidt, W., Knothig, M., Liebel, K., Zobel, R. & Kaiser, H. (2000) Amphibians and Reptiles of Madagascar, the Mascarene, the Seychelles, and the Comoro Islands. Krieger Publishing Company, Malabar, FL. Jackman, T.R., Bauer, A.M., Greenbaum, E., Glaw, F. & Vences, M. (2008) Molecular phylogenetic relationships among species of the Malagasy- Comoran gecko genus Paroedura (Squamata: Gekkonidae). Molecular Phylogenetics and Evolution, 46, 74 81. Ji, X. & Wang, Z.W. (2005) Geographic variation in reproductive traits and trade-offs between size and number of eggs of the Chinese cobra (Naja atra). Biological Journal of the Linnean Society, 85, 27 40. Kearney, M. & Shine, R. (2005) Lower fecundity in parthenogenetic geckos than sexual relatives in the Australian arid zone. Journal of Evolutionary Biology, 18, 609 618. Kearny, M. & Shine, R. (2004) Developmental success, stability, and plasticity in closely related parthenogenetic and sexual lizards (Heteronotia, Gekkonidae). Evolution, 58, 1560 1572. Knops, J.M.H., Koenig, W.D. & Carmen, W.J. (2007) Negative correlation does not imply a tradeoff between growth and reproduction in California oaks. Proceedings of the National Academy of Sciences,USA, 104, 16982 16985. Kratochvíl, L. & Frynta, D. (2006) Egg shape and size allometry in geckos (Squamata: Gekkota), lizards with contrasting eggshell structure: Why to lay spherical eggs? Journal of Zoological Systematics and Evolutionary Research, 44, 217 222. Kratochvíl, L. & Kubicka, L. (2007) Why reduce clutch size to one or two eggs? Reproductive allometries reveal different evolutionary causes of invariant clutch size in lizards. Functional Ecology, 21, 171 177. Kratochvíl, L., Kubicka, L. & Landová, E. (2006) Yolk hormone levels in the synchronously developing eggs of Paroedura picta, a gecko with genetic sex determination. Canadian Journal of Zoology, 84, 1683 1687. Kratochvíl, L., Kubicka, L. & Landová, E. (2008) Does mechanism of sex determination constrain the potential for sex manipulation? A test in geckos with contrasting sex-determining systems. Naturwissenschaften, 95, 209 215. Lourdais, O., Bonnet, X., Shine, R. & Taylor, E. (2003) When does a reproducing female viper (Vipera aspis) decide on her litter size? Journal of Zoology, 259, 123 129. Madsen, T. & Shine, R. (1992) Determinants of reproductive success in female adders, Vipera berus. Oecologia, 92, 40 47. Madsen, T. & Shine, R. (1993) Phenotypic plasticity in body sizes and sexual size dimorphism in European grass snakes. Evolution, 47, 321 325. Naulleau, G. & Bonnet, X. (1996) Body condition threshold for breeding in a viviparous snake. Oecologia, 107, 301 306. Olsson, M. (1994) Nuptial colouration in the sand lizard, Lacerta agilis: an intra-sexually selected cue to fighting ability. Animal Behaviour, 48, 607 613. Olsson, M. & Shine, R. (1997) The limits to reproductive output: offspring size versus number in the sand lizard (Lacerta agilis). The American Naturalist, 149, 179 188. Powell, G.L. & Russell, A.P. (1985) Growth and sexual size dimorphism in Alberta populations of the eastern short-horned lizard, Phrynosoma douglassi brevirostre. Canadian Journal of Zoology, 63, 139 154. Reznick, D., Callahan, H. & Llauredo, R. (1996) Maternal effects on offspring quality in poeciliid fishes. American Zoologist, 36, 147 156. Reznick, D., Nunney, L. & Tessier, A. (2000) Big houses, big cars, superfleas and the costs of reproduction. Trends in Ecology and Evolution, 15, 421 425. Roff, D.A. (2002). Life history evolution. Sinauer Associates Inc., Sunderland. Schoener, T.W. & Schoener, A. (1978) Estimating and interpreting body-size growth in some Anolis lizards. Copeia, 1978, 390 405. Seigel, R.A. & Ford, N.B. (1991) Phenotypic plasticity in the reproductive characteristics of an oviparous snake, Elaphe guttata implications for life-history studies. Herpetologica, 47, 301 307. Shine, R. & Greer, A.E. (1991) Why are clutch sizes more variable in some species than in others? Evolution, 45, 1696 1706. Shine, R. & Charnov, E.L. (1992) Patterns of survival, growth, and maturation in snakes and lizards. The American Naturalist, 139, 1257 1269. StatSoft, Inc. (2001) STATISTICA for Windows, Vers. 6 0. StatSoft, Tulsa, Oklahoma. Stearns, S.C. (1992) The evolution of life histories. Oxford University Press, Oxford. Stearns, C.S. & Kawecki, T.J. (1994) Fitness sensitivity and the canalization of life-history traits. Evolution, 48, 1438 1450. Stelzer, C.-P. (2001) Resource limitation and reproductive effort in a planktonic rotifer. Ecology, 82, 2521 2533. van Noordwijk, A.J. & de Jong, G. (1986) Acquisition and allocation of resources: their influence on variation in life history tactics. The American Naturalist, 128, 137 142. Warner, D.A., Lovern, M.B. & Shine, R. (2007) Maternal nutrition affects reproductive output and sex allocation in a lizard with environmental sex determination. Proceedings of Royal Society B, 274, 883 890. Worley, A.C., Houle, D. & Barrett, S.C.H. (2003) Consequences of hierarchical allocation for the evolution of life-history traits. The American Naturalist, 161, 153 167. Received 9 July 2008; accepted 6 November 2008 Handling Editor: Frank Messina