Heikki Pöykkö, Department of Biology, University of Oulu, Oulu 90014, Finland. E-mail: firstname.lastname@example.org
1. Egg size is often used as a proxy of egg quality although size and composition may vary, e.g. in insects egg size usually decreases as female ages. Whether this decrease in size reflects reduced concentrations of essential nutrients such as lipids and proteins of eggs laid by ageing females, or does reduced size per se explain often observed lower fitness of later laid eggs is poorly explored.
2. Egg properties were compared with fitness parameters of offspring laid on the first and fourth night during the oviposition period of a capital breeding moth, Cleorodes lichenaria (Hufnagel). The study aim was to explore whether decreased egg size is caused by decreased provisioning into later laid eggs measured as egg protein and lipid concentration, and whether it results in lower fitness of later laid offspring.
3. The fresh and dry weight of eggs decreased over the oviposition period, but the protein and lipid concentration remained constant. Survival of larvae was lower among the fourth night laid offspring on a low quality host Parmelia sulcata Taylor compared to a high quality host Ramalina fraxinea (L.) Ach. No differences were observed in egg fertility or hatchability, neonate survival without food and pupal mass between the offspring produced on different nights.
4. Decreased survival of offspring produced later was rather attributable to absolute provisioning (i.e. lower weight of eggs) than relative provisioning (i.e. decreased concentrations of nutrients in eggs). It is argued that lower survival of later laid smaller eggs on low quality diet is likely attributable to physical and chemical characteristics of host lichens and/or physical properties of tiny neonate larvae.
Decrease in the size of eggs during the female's lifespan is commonly assumed to reflect decreased provisioning of nutrients and/or energy into eggs (Begon & Parker, 1986; Bernardo, 1996; Fox & Czesak, 2000). While the decrease in the size of eggs likely reflects decreased provisioning in absolute terms (i.e. lower absolute amount of nutrients and/or energy in an egg), it needs to be separated from relative provisioning (i.e. changes in concentration of nutrients in an egg) to reveal changes in the quality of eggs other than size per se. Thus, absolute provisioning is an applicable measurement when provisioning is considered from the females' viewpoint, but relative provisioning indicating qualitative changes in eggs might better apply when predicting offspring performance. There are some examples showing that concentrations of essential nutrients in eggs, i.e. proteins (Giron & Casas, 2003) or lipids (McIntyre & Gooding, 2000), decrease as the female ages. In other cases either no differences exist in the composition of eggs, or females may even increase their provisioning when ageing depending, for example, on the nutritional environment (Karl et al., 2007; Geister et al., 2008).
The egg contains all nutrients and energy necessary for a developing embryo. Water is the most voluminous ingredient in insect eggs composing ca. 64–90% of fresh weight, followed by lipids, proteins, and minor amounts of glycogen and carbohydrates (Chapman, 1975; Kawooya & Law, 1988; Garcia-Barros, 2006; Geister et al., 2008). Lipids are the main source of energy for a developing embryo while proteins are mostly used for tissue construction during embryogenesis (Chapman, 1975; Van Handel, 1993; Forte et al., 2002). The importance of egg proteins for offspring development is highlighted by the fact that all essential amino acids need to be obtained during the larval stage (O’Brien et al., 2005). High water concentration in the egg may increase the survival of developing embryo and hatching success of neonate larvae (e.g. Geister et al., 2008).
The decreased provisioning of ageing females into their offspring, decrease in egg size during the oviposition period, and impact of egg size on the performance of larvae are all relatively well documented phenomena. However, there is a surprising lack of studies showing causality between these variables. For example, it is still unclear whether decrease in egg size during the oviposition period and its potential consequences for performance of larvae are caused by decreased relative provisioning (decreased concentration of nutrients) or absolute provisioning (decreased size of eggs per se) of ageing females (McIntyre & Gooding, 2000). To fill this gap, we studied the causes of decreased egg size in ageing females of a geometrid moth, Cleorodes lichenaria, and explored whether there are any differences in the performance between offspring laid early in the reproductive period compared with those laid later. As C. lichenaria is a capital breeder (Pöykkö, 2009) and, consequently, unable to replenish nutrient reserves at the adult stage, we predicted that decreased allocation per individual egg (absolute and/or relative) during the oviposition period should result in decreased performance of the offspring laid later along the female's life, i.e. egg fertility, egg hatchability, starvation resistance, survival, and attained final pupal size of later laid individuals should be lower compared with first laid bigger eggs. Finally, this difference should be clearer among larvae fed on low quality diet than among those fed on high quality diet.
Materials and methods
Cleorodes lichenaria (Hufnagel) is a geometrid moth that feeds on lichens during the larval stage (Robinson et al., 2000). Adults fly from the end of June until the beginning of August. Females lay eggs singly on lichens and hatched larvae overwinter as middle-sized larvae (fourth or fifth instar) on tree trunks and resume feeding next spring. Females mature with half of their egg load ready for oviposition, which they oviposit during the first night after mating (Pöykkö, 2009). The rest of the eggs can be manufactured without any feeding at the adult stage, which lasts ca. 12 days. Approximately 70% of eggs will be laid during the first three nights (Pöykkö, 2009). Larvae pupate inside cocoons made from silk and pieces of lichens in May or the beginning of June.
Two lichen species were selected as fodder for larvae in the experiments. Ramalina fraxinea (L.) Ach. is preferred both by females and neonate larvae as oviposition and feeding substrate to other potential hosts, such as Parmelia sulcata Taylor and larvae receive enemy-free space against generalist natural enemies on R. fraxinea (Pöykkö, 2006, 2011). Previous laboratory experiments have shown that on the preferred R. fraxinea larvae grow faster, have higher survival and pupate with higher pupal mass compared with larvae reared on P. sulcata (Pöykkö, 2006; Pöykköet al., 2010). Growth of C. lichenaria larvae on P. sulcata is hindered by its secondary metabolites, atranorin and salazinic acid (Pöykköet al., 2010). These results unequivocally show the superior host quality of R. fraxinea to larvae of C. lichenaria compared to P. sulcata. Parmelia sulcata is even more common than R. fraxinea in the population where females originated (Pöykkö, 2006), so ovipositing females and wandering larvae are likely to encounter lower quality hosts regularly. Thus, lower performance of smaller eggs and larvae especially on P. sulcata would indicate real-life fitness costs for decreased egg size.
Larvae as well as lichens used as a fodder originated from the Jomala population from the Åland Islands (60°11′N, 20°00′E) (Pöykkö, 2006). All eggs and hatching larvae used in the experiments were obtained from females reared on R. fraxinea throughout their larval period. Each female was mated to a male in a small plastic box (0.5 l) with a small piece of R. fraxinea for a female to lay her eggs. Eggs were carefully detached from lichens, counted and weighed daily during the first 8 days of the oviposition period. Oviposition experiments were carried out in the LD 18:6 photoperiod and 20:16 °C temperature to reflect natural conditions during the oviposition period of the species.
Measurements of egg properties
Eggs from 8 and 10 females, respectively, were used for measurements of egg weights and contents. Ten to 15 eggs per female in each day were randomly selected for the measurements of fresh and dry weights of eggs and another 10–15 eggs for the measurements of protein and lipid concentration. If females laid fewer than 15 eggs during a night, eggs were divided in those two treatments so that at least eight eggs were distributed for protein and lipid analyses, the rest remaining for weight analyses. The eggs laid between the fifth and eighth night were pooled in to two samples, the first containing eggs oviposited during the fifth and sixth night and the second eggs from the seventh and eighth night. This was essential to ensure enough material for protein and lipid analyses, as the number of eggs decreases towards the end of the oviposition period (Pöykkö, 2009). The eggs laid on the fifth and seventh night were stored in a dark fridge (+ 2 °C) until the next day, when the eggs laid on the sixth and eighth night were added to the samples. With this procedure, we aimed to inhibit the developing of embryos and to prevent changes in egg composition of eggs laid on the fifth and seventh nights. Females lay over 90% of their eggs during the first 8 days, which is almost the entire oviposition period of the study species (Pöykkö, 2009).
The clutch of eggs were weighed immediately after removal from lichens (or after the sixth and eighth night during the second half of the oviposition experiment), killed in a freezer and dried for 48 h in 55 °C in an oven and reweighed to measure the dry weight of eggs. Measurements were performed with a precision balance (Mettler-Toledo MT 5) to the nearest 0.1 µg.
Colorimetric determination of total protein and lipid content
Total lipid content was measured using the sulphophosphovanillin method (Frings et al., 1972; Lorenz, 2003) with some modifications. Samples were homogenised manually (plastic to plastic) in 100 µl of methanol. After homogenisation, 10 µl was transferred into a borosilicate test tube, 100 µl sulphuric acid (95%) was added and the sample was evaporated to dryness at 100 °C and cooled for 5 min. Five millilitres of phosphovanillin reagent was added and the samples were incubated for 15 min. After cooling, the samples were measured spectrophotometrically against olive oil standards at 530 nm.
The amount of total protein in the samples was determined according to Bradford (1976). Ten microlitres of the homogenate was diluted with 0.9% NaCl solution 1 : 20. Two hundred microlitres of Bradford protein assay dye reagent (Bio-Rad, Hercules, CA, USA) was added to 10 µl of diluted homogenates and standards (bovine serum albumin in NaCl solution). The absorbance of samples was measured at 595 nm with Victor3 (PerkinElmer, Waltham, MA, USA) spectrometer.
Although carbohydrates (e.g. glycogen) occur in eggs that are also used as an energy source for metabolism, we had to ignore them in our analyses. Egg fresh mass used for chemical analyses varied between 1 and 2 mg and thus it was not possible to measure carbohydrates as we concentrated on most important nutrients. Carbohydrates occur in relatively low concentrations in insect eggs and are not as crucial for fuelling metabolism as lipids (Kawooya & Law, 1988; Van Handel, 1993; Geister et al., 2008; Sloggett & Lorenz, 2008).
Absolute versus relative provisioning
Decreased provisioning by ageing females may be a result of decreased allocation of any crucial nutrient to eggs or simply result from the decreased total amount of nutrients allocated to eggs. Thus, to separate absolute (mass of any individual nutrient in an egg) and relative provisioning (percentage of individual nutrients) we calculated both absolute and relative amount of lipids and proteins in eggs. Absolute allocation was measured as micrograms of nutrients in an egg and relative allocation as a percentage of individual nutrients of total egg fresh weight.
Performance of the first and fourth night laid offspring
The offspring of six and four females in 2007 and 2008, respectively, were used to compare the performance of the first and fourth night laid offspring of C. lichenaria. We had to use the offspring of different females, as in the analyses of egg compositions, because the number of eggs oviposited during the fourth night is usually between 20 and 30 and not enough to perform all analyses from the same clutch. In 2007, 30 randomly selected eggs per female laid during both the first and fourth night of the oviposition period were selected for the experiments. The eggs oviposited during the fourth night were used in the experiment for two reasons. First, there is a significant difference in the size of eggs oviposited during the first night compared to those oviposited on the fourth night (Pöykkö, 2009). Second, the number of eggs oviposited on the fourth night was enough for the experiment (usually about 20). If females laid fewer than 30 eggs during the fourth night, eggs were divided evenly into the three treatments described above.
The eggs were weighed and placed singly on Petri dishes (diameter 35 mm). The viability of eggs was monitored and mortality factors attributed either to sterility of eggs or to mortality at the egg stage according to egg colour (N = 57 and 32 for eggs laid on the first and fourth night, respectively). Immediately after ovipositioning eggs are bright green and when developing, fertilised eggs turn to dark brown and finally almost black, whereas sterile eggs remain green (Pöykkö, 2009). Accordingly, eggs remaining green were regarded sterile and eggs turning to dark brown but remaining unhatched were regarded dead at egg stage. Thus, fertility was determined as a proportion of eggs changing their colour into dark brown of all laid eggs and egg hatchability as a proportion of fertile eggs that hatched. The egg development time was recorded for hatched eggs. The larvae hatching from these eggs were again weighed. These hatched larvae were divided into three treatments: to be reared on high quality R. fraxinea, low quality P. sulcata, or to a neonate survival experiment. In laboratory conditions, C. lichenaria larvae grow slower and have a higher mortality on P. sulcata compared with larvae reared on R. fraxinea (Pöykköet al., 2010). The larvae in the neonate survival experiment were kept in dishes without food and they were monitored daily to determine how many days they survived. Larvae were regarded as dead when they did not react when gently touched. The experiments were carried out in a climate chamber with LD 18:6 photoperiod and 20:16 °C temperature to reflect natural conditions. In 2008, the experimental procedure was similar to that in 2007 except that the offspring were not monitored individually during the egg stage and were distributed only to the two host lichen treatments.
Before offering to the larvae, lichens were kept overnight in a chamber with 100% relative humidity. After that, lichens were sprayed with deionised water and placed in Petri dishes for the larvae to feed on. A similar amount of lichen material was offered to larvae in each treatment. Lichens were moistened every 3 days with a few drops of deionised water and changed for new ones in 2 weeks. The larvae were reared for 2 months in LD 16:8 photoperiod and 20:16 °C temperature, and during the next 50 days environmental conditions were gradually changed to the LD 8:16 photoperiod and 8:4 °C temperature to reflect ending of the season (N = 60 and 32 for first and fourth night laid larvae, respectively on both R. fraxinea and P. sulcata in 2007; N = 30 and 26 for first and fourth night laid larvae, respectively on both R. fraxinea and P. sulcata in 2008). After that, the larvae were overwintered in a dark cold-storage room (2 °C) for 3 months. After overwinterting the larvae were reared on plastic cups (45 ml) covered with veiling and sprayed every third day with deionised water. The larvae were reared in similar environmental conditions as before overwintering but in a reverse order (i.e. starting with the LD 8:16 photoperiod and 8:4 °C temperature). Survival, pupal mass, length of larval period, and sex were recorded. Sex of pupae was determined according to genital scars.
Linear mixed models in statistical package R 2.8.1 were used for statistical analyses (R Development Core Team, 2009). Repeated-measured analyses of changes in egg properties (dry and fresh weight, absolute and relative protein and lipid concentration) were performed with linear mixed effect models (function ‘lme’) with oviposition day as an explanatory factor and brood (i.e. individual female) as a random factor. Day was set as an ordered factor to get orthogonal polynomial contrasts for oviposition day to study whether egg weights and lipid and protein concentrations decrease over the oviposition period (Bates et al., 2008). We also added female pupal mass as a covariate in the analyses to explore whether females with different body mass differ in provisioning. As inclusion of pupal mass decreased model-goodness-of-fit of all models indicated as increased Akaike's information criterion (AIC) values, it was omitted from final models. Owing to variable residual variances between oviposition days in all measured egg traits (Fig. 1), heteroscedasticity was modelled with power variance function (function varPower) (Pinheiro et al., 2008). Pearson correlation was used to explore the relationship between the dry and fresh weight of eggs as well as fresh weight of eggs and hatching neonate larval weight.
The generalised linear mixed effect model (GLMM) for the survival of the larvae was fitted with the function ‘glmer’ with binomial error distribution and logit link function with host lichen and day of egg oviposition as fixed factors using the Laplace approximation (Bates et al., 2008). Egg fertility and egg hatchability were similarly analysed with GLMM models. Linear mixed effect models (function ‘lme’) were used to compare the duration of egg period, endurance of unfed neonate larvae, pupal mass, and length of development time between the first and fourth night laid offspring (Pinheiro et al., 2008). Duration of neonate larval survival without food was analysed with the day of egg oviposition as a fixed factor. Pupal mass and length of larval period were analysed with host lichen, day of egg oviposition, and sex of an individual as fixed factors. Fitting a random factor to the models was started with brood nested within year and proceeded with only brood as random factor, except in the analysis of neonate survival without food, when only brood was used as a random factor (experiment was performed only in 2007). Finally, to explore the need for the used random factors, GLM models without any random factor with ‘glm’ function for binomial variables (survivals, egg hatchability and fertility) or a generalised least square linear models with ‘gls’ function for continuous variables (egg time, neonate survival time, pupal mass, and length of larval development) were fitted to the data. Estimations of goodness-of-fit of all models were checked by visual evaluation of residual plots and with AIC. Subsequently, fixed factors were hierarchically reduced by removing insignificant interaction terms to get the definitive models.
Both linear and quadratic contrasts explained variation in the fresh and dry mass of eggs during the oviposition period of C. lichenaria (Table 1). The negative linear contrasts imply that egg weights decreased over the oviposition period and positive quadratic contrasts imply that this reduction in size decreased towards the end of the oviposition period and even slightly increased during the very last nights (Fig. 1a). Both fresh and dry masses of eggs decreased almost in a parallel way (Fig. 1a,d). Negative linear contrasts for the absolute amount of proteins and lipids imply that the amount of these nutrients in eggs decreased during the oviposition period (Fig. 1b,c, Table 1). Changes in egg composition over time (relative provisioning), however, were statistically insignificant (Fig. 1b,c, Table 1). There was also a significant quadratic contrast for the change in protein concentration of eggs, attributable to a slight increase during the second oviposition night and subsequent decreasing protein concentration of the eggs (Fig. 1b). No observable differences, however, existed in protein concentration between eggs laid on the first and fourth night (Fig. 1b). In addition, lipid concentration remained constant during the oviposition period (Fig. 1c, Table 1). Water concentration increased slightly but significantly from 68.6% on the first night to 70% during the last nights as indicated by significant positive linear contrast (Fig. 1d, Table 1). The dry weight of eggs correlated positively with the fresh weight of eggs (r = 0.771, d.f.= 219, P < 0.001). Moreover, egg weight correlated positively also with neonate larval weight (Fig. 2).
Table 1. Orthogonal polynomial contrasts for models of egg properties of Cleorodes lichenaria exploring changes of egg characteristics during the oviposition period.
Only linear and quadratic contrasts are shown.
No differences were found in egg fertility [96.6 ± 1.3 and 94.1 ± 2.4 (mean % ± 1 SE)] or egg hatchability [95.4 ± 1.5 and 95.8 ± 2.1 (mean % ± 1 SE)] between eggs laid on the first and fourth night, respectively (Table 2). The offspring laid on the fourth night had statistically significantly longer development time [12.8 ± 0.05 and 13.0 ± 0.06 (mean days ± 1 SE), Table 2], although this difference of ca. 5 h is probably without any biological significance. In the neonate survival experiment without food no differences were found in the survival time between the larvae eclosing from the eggs laid on the first and fourth night (4.19 ± 0.12 and 4.06 ± 0.14 days ± SE, respectively, Table 2). Results of the GLMM for larval survival during the whole larval period, however, revealed that oviposition day and host lichen had a significant impact on the survival of larvae and that survival of larvae from the eggs laid on the first versus the fourth night differed between the host lichens (Table 2). There were no differences in survival of the larvae laid on different nights on R. fraxinea, but overall survival was lower on P. sulcata and larvae from eggs laid on the fourth night had lower survival than from eggs laid on the first night on low quality diet (P. sulcata) in both years (Fig. 3). There was also some variation in the survival of larvae between years, overall survival being higher in 2007 and lower on P. sulcata in 2008 compared with that observed in 2007 (Fig. 3). This yearly difference may be attributable to differences in resource allocation to offspring between individual females (only four broods used in 2008), or alternatively to yearly differences in nutritional value of P. sulcata as a host for larvae. The previous explanation may be supported by the fact that the developmental period was also in general longer in 2008 (Fig. 5). Trends in the survival of offspring laid on different days, however, were similar in both years on P. sulcata (Fig. 3). Inclusion of year into a random factor resulted in lowest AIC value only when the length of larval period was considered, indicating that yearly differences in studied traits were negligible (Table 3).
Table 2. Fixed effects of generalized linear mixed models for offspring life-history traits exploring differences between the first and the fourth night laid offspring of Cleorodes lichenaria in the explored life-history traits.
*Z-statistics and number of replicates are given for the results of analyses of egg fertility, egg hatchability and survival of larvae and t-statistics and d.f. values for egg time, neonate survival time without food, larval period, and pupal mass.
−0.57 ± 0.57
3.34 ± 0.38
1.35 ± 1.01
−5.77 ± 1.48
0.18 ± 0.08
12.85 ± 0.11
Neonate survival time without food
−0.15 ± 0.17
4.11 ± 0.18
0.94 ± 0.33
−1.09 ± 0.38
Host lichen × day
1.13 ± 0.55
11.43 ± 1.11
−0.92 ± 1.11
18.29 ± 1.04
55.34 ± 1.82
−9.86 ± 0.88
−2.79 ± 0.57
2.52 ± 0.94
Host lichen × sex
2.57 ± 1.16
128.99 ± 8.79
Table 3. Comparisons of GLMM models (with random factors) and GLM models (without random factors), or LME models (with random factors) and GLS models (without random factors) for each measured life-history trait to explore whether variation caused by an individual female and a year need to be taken into account.
Oviposition day had no effect on the attained pupal mass of C. lichenaria larvae (Fig. 4, Table 2). Only host lichen and sex were significant determinants of pupal masses (Fig. 4, Table 2). Larvae from eggs laid on the fourth night had a slightly but significantly shorter larval period compared to larvae eclosed from eggs laid on the first night (Fig. 5, Table 2.). In addition, males and larvae reared on R. fraxinea attained pupal stage earlier than larvae reared on P. sulcata (Table 3, Fig. 5). There was also statistically significant host lichen × sex interaction for the larval period [males 123.0 ± 1.8 and 118 ± 1.2, females 126.3 ± 1.6 and 121.0 ± 1.3 (mean days ± SE) on P. sulcata and R. fraxinea, respectively, Table 2], although this 0.3-day longer development between host lichens among females compared to males likely has no biological significance (Fig. 5).
By comparing both absolute and relative egg provisioning of ageing females, our aim was to explore whether presumed lower performance of later laid offspring is attributable to egg size per se or decreased concentration of essential nutrients in eggs. We found significant differences in the survival between the first and the fourth night laid offspring of C. lichenaria, while no differences were observed in other studied life-history traits, e.g. in pupal mass. The larvae that hatched from eggs laid on the fourth night had a higher overall mortality and were more prone to die on low quality P. sulcata compared with the larvae hatching from eggs laid on the first night and on high quality R. fraxinea. Fresh and dry weights of eggs as well as absolute amount of proteins and lipids in an egg decreased during the oviposition period, whereas no clear changes were observed in the total lipid or protein concentration, i.e. in relative provisioning. Although there was a significant quadratic contrast for egg protein concentration, overlapping standard errors reveal that no differences in egg protein concentration occurred between eggs laid during the first and fourth night (Fig. 1). Furthermore, similar egg fertility, egg hatchability, egg development time, and neonate survival without food between eggs laid during the first and fourth night also enhances our view that the quality of eggs remained otherwise equal except that only the size of eggs and larvae decreased during the oviposition period of C. lichenaria. Resources (i.e. proteins and lipids) in relation to larval size were similar for earlier laid larger and later laid smaller larvae and resulted in similar responses of earlier and later laid offspring in all traits measured during egg stage and early larval development. The relatively long inspection interval of larvae (24 h), however, may have delayed the time to find out differences in survival time of neonate larvae. All things considered, the lower survival of later laid offspring in this study species is attributable to the smaller size of later laid offspring than to reduced relative provisioning.
The increase in water concentration was not reflected on the performance of eggs or hatching success as proposed for other insect species (e.g. Geister et al., 2008). This slight increase in water concentration of eggs during the oviposition period, however, reveals that the relative provisioning may decrease when females grow old. The increase in water concentration may be attributable to the combined decrease of all nutrients in an egg. Thus, although changes in relative provisioning of single nutrients during the oviposition period were minimal, the combined impact of nutrients may also reveal a decrease in relative provisioning. It is also important to point out that these methods may have limitations to extract all the different lipids and proteins (Kawooya & Law, 1988; Stoscheck, 1990; Geister et al., 2008). Based on the previous study of Lorenz (2003), we propose, however, that the methods used are appropriate to extract most voluminous ingredients of eggs (e.g. vitellogenin and triacylglycerol) and thus qualitatively reflecting changes in nutrients during the oviposition period.
Our results are in line with other results regarding resource allocation into eggs of invertebrates. First, although relative provisioning remained constant during the oviposition period, decreased absolute provisioning showed that females allocated fewer resources to later laid eggs compared with first laid eggs. Thus, our results are consistent with a theoretical prediction of resource depletion hypotheses that decrease in egg size during the oviposition period is an unchangeable product of remaining reproductive resources (Wiklund & Karlsson, 1984). Second, in line with many previous experiments, we found that the absolute amount of resources allocated to eggs measured as egg weight, number, and absolute provisioning decreased as females aged (Begon & Parker, 1986; Bernardo, 1996; Bridges & Heppell, 1996; Jann & Ward, 1999; Fox & Czesak, 2000; McIntyre & Gooding, 2000; Fox et al., 2003; Giron & Casas, 2003; Geister et al., 2008; Gibbs et al., 2010a,b). More studies with capital breeders are needed to explore whether our results apply only to this study species or whether they reflect capital breeders in general. Furthermore, our current knowledge of resource allocation of eggs of invertebrates is mainly based on studies with insects. Experiments with other invertebrates are necessary to reveal a whole picture of egg provisioning of invertebrates (see e.g. Glazier, 1992; Bridges & Heppell, 1996; Ito, 1997).
Also, in line with a previous study, our results suggest that in capital breeders egg composition is rather constant (Diss et al., 1996) and contradict our predictions that in capital breeders relative provisioning should decrease as the female ages (Bernardo, 1996; Fox & Czesak, 2000). Diss et al. (1996) compared the egg composition of Lymantria dispar in relation to feeding history of females and found no differences in vitellin or glycin-rich protein concentrations between females from defoliated and undefoliated sites. Although Diss et al. (1996) measured egg properties during first-, centre-, and last-laid sections of egg mass, data were not shown or analysed in relation to timing of egg laying and thus it is impossible to say anything about changes in egg composition during the whole oviposition period in L. dispar. In C. lichenaria there was a trend from the fifth night onwards that egg protein content may decrease also in capital breeders (Fig. 1). However, whether this trend reflects decreased provisioning of this specific capital breeder and capital breeders in general, and how it contributes to fitness of larvae needs to be explored further.
Decreased survival of smaller neonate larvae compared with larger ones may be attributable to several properties of neonate larvae and physical and chemical characteristics of their host lichens. Lichen secondary metabolites, atranorin and chloratranorin in P. sulcata hinder the growth of neonate C. lichenaria larvae and larvae prefer to feed on Ramalina species whose lichen secondary metabolite, usnic acid has no detrimental effect on the growth or survival of larvae (Pöykkö, 2006; Pöykköet al., 2010). Reduced feeding by deterrent impact of secondary metabolites on P. sulcata may be more detrimental to smaller larvae causing the observed survival difference (Zalucki et al., 2002). Lichen species also differ in the thallus structure R. fraxinea being thicker than P. sulcata. Moreover, algal layer which lichen-feeders likely prefer (Rawlins, 1984; Backor et al., 2003) occurs on both sides of the Ramalina thallus but only on the upper side of the P. sulcata thallus. This ensures that larvae have more opportunities to find places with thin cortex and easier access to the algal layer on R. fraxinea than on P. sulcata. Finally, smaller larvae likely have smaller and less efficient mandibles and/or mandibular muscles to deal with their fodder. Thus, rather than one specific reason for lower survival, mortality of later laid C. lichenaria larvae is more probably caused by several interactive factors.
No differences were observed in pupal masses between individuals hatched from eggs laid on the first and fourth night and only a minor difference in larval developmental time with fourth night laid offspring having a slightly shorter larval period. Accordingly, a difference of 3 days in the beginning of the larval period was not reflected in the achieved pupal mass. Life cycle events in insects are largely determined by environmental effects, such as photoperiod and temperature (Tauber et al., 1986). As the larval period of C. lichenaria is relatively long-lasting from July to May in the next year, the fourth night laid offspring may have at least two stages to achieve similar body size as first night laid offspring. First, when larvae prepare to overwinter, fourth night laid offspring may continue growing longer than first night laid offspring to achieve the appropriate stage for overwintering. Second, as biomass of Lepidopteran larvae accumulates during their last instars, in case of C. lichenaria after overwintering, fourth night laid offspring (if still smaller after overwintering), may need to increase their growth according to developing season to attain optimal adult size more than the first night laid offspring. Differences between years, especially in the length of larval period, may be attributable, for example, to yearly differences in the quality of host lichens. However, in both years, fourth night laid offspring had a trend for a shorter developmental period (Fig. 5).
As egg size had a significant impact on the survival of larvae, it is justified to ask why egg size decreases as female ages. Theoretical models predict that egg number and size should decrease with female age if the mortality risk increases simultaneously (Begon & Parker, 1986; Stearns, 1992; Roff, 2002; Sakai & Harada, 2004). Although we are unaware of any specific mortality factors for adult C. lichenaria moths, there evidently exist plenty of invertebrate and avian predators feeding on tree trunks and branches. Thus, concentrating reproduction to early stages, females of C. lichenaria ensure that at least most of their reproductive potential can be realised. Decrease in the size of eggs may be attributable also to the fact that after the egg load of the first night has been oviposited, 1 day may not be enough for developing eggs to attain a similar size as in the first night. Instead, to maximise total reproductive effort, it may be better to mature more but smaller eggs than fewer larger eggs. Decrease in egg size and number is also predicted if there are benefits of producing sequential cohorts (Sakai & Harada, 2004). For insect larvae, a head start of a few days in development may considerably increase competitive advantage of early laid offspring (Averill & Prokopy, 1987; Messina, 1991; Kivelä & Välimäki, 2008). Accordingly, females of C. lichenaria may lay sequential cohorts with decreasing size of eggs to decrease potential competition among larvae.
In C. lichenaria almost half of the eggs are ready for oviposition at the time of eclosion and the rest mature during the adult stage (Pöykkö, 2009) but the composition of eggs remains relatively stable (Fig. 1). This intermediate ovigeny index is, in general, associated with several other life-history traits, such as intermediate lifespan, female flight ability, and relatively high diet breadth and availability (Jervis et al., 2005, 2007). At maturation, C. lichenaria females have their abdomen full of eggs and developing oocytes, which considerably restrict their flight ability. Cleorodes lichenaria, although capable of using multiple lichens as hosts, prefers Ramalina species that are among the most common lichens at the habitat of origin serving as oviposition substrates (Pöykkö, 2006). Egg load is known to affect the readiness for oviposition. Insects with high egg loads are more willing to lay eggs than insects with low egg loads (Jallow & Zalucki, 1998). Accordingly, towards the end of the oviposition period with a decreased abdomen load and increased flight ability, females may have more time to carefully search for suitable oviposition sites as they have fewer eggs to deposit. Later laid offspring would benefit relatively more of being laid on high quality hosts than early laid larvae. As egg composition did not change during the oviposition period, a decrease in egg size may be one way to increase total reproductive effort if females are able increase their selectiveness and reduce intraspecific offspring competition.
In conclusion, egg composition remained relatively constant during the oviposition period of C. lichenaria. The decreased size of later laid eggs, however, incurred a mortality cost especially on low quality P. sulcata and is likely attributable to physical characteristics of larvae and chemical and physical properties of their host lichens. Differences in other measured life-history traits, such as egg fertility and hatchability, neonate survival or pupal mass between offspring laid on the first and fourth night were not, however, observed. The decrease in egg size supports the resource-depletion hypothesis according to a reduced total amount of resources allocated to eggs, which may result from physiological attributes as well as selective forces favouring maximal reproductive output. As a whole, a decrease in egg size and number may thus reflect trade-offs between multiple competing life-history traits.
We are grateful to Eija Hurme, Laura Härkönen, Arja Kaitala, Sami Kivelä, Maria Tuomaala, and Panu Välimäki for their invaluable comments on the manuscript and Sami Kivelä for statistical advice. We also thank Marja-Liisa Martimo-Halmetoja and Minna Orreveteläinen for help in the protein and lipid analyses. This study was financially supported by Finnish Cultural Foundation and Finnish Academy (project no. 120964)