Introduction

Life history traits are considered the clearest, indirect measures of an organism’s fitness in nature (Stearns 1992); traits such as survival, growth rate, fecundity, age at maturity, progeny size, etc., exceed the individual level, greatly affecting the evolutionary ecology of organisms. The expression of these phenotypic traits, like any others, is mediated through environmental influences; nevertheless, life history traits have a particularly strong influence on an organism’s fitness (Roff 2002). One important source of environmental variation is the maternal effect, which is defined as nongenetic influences of the maternal phenotype (or maternal environment) on progeny phenotype, independent of progeny genotype (Mousseau and Dingle 1991). Several studies suggest that, although maternal effects often have large effects on progeny phenotype early in ontogeny, they are often undetectable later in ontogeny (Mousseau and Dingle 1991; Fox and Mousseau 1998; Mousseau and Fox 1998; Fox 2000). Nevertheless, some traits, such as growth rate, can affect young individuals and also have profound consequences on their later life cycle.

Many arthropods, including crustaceans, exhibit indeterminate growth; in some terrestrial isopod species larger females can weigh nearly 10 times the live weight of the smallest sexually mature females (Lawlor 1976; Warburg 1987). Even though the young from different females differ in weight and/or size by only a few micrograms/millimeters, since juvenile growth is exponential, the smallest offspring grow slower, and can be expected to reach a smaller size at maturity than the larger young (Warburg 1987). That is, the absolute difference in offspring size is amplified by their exponential growth.

Life history traits are often negatively associated with each other (Stearns 1992; Rose et al. 1996; Roff 2002). Theoretical studies of trade-offs clearly show that the sign of a correlation between two traits cannot be used as an unambiguous indicator of whether those traits functionally interact in a negative manner (Zera and Harshman 2001). Trade-offs are often difficult to observe within natural populations (Stearns 1989). One reason for this problem is that variation in resource acquisition among individuals can confound the detection of an expected trade-off (Tuomi et al. 1983; Jordan and Snell 2002). Quantifying nutrient input has long been regarded as an essential aspect of energy budget studies (Congdon et al. 1982; Withers 1992; Niewiarowski 2001). For many organisms, however, energy may be superabundant relative to the supply of various nutrients, and the acquisition and expenditure of certain nutrients may be far more critical than the allocation of energy (Slansky and Feeny 1977; Coll and Guershon 2002). Traditional studies of life history trade-offs have focused almost exclusively on variation in energy input (Calow 1979; Townsend and Calow 1981; Zera and Harshmann 2001); however, in these studies it is not clear which internal nutrients are limiting the physiological and reproductive functions of the organisms (Zera and Larsen 2001).

Food quality greatly influences survival, development and reproduction in many arthropod herbivores (Karowe and Martin 1989; Rossi and Strong 1991; Yang and Joern 1994). Different species of plants have been shown to differ greatly in their suitability and availability as a herbivore food source, as have different populations of plants from the same species (Stockhoff 1993). This variation depends on geographic or seasonal factors, and may also depend on the stage of plant development or its physiological state, which may also fluctuate in space and time (Schultz et al. 1982; Stockhoff 1993). Terrestrial isopods feed primarily on leaf litter, detritus, live vegetation, and are sometimes predators of microarthropods (David et al. 2001), but it should be emphasized that the main food source for these organisms is detritus, originating from flora (Warburg 1987).

Since their main food source is composed of vegetable detritus, the differential availability and composition of the vegetation affects the biology of these isopods (Rushton and Hassall 1983a, 1983b; Meriam 1971; Zimmer and Topp 1997; Lavy et al. 2001). In this study we empirically quantify the effects of diet quality on the life history traits of females, and the trade-offs in these traits, evaluated as effects on their progeny. We hypothesized that the effects of diet on the life history of females can greatly affect offspring performance (considered as growth rate). Using the terrestrial isopod Porcellio laevis as a model we tested this prediction. Our study had the following three objectives: (1) to evaluate the effect of diet quality on life history traits; (2) to determine which essential nutrient cause, or change, the trade-offs under study; and (3) to quantify maternal effect on offspring growth rate.

Materials and methods

Animals and study site

P. laevis is widely distributed in Chile, and in several other zones throughout the world (Leistikow and Wägele 1999). The individuals utilized in the experiment were obtained from EDIEM, the university terrestrial station (Estación de Investigaciones Ecológicas Mediterráneas) in San Carlos de Apoquindo (33° 23’S, 70° 31’W). The station is located 20 km to the east of Santiago, in the Andean foothills, and encompasses roughly 835 ha. The climate in the study area is Mediterranean, with annual mean rainfall of 376.4 mm, concentrated (65%) during the austral winter months, from June to August (Jaksic 2001). The annual mean temperature averages 15.9 °C, with the mean minimum temperature 6°C below, and the maximum mean temperature 7°C above this annual mean. The study area at San Carlos de Apoquindo is covered by sclerophyllous vegetation which, physiognomically, may be described as evergreen scrub (for a complete description of the study site see Jaksic 2001). While there is no general phenological description of the shrub vegetation at San Carlos de Apoquindo, there is a fairly detailed one regarding the flowering and fruiting period of the most common woody shrubs at the site (Jaksic 2001).

We collected all isopods by hand, from under stones, pieces of wood, and soil litter. We placed all specimens in plastic containers for transfer to the laboratory in the Department of Ecology of Pontificia Universidad Católica de Chile.

Maintenance and culture

We sexed all individuals and later assigned each to one of three different diet treatments (see Table 1). We determined the gross energy of each diet using a Parr 1261 computerized calorimeter, with three replicates per treatment. All diets were isocalorific (i.e. differences in calories between treatments were smaller than 3%), with average values of 18.8±0.12, 18.1±0.07, and 18.9±0.14 kJ/g for control (C), high-carbohydrate (HC) and high protein (HP) diets, respectively. In each petri dish (50 mm diameter; base layer of plaster of Paris) we placed three females and one male. For each experimental condition we utilized 30 replicate petri dishes (i.e. 120 individuals: 90 females and 30 males). We maintained all treatments in an environmental chamber at 21 °C and LD14:10. The adult cultures were maintained for 10 months; during the course of the experiment we weighed all individuals weekly (CHYO JK-180 with precision ±0.01 mg), and provided food ad libitum. We placed females and males together to allow fertilization and egg extrusion in the females.

Table 1 Composition of experimentally purified high-carbohydrate, high-protein and control diets. Weights are expressed in grams on an air-dry basis
Full size table

We maintained the resulting juveniles from each female in laboratory cultures. For simplicity, in this text we use the term “juvenile” to refer to the larval manca stages (characterized by having six thoracic segments) both within the maternal brood pouch and following release from the female. We kept juveniles in clear plastic plates (10×10×2 cm; base layer of plaster of Paris), which were subdivided into 25 cells, with one juvenile in each cell. The juvenile cultures were also maintained for 15 weeks; during the course of the experiment we weighed the juveniles weekly, and provided food from the control diet ad libitum.

We measured the following life history traits: egg incubation period (IP), offspring number per female (ON), offspring size at release (OS), and growth rate of juveniles and adults.

The IP was considered as the total number of days from egg extrusion to the moment of juvenile release (this is denoted by a change in coloration in the anterior ventral zone, from dark orange to yellow). At this moment we separated females, placing them in individual incubation chambers until the day of offspring release, for which we assessed the females daily.

Immediately following emergence from the maternal marsupium we fixed offspring in 50% alcohol. We measured OS as the total length of offspring (i.e. the distance between the mid-dorsal anterior margin of the carapace and the distal margin of the pleotelson) in a subsample of ten offspring from each female. We measured females and their offspring using a compound microscope equipped with a calibrated ocular micrometer. We quantified the ON by counting all the young of each female under a compound microscope.

To calculate individual growth rates of both adults and juveniles we utilized the differences in individual masses at successive time periods. Through weekly weighing of the individuals in an analytical balance (precision of ±0.01 mg) we obtained the growth rates of both adults (GRA) and juveniles (GRJ). In the case of GRA we considered only measurements conducted on males, because females showed strong variation in weight depending on their reproductive state. We assessed adult growth in terms of absolute growth rate, which we calculated as the difference in weight at the beginning and end of the experiment, divided by time (in days). For GRJ, we adopted the modification of the von Bertalanffy growth model realized by Folker-Hansen et al. (1996), for fitting the weight data of individuals for each diet treatment:

$$ W{\left( t \right)} = W_{\infty } {\left( {1 - e^{{{\left( { - k{\left( {t - t_{0} } \right)}} \right)}}} } \right)}; $$

where W(t) is weight as a function of time; W ∞ is asymptotic weight; k is the growth rate coefficient; and t 0 is the hypothetical negative time for an individual with weight 0, estimated from the hatching date. In our analysis we considered the individual k coefficients for each diet.

Statistical analysis

To test for differences in the traits IP, GRA, and GRJ between diet treatments we used a one-way ANOVA. To evaluate differences in the OS and ON between treatments, we performed an ANCOVA, with female body weight as the covariate (Sokal and Rohlf 1997). Prior to analysis, all data were tested for normality and homoscedasticity (Kolgomorov-Smirnov and Cochran C tests, respectively). When differences in the means were significant at the P <0.05 level, they were also tested with an a posteriori Tukey test (HSD). Product-moment (Pearson) correlations between different life history traits were computed separately for each diet treatment. When female body mass was related to a life history variable, we utilized the residuals from these relationships. All statistical analyses were conducted using STATISTICA 6.0 software (Statsoft, Tulsa, Okla.).

Results

Offspring size and number

In the laboratory, OS and ON of female P. laevis developed in markedly different ways with respect to food quality. Average ON varied from 46 for females on the HP diet (range: 10–80), to 41 for females on the C diet (range: 15–94), to 40 for females on the HC diet (range: 10–73; Fig. 1). We found a linear relationship between female body mass and ON for each of the different diets studied (Fig. 1). Female body mass was not significantly different between treatments (one-way ANOVA; F 2,225 =2.371, P=0.311). We did, however, observe significant differences in ON per female between the HP diet and the other two diets (ANCOVA; F 2,225 =27.315, P <0.05; a posteriori Tukey test P <0.0001).

Fig. 1
figure 1

Fecundity of females from the three diet groups. The regressions are based on data obtained from females that recently released offspring (high carbohydrate: ON =0.492–3.107 BM, R 2 =0.59, n =72, P <0.05; high protein: ON =0.452–1.076 BM, R 2 =0.61, n =71, P <0.05; control: ON =0.508−5.581 BM, R 2 =0.65, n =80, P <0.05; ON offspring number; BM body mass; n number of observations)

Full size image

The smallest offspring were found in females fed the HP diet, for which OS averaged 1.575 mm (SD=0.39; Fig. 2). Females on the HC diet (x̄ =1.765; SD=0.17) and C diet (x̄=1.787, SD=0.19) produced considerably larger offspring than females on the HP diet (Fig. 2A). The recently released offspring were significantly different between HP/HC and HP/C (ANCOVA; F 2,1408 =12.580, P <0.0001; a posteriori Tukey test, P <0.0001), but not between HC/C (a posteriori Tukey test, P=0.071; Fig. 2A).

Fig. 2 A
figure 2

Offspring size at release (in mm) and B duration of embryonic development (in days) from females maintained on the three different food quality treatments. An asterisk indicates significant differences at the 1% level

Full size image

Incubation period

Embryonic development under laboratory conditions (21°C and 14L:10D) averaged 22.77 days (Fig. 2B). The maximum and minimum incubation periods were 40 and 15 days for the HC and C diets, respectively. IP of eggs was significantly different between food treatments (one-way ANOVA; F 2,225 =5.479, P <0.0001). An a posteriori test revealed that the incubation period was significantly smaller for the HC diet group (x̄ =21.0 days, SD=3.52) compared with the HP (x̄=23.7 days, SD=4.11) and C (x̄ =23.4 days, SD =3.84) diet groups (Tukey test, P <0.0001; Fig. 2B).

Growth rate

Adult male P. laevis from all three diets maintained their weight from the beginning to the end of the experiment. Analysis of adult (males) growth curves for the three diets indicated that GRA was similar for all isopods, independent of the diet that they were fed (Table 2). There was no significant difference in GRA between the three diets (one-way ANOVA; F 2,82 =1.521, P =0.152). However, GRA for the HP diet was similar to that of the C diet, and GRAs in both of these diets were lower than for the HC diet (Table 2).

Table 2 Effects of food quality on growth rates of adults, and subsequent effects on parameters of juvenile growth rates: W ∞ is asymptotic weight; k is the growth rate coefficient; t 0 is time to 0 mass; n is the number of observations; R 2 is the fit of data to the von Bertalanffy growth model
Full size table

Juvenile P. laevis from mothers reared on the HP diet grew significantly more rapidly (by about 77%) during the 15-week period than isopods reared on the HC and C diets (one-way ANOVA; F 2,1650 =79.808, P <0.0001; Table 2). These latter two diets were not significantly different from each other (a posteriori Tukey test, P =0.381). For juveniles, the coefficient of growth rate, k, was higher for the HP diet (x̄ =0.328, SD =0.03) than the HC (x̄ =0.152, SD =0.02) and C (x̄ =0.155, SD =0.04) diets (see Table 2).

Life history trade-offs

Female body mass was negatively correlated with IP (R 2 =0.14, P <0.0001, R 2 =0.21, P <0.0001, R 2 =0.15, P <0.0001; for C, HP and HC diet, respectively), and strongly positively correlated with ON (see Fig. 1). However, body mass was not correlated with OS (C: R 2 =0.007, P =0.0658; HP: R 2 =0.002, P =0.362; HC: R 2 =0.005, P =0.173) or GRJ (C: R 2 =0.002, P =0.384; HP: R 2 =0.003, P =0.211; HC: R 2 =0.006, P =0.471). Using residuals from the significant relationships with body mass (i.e. OS and ON), we found a negative correlation (i.e. trade-off) between OS and ON for the HP diet (r p =−0.46, P <0.001), but this significant correlation did not hold for the other two diets (C: r p =−0.051, P =0.267; HC: r p =−0.071, P =0.183; Fig. 3A–C). Furthermore, using residuals of IP and ON, we found a trade-off between these two traits for the HP group (r p =−0.21, P <0.05) and C (r p =−0.20, P <0.0001), and a change to a positive, significant correlation for the HC diet group (r p =0.31, P <0.0001; Fig. 3D–F). Finally, GRJ did not reveal significant correlations with the other studied traits for any of the different diets of the females: OS (C: r p =0.022, P =0.172; HP: r p =0.047, P =0.087; HC: r p =0.011, P =0.364), ON (C: r p =−0.011, P =0.138; HP: r p =0.022, P =0.324; HC: r p =−0.055, P =0.125) and IP (C: r p =0.052, P =0.254; HP: r p =0.078, P =0.072; HC: r p =0.068, P =0.155).

Fig. 3
figure 3

The relationship between residual offspring number and offspring size (A–C) and residuals of incubation period and residual offspring number (D–F) in the terrestrial isopod Porcellio laevis. Dotted lines indicate the correlation within the high-carbohydrate (HC) treatment. Continuous lines denote the correlation within the high-protein (HP) treatment. Point-segment lines indicate the correlation within the control treatment (C). An asterisk indicates statistically significant within treatment correlations (P <0.0001)

Full size image

Discussion

All life history traits, i.e. offspring size, clutch size, incubation period, and juvenile growth rate, varied significantly with food quality in P. laevis. The quality of the nutritional items included in the diet evidently has important consequences on female life history and offspring growth, but this effect acts in different directions. Our results support the notion that protein, rather than carbohydrate, concentrations in food constrain the performance of herbivorous arthropods, and have significant consequences on life history traits, as was seen in P. laevis. The most important findings of the present study are as follows: (1) there is a substantially elevated maternal effect on offspring growth rate; (2) the nutritional quality of resources available to mothers has important effects on offspring phenotype (size and number); and (3) our results are consistent with the expectation that differences in resource acquisition among mothers can influence the detection of life history trade-offs (van Noordwijk and de Jong 1986), and in this study we show that differential availability of essential nutrients can also affect these trade-offs.

Life history traits

Food quality affected the life history traits in our study model in contrasting ways. Offspring size was smaller in animals fed the HP diet. Several authors have agreed that, especially for arthropod herbivores, mothers in resource-rich environments produce smaller offspring compared with females in resource-poor environments (Taylor 1985; Brody and Lawlor 1984; Reznick et al. 1996). Generally, in the above studies resource-rich environments are associated with environments with the highest protein availability. Results from our study suggest that protein availability may be one of the most important causes of the observed differences in offspring size at release for P. laevis. Furthermore, other studies have shown that OS can be quite responsive to changes in dietary nitrogen in several arthropod species (see Joern and Behmer 1997). Juvenile size at release and/or egg size is related to the fitness of individual offspring in many arthropods (Capinera 1979; Fox and Czesak 2000), including terrestrial isopods (Lawlor 1976; Brody and Lawlor 1984). In nature, a smaller initial size could result in higher juvenile mortality for at least two reasons. First, smaller young may have lower energy reserves and, consequently, would be more sensitive to periodic food shortages. Second, and probably a more important effect of smaller initial size, is that the juveniles remain small for a longer time. Consequently, these smaller juveniles are exposed to a greater predation risk (Lawlor 1976; Warburg 1987). That is, diet quality, especially total nitrogen content can significantly affect juvenile survival, and at the same time affect the fitness of the females.

Carefoot (1993a, 1993b) did not observe dietary requirements for amino acids in a terrestrial isopod, suggesting a quite remarkable and unique independence for nutrients known to be essential for other animals. Our study demonstrates that the presence of protein in the diet does significantly contribute to higher production of offspring by the mother, which would have profound consequences on female fitness. When considering offspring production in equally sized mothers of P. laevis, the number of offspring per unit weight of the mother decreased from the high protein (HP) diet to the low protein diet (HC; see Fig. 1). That is, the main effect of protein supplementation was to increase the number of offspring produced. Similar observations have been reported for ants, butterflies, flies, and grasshoppers, among other arthropods (Joern and Behmer 1997; Aron et al. 2001; Romeis and Wäckers 2002; Wall et al. 2002).

In contrast to the significant variation observed in juveniles, adult growth rate was not significantly affected by diet quality (see Table 2). Helden and Hassall (1998) reported that adult growth rate was strongly influenced by food quality in the terrestrial isopod, Armadillidium vulgare. The difference between these results and our own can be explained by the fact that Helden and Hassall (1998) determined growth rate in adult females while we measured growth rate in adult males. Obviously, the growth pattern between sexes is very different in woodlice species (Warburg 1993), and growth is more variable in reproductive females.

We found that increased carbohydrate concentrations in the diet shortened the duration of the incubation period in the experimental group; this is likely because the time required for egg synthesis is biochemically dependent. A shortening of the incubation period with added carbohydrates has also has been observed in insects such as Lepidoptera and Hymenoptera (Romeis and Wäckers 2002; Özalp and Emre 2001). On the other hand, dietary carbohydrates seem to be nonessential for crustaceans (Harrison 1990), only their specific roles in the production of nucleic acids and as a component in ovarian pigments may be especially important for maturation processes, such as oogenesis, vitellogenesis, and embryogenesis. Although, in general, crustaceans produce a large number of heavily yolked eggs, between different species, there is an apparent correlation between the amount of yolk provided and the duration of embryogenesis as well as with the level of development of the hatched offspring.

In crustaceans the total dietary protein requirement is typically higher for young, rapidly growing animals than for adults, due to the need for rapid tissue synthesis (see Harrison 1990). Nitrogen in particular, has been implicated as a critical nutritional element for the growth of many herbivorous arthropods. These beneficial effects are manifested in many ways, through increases in individual growth rate, longevity, fecundity, reproductive state, and/or survivorship. In this study we have shown that the availability of high-protein (i.e. nitrogen) plays a dominant role in determining characteristics of offspring growth in P. laevis. Regarding diet, studies conducted in terrestrial isopods indicate that food quality can cause important changes in individual and population parameters. Merriam (1971) and Rushton and Hassall (1983a) demonstrated that food quality can significantly change growth rate in Armadillidium vulgare, directly affecting the age at the first reproduction, which is a key factor controlling population growth.

Our findings may apply to other herbivore arthropods, since the proportion of nitrogen in animal biomass is usually much higher than that of their food plants (Awmack and Leather 2002). For arthropod herbivores, available dietary nitrogen (in the form of protein and free amino acids) can potentially limit many important processes (McNeill and Southwood 1978; White 1993; Joern and Behmer 1997). An implication of our results is that, for juveniles of P. laevis, and perhaps for the growth stages of other herbivorous animals, energy budgets may be of less significance than nitrogen budgets.

Given the seasonal variability in reproductive characteristics of other woodlice species (Sunderland et al. 1976; Davis 1978; Warburg 1987; Willows 1987; Helden and Hassall 1998) the ability to vary life history traits (such as incubation period, offspring size, and offspring number) to match changing environmental conditions may be an adaptive strategy common among woodlice.

Trade-offs and maternal effects

Life history theory predictions regarding the ubiquity of trade-offs are rarely confirmed in empirical studies (Stearns 1992; Jordan and Snell 2002; Roff 2002). The amount of resources a female can provide to her progeny is limited by her ability to garner resources, and by her own needs for assimilated energy, including maintenance (Dunham 1978). In this sense, if parental resources are limited, increasing resource allocation to one offspring means that the other offspring must either be smaller or fewer in number. This intuitive trade-off between offspring size and number has received a great deal of theoretical and empirical study (McGinley and Charnov 1988; Bernardo 1996). In our study, implicit in the observation that trade-offs are obscured by differences in diet quality (i.e. essential nutrients) is the suggestion that diet quality differences are the main cause of changes in trait values. The negative correlation between offspring number and offspring size in the HP diet indicates that carbohydrates are the limiting essential nutrient. Carbohydrates play a critical role in the production of glucosamine, which is the precursor in the synthesis of chitin, the principal constituent of crustacean exoskeletons (Harrison 1990). Therefore, in our study model progeny size was limited by the availability of precursors for their exoskeletons. These results match those of Lawlor (1976) and Willows (1987) who found a negative correlation between fecundity and offspring size in the natural environment for the terrestrial isopods Armadillidium vulgare and Ligia oceanica, respectively

Assuming that the same amount of energy is invested in offspring production, a reproducing female can follow one of two different strategies: produce a large quantity of smaller offspring or a lesser quantity of larger offspring. Females in HP diet group chose the former strategy, producing a large number of small offspring. So the question arises as to why females did not produce fewer, larger progeny? We suggest two possible explanations. First, offspring growth rate is higher for individuals on the HP diet compared to other experimental diets, therefore these smaller individuals should be able to reach a larger size in a short amount of time; so mothers would favor the strategy producing more individuals with high performance (i.e. growth rate). Second, the HP diet is devoid of carbohydrates, and juvenile size is probably limited by the precursors needed to build the chitin exoskeleton.

The significant correlations between incubation period and offspring number were opposite for the HP/C diet groups with respect to the HC diet group. This is an empirical demonstration of the context dependence of life history trade-offs. In the HP and C diets, the time wasted in incubation is dependent on clutch size; females that produce a larger quantity of offspring have a shorter period of incubation. This trade-off can be explained by the fact that egg volume increases throughout embryonic development, ranging from 87% to 146% (Lardies et al., in preparation). In crustaceans, the physical space available to carry eggs under the cephalotorax, or in the marsupium, is a limiting factor for egg production (see Corey and Reid 1991; Hines 1992; Warburg 1993). Therefore, in the HP and C diets, females with greater offspring number may incite an earlier juvenile release due to space limitation, while females with fewer offspring could accommodate juveniles for a longer period. However, this does not explain the positive correlation between incubation period and offspring number observed for the HC diet. We do not know for sure the reasons for the increase in incubation period with offspring number in the HC diet. Since energy retention in crustaceans is more efficient under a higher protein diet (Cuzon et al. 2000), we suggest that a larger quantity of eggs would increase the incubation time for energetic reasons. This idea is supported by the significantly shorter incubation period for females on the HC diet. Nutrition and excretion processes from egg mass would be more energetically expensive with more progeny in the marsupium.

In the case of females placed on the HC diet, if resources are superabundant and individual energy acquisition is the same, a trade-off between IP and ON could be masked by an increase in the values of both traits. However on the HP and C diets, the lack of carbohydrates will trigger the trade-off between incubation period and offspring number. In general, there are two possible explanations for changes in trait correlations. First, trait correlations can be plastic, so the correlation between two traits can be altered by the environment, and plasticities themselves can be correlated (i.e. the reaction norms of different traits can be correlated to yield coherent multivariate phenotypic plasticity; see Schlichting and Pigliucci 1998).

Intra-individual trade-offs have played a much larger role to date than intergenerational trade-offs. In our study, diet quality changed the female allocation strategy, which has profound implications on juvenile size and, probably, on size at maturity. Thus, in many ectothermic animals the fastest growing phenotypes produce the smallest adults (Stearns 1992). In our study, offspring growth rate was significantly affected by the quality of the diet fed to the mother. In woodlice juvenile growth rates are a critical factor determining age at reproduction (Grundy and Sutton 1989). Our study supports this tendency since the asymptotic weight (W ∞) of juvenile P. laevis is smaller for the HP diet group (~89 days) compared to juveniles on the C and HC diets, which have W ∞ of about 138 and 145 days, respectively (see Table 2). Terrestrial isopods exhibit a high level of parental care, first by carrying their offspring in a marsupium while they undergo embryonic development, and second, by supplying them with nutrients within the marsupium (Warburg 1987; Warburg and Rosenberg 1996). These findings indicate a significant maternal provisioning of organic matter in vivo, presumably via the cotyledons, as proposed by Hoese and Janssen (1989). This is further supported by the intimate association between the cotyledons, the maternal fat body, and the depletion of maternal fat reserves observed during the course of development (Wright and Surbida 2001). In our study, these facts may explain the higher growth rates for the offspring of females fed a high protein diet. That is, offspring are provided with food by the mother during embryonic development, which alters their growth via the availability of protein for biosynthesis. However, the growth rate of P. laevis juveniles from females fed the HP diet was higher for at least 100 days, indicating that our latter explanation is not sufficient to explain these strong differences. A more complete explanation for this dramatic change in growth rate is that the studied traits, which are mostly physiological, can be switched on or off as a consequence, or as a response to, environmental challenge (Spicer and Gaston 1999). That is, an early stimulus, or environmental challenge, that operates at a critical or sensitive period, can result in a long-term change in the structure and function of an organism (Lucas 1991).

There is abundant evidence that the impact of offspring size phenotype on offspring performance (growth, survival, etc.) is contingent on larval environments, where variation in prey availability, conspecific density, and temperature have been shown to modify egg size effects (for review see Fox and Czesak 2000). Nevertheless, in our study there is no relationship between offspring size and their growth rate, rather both characteristics are affected in different ways depending on the maternal diet. We presume that the reproductive strategy of females from the HP diet was to maximize progeny production, although they produced smaller young which grew at a higher rate compared with other experimental environments. Thus, the females were somehow able to register their environment, and to choose a strategy that favored their own performance and that of their progeny. On the other hand, females under the C and HC diet treatments, that is, diets more similar to the actual composition of available food in the natural environment (see Joern and Behmer 1997; Awmack and Leather 2002), produced fewer, larger progeny, which had slower growth rates. We believe that this strategy has its basis in the fact that larger progeny have reduced predation risk and are more resistant to periodic food shortages (Fox and Czesak 2000), which consequently increases juvenile survival.

Female fitness is ultimately determined by two factors: the number of offspring produced, and the subsequent fitness of those offspring. Any trade-off between female fecundity and subsequent offspring fitness can be an important selective factor in the evolution of fecundity (McGinley and Charnov 1988). We have demonstrated that environment, through food quality, differentially affects female trait strategies as a response to shortage of essential nutrients. Even though, in the long-term, this would become an important selective pressure, exerting directional selection on the evolution of fecundity. Our results demonstrate the plasticity of reproductive traits in females and how changes in relation to diet can have profound effects on the progeny. Since food quality varies greatly for arthropod herbivores, on both temporal and spatial scales (e.g. leaves vary in composition between different species), we believe that the plasticity of female life history traits in response to this environmental factor (food quality) is adaptive.

The strong maternal effect detected in the growth rates of juvenile P. laevis, demonstrates that mothers can directly affect the condition of their offspring. These experiments support our hypothesis regarding the importance of the maternal effect: we predicted that the rearing diet of females would significantly affect the size and growth rate of offspring. Thus, the quality of resources an individual offspring obtains during its development and growth, the length of its incubation period, and the environment into which it is born is, to a large extent, determined by its mother. All of the parameters evaluated in this study changed as a function of maternal diet. However, variation in maternal resources has diverse effects on offspring phenotype, making it difficult to state, with generality, how resource availability affects offspring phenotype. This is because the offspring number/size phenotype that a mother produces, and the relative fitness that she confers to her developing progeny, depend on the distinct environmental contexts in which these activities occur (Bernardo 1996).

Finally, we hope that this kind of empirical approach will stimulate researchers to examine not only intra-generational variation in life history traits and trade-offs, but also intergenerational trade-offs, which are far more informative regarding the evolutionary consequences of maternal effects on offspring phenotypes beyond genetic inheritance.