- Top of page
- Materials and methods
Mothers sometimes use epigenetic mechanisms to manipulate the phenotype of their progeny in ways that improve their fitness under prevailing or anticipated environmental conditions (Fox & Mousseau, 1998). It is now generally accepted that such ‘maternal effects’ are transgenerational signals subject to selection that serve to canalise the development of phenotypic plasticity, with potentially life-long impacts on progeny life histories (Mousseau et al., 2009). A large body of medical research has examined the role of maternal effects in human child development, focusing mostly on the negative consequences of drug use and various maternal pathologies. However, natural maternal modifications of progeny phenotype may be adaptive when the optimal phenotype varies either spatially or temporally within the reproductive lifespan of the mother (Mousseau & Dingle, 1991). Consequently, empirical studies of maternal effects have examined the influence of various aspects of the maternal environment such as food availability (Bashey, 2006), food quality (Takakura, 2004; Gonzalez-Teuber et al., 2008), availability of oviposition sites (Gottlieb et al., 2011), intensity of intra-specific competition (Marshall et al., 2006), and temperature (Huestis & Marshall, 2006; Scharf et al., 2010). These have been characterised as ‘detection-based’ maternal effects (Shea et al., 2011). In contrast, ‘selection-based’ maternal effects—those that occur independent of the maternal environment, such as changes in progeny phenotype as a function of maternal age or birth order—have received far less empirical attention. For example, immature fresh water rotifers, Brachionus calyciflorus Pallas, develop longer defensive spines when their mothers are exposed to predation risk, but this phenotype is also expressed in offspring born late in the mother's life, regardless of maternal environment (Schröder & Gilbert, 2009). Later born progeny of the waterflea Daphnia galeata Sars develop faster and become larger adults, illustrating a fixed birth order effect on offspring size (Sawinska, 2004).
Many aphidophagous coccinellid beetles depend on sporadic and ephemeral aphid outbreaks to supply the resources critical for both reproduction and subsequent offspring development (Borges et al., 2006). The larvae of such species face a challenging, but largely predictable, trajectory of food availability. Aphid populations typically begin with low numbers (the initiation stage) and follow a trajectory of rapidly accelerating abundance (the exponential growth stage) followed inevitably by a precipitous decline in numbers (the collapse stage; Smith, 1966; Michaud & Harwood, 2012). The optimum time for coccinellid reproduction, the ‘oviposition window’, is early in the exponential growth phase, as both intra- and inter-specific competition for dwindling resources intensifies during the collapse phase (Kindlmann & Dixon, 1993, 2010). It follows that progeny produced sequentially over this period do not experience equivalent conditions; those produced later face increasingly difficult conditions as the aphid bloom matures, such that selection-based maternal effects could evolve to improve maternal fitness via adaptive adjustments to offspring phenotype.
Variation in egg size has the potential to affect both developmental rate and final offspring size in many animals; in general, progeny hatching from larger eggs have higher survival, faster development and achieve a larger adult size than those hatching from smaller eggs (Fox, 1994; Bernardo, 1996; Fox & Mousseau, 1998). Although most arthropods tend to decrease egg size with advancing maternal age (Fox & Czesak, 2000), periods of an increase in egg size have been reported in the aphidophagous species Coleomegilla maculata De Geer and Hippodamia convergens Guerin-Meneville (Coleoptera: Coccinellidae) (Vargas et al., 2012a,b), and interpreted as a maternal effect that may enhance offspring fitness over the course of the female's reproductive cycle. However, egg size is only one measure of offspring quality and other, more cryptic, maternal signals may also influence progeny fitness (Bernardo, 1996). For example, offspring of the soil mite Sancassania berlesei (Michael) produced by older mothers mature to larger body sizes, even when the effects of egg size are controlled, better enabling them to compete for limiting resources with earlier born siblings that are developmentally advanced (Benton et al., 2008). Cryptic maternal effects, for example subtle changes in development or behaviour, have also been reported in birds (Groothuis et al., 2005) and mammals (Dloniak et al., 2006). Recently, Vargas et al. (2012c), demonstrated age-specific maternal effects in H. convergens that appeared unrelated to changes in egg mass, but affected progeny developmental rate, the relative duration of particular developmental stages (fourth instar and pupa) and final adult size.
Although C. maculata is an exceptionally polyphagous lady beetle (Hodek, 1996), North American populations commonly rely on aphid resources for reproduction (Wright & Laing, 1980; Michaud & Jyoti, 2008). Coccinellids that exploit cereal aphids on the High Plains of the USA normally produce two generations per year, one in the fall and the other in spring, with the potential for additional generations if summer weather is cool enough to permit additional aphid outbreaks (Michaud & Qureshi, 2006). As summer is normally passed in reproductive diapause due to food shortage, and winter in hibernation, cohorts of beetles normally complete development in one aphid season and reproduce in the next, with most females partitioning their reproductive effort over a single aphid population cycle.
The objective of the present study was to test for age-specific maternal effects (e.g. Harvey, 1977) and the possible interactions of these with larval environment (e.g. Ng, 1988). We hypothesised that age-specific maternal effects would result in progeny phenotype alterations consistent with optimisation of maternal fitness on a ‘boom and bust’ cycle of prey availability. This was tested by rearing a series of four larval cohorts, obtained from the same mothers at four different points in their reproductive cycle, and recording their developmental data. Second, we hypothesised that certain age-specific maternal effects might improve progeny survival, specifically under harsh conditions such as food shortage; this was tested in the same experiment by dividing each of the four larval cohorts obtained from each female into two groups, one reared with abundant food and the other with very limited food access.
- Top of page
- Materials and methods
Among the 53 pairs of beetles established in the experiment, 42 females completed 36 oviposition days and were included in the analysis. Changes in daily fecundity and fertility as a function of female age were both described with high levels of significance by second order regressions (F = 136.62; d.f. = 2,33; P < 0.001, R2 = 0.89 and F = 52.15 P < 0.001, R2 = 0.76, respectively) (Fig. 1a,b). Changes in egg mass best fit a third order regression (F = 295.05; d.f. = 3,32; P < 0.001, R2 = 0.96) (Fig. 1c). Females increased egg number and egg mass up to around the 30th oviposition day (by 10% and 35%, respectively), whereas fertility peaked around day 18.
Figure 1. Changes in mean daily fecundity (a), egg fertility (b), and egg mass (c) of 42 Coleomegilla maculata females over the course of 36 oviposition days. Pairs of beetles were held together for 28 days, whereupon males were removed. Females were fed eggs of Ephestia kuehniella ad libitum during both development and reproduction.
Download figure to PowerPoint
A total of 1493 larvae were reared in the experiment and a larger proportion of those receiving ad libitum food survived compared with those receiving 30 min access daily (97.3% vs. 88.0%; F = 103.95; d.f. = 1,171.3; P < 0.001). There was also a significant main effect of oviposition day on larval survival (F = 3.94; d.f. = 3,156.7; P = 0.009), but the interaction between oviposition day and treatment was not significant (F = 0.47; d.f. = 3,244.8; P = 0.360) (Fig. 2). In both treatments, larvae derived from the first day of oviposition had a lower survival than those derived from later oviposition days (LSM test, α = 0.05), although survival of those derived from the 36th oviposition day of 30-min females were intermediate and not significantly different from other oviposition days in this treatment. Sex ratios were not significantly different from 1 : 1, and were not affected by feeding regime (F = 2.08; d.f. = 1,118.3; P = 0.151) or oviposition day (F = 1.79; d.f. = 3,232.1; P = 0.149).
Figure 2. Mean (± SE) survival of Coleomegilla maculata progeny produced from the 1st, 12th, 24th, and 36th oviposition days of their mothers (n = 42) and reared under two periods of daily access to food (eggs of Ephestia kuehniella), 30 min versus ad libitum. Larvae fed ad libitum had a higher survival (P < 0.001) than those reared with 30 min daily access; different letters denote significant differences (LSM, α = 0.05) among oviposition days within treatments.
Download figure to PowerPoint
All viable eggs hatched in 3 days, regardless of the cohort. There was a significant main effect of treatment on the total duration of larval development (F = 2,849.01; d.f. = 1,295; P < 0.001), but the main effect of oviposition day was not significant (F = 0.54; d.f. = 3,552.2; P = 0.652), and there was no significant interaction between treatment and oviposition day (F = 0.65; d.f. = 3,514.3; P = 0.582). The ad libitum treatment generated faster larval development (mean ± SE = 15.2 ± 0.03 days) than the 30-min feeding treatment (26.0 ± 0.02 days). The main effect of treatment was not significant for pupation time (F = 0.47; d.f. = 1,480.2; P = 0.494), whereas the main effect for oviposition day was (F = 6.78; d.f. = 3,664; P < 0.001), and the interaction between these independent variables was not significant (F = 0.69; d.f. = 3,636.9; P = 0.559). Regardless of feeding treatment, progeny produced on the 36th oviposition day spent less time in the pupal stage than those produced on either the 1st or 12th oviposition day.
Both feeding treatment and oviposition day affected the number of larval stadia. A majority of individuals pupated after four larval instars in both feeding treatments (Fig. 3), but the percentage was greater in the ad libitum feeding treatment than in the restricted diet treatment (90.3% vs. 66.9%; F = 74.11; d.f. = 1,148.6; P < 0.001). Oviposition day affected the percentage of individuals pupating after four instars (F = 2.91; d.f. = 3, 283; P = 0.033); the percentage was higher on the two early days than on the two later days in the ad libitum feeding treatment, whereas it was higher on the first oviposition day compared with the 24th day on the restricted diet (LSM test, α = 0.05). There was no significant interaction between treatment and oviposition day for the percentage of individuals pupating after four instars (F = 1.93; d.f. = 3,246.8; P = 0.125). A larger percentage of individuals pupated after only three instars on the restricted diet than in the ad libitum treatment (28.8% vs. 8.6%; F = 56.35; d.f. = 1,152.5; P < 0.001) and there was a significant effect of oviposition day in both treatments (F = 5.04; d.f. = 3,239.1; P = 0.002). In both feeding treatments, the percentage of third instar pupations was higher on later oviposition days than on earlier ones (LSM test, α = 0.05) and the interaction between treatment and oviposition day was not significant (F = 1.07; d.f. = 3,247.7; P = 0.361). A small fraction of individuals underwent a supernumary fifth larval instar and the percentage doing so was higher in the restricted diet treatment than in the ad libitum treatment (4.3% vs. 1.2%; F = 9.54; d.f. = 1,111.4; P = 0.002) but the effect of oviposition day was not significant (F = 0.76; d.f. = 3,198; P = 0.515), nor was the two-way interaction term (F = 1.59; d.f. = 3,232.2; P = 0.191).
Figure 3. Percentages of Coleomegilla maculata larvae reared from four different maternal cohorts (oviposition days) that pupated after three (open segments), four (shaded segments), or five (solid segments) stadia when reared with access to food (eggs of E. kuehniella) for 30 min daily (a) or ad libitum (b). Treatment contrasts were the same across all oviposition days: three and five instars, A > B; four instars A < B. Within feeding treatments, column segments bearing different letters were significantly different among oviposition days (LSM, α = 0.05). The percentage of larvae pupating after five instars did not vary significantly among oviposition days in either treatment. Numbers indicate sample sizes for each stacked column.
Download figure to PowerPoint
A three-way model considering ‘treatment’, ‘oviposition day’, and ‘gender’ as independent variables was significant overall for adult weight with the significant main effects of treatment and gender, but not oviposition day, and with significant treatment × gender interaction (Table 1), so data were pooled across oviposition days for further analysis. Both males and females were almost three times as heavy when reared on the ad libitum diet as when restricted to 30 min daily food access (males: 3.03 ± 0.02 vs. 1.12 ± 0.01 mg; females: 3.36 ± 0.03 vs. 1.23 ± 0.01; P < 0.001 in both cases).
Table 1. Three-way anova showing the effects of larval feeding treatment (30 min daily access vs. ad libitum), maternal oviposition day (1st, 12th, 24th or 36th), and gender (male/female) on Coleomegilla maculata adult dry mass at emergence
|Source of variation||F||d.f.||P|
|Corrected model||683.23||15||< 0.001|
|Treatment × oviposition day||0.19||3||0.249|
|Treatment × gender||29.29||1||< 0.001|
|Oviposition day × gender||0.61||3||0.612|
|Treatment × oviposition day × gender||0.35||3||0.792|
The number of larval stadia did not affect the final adult body weight of either males or females in either feeding treatment (restricted diet males: F2,277 = 1.57, P = 0.209; restricted diet females: F2,322 = 0.35, P = 0.707; ad libitum males: F2,391 = 2.26, P = 0.106; and ad libitum females: F2,381 = 0.14, P = 0.868). However, the total larval development time was significantly longer for larvae on a restricted diet that pupated after five larval instars compared with either three or four (mean ± SE = 32.2 ± 0.9 vs. 28.8 ± 0.4 and 29.4 ± 0.2, respectively; F2,602 = 5.14, P = 0.005), whereas the number of stadia did not affect developmental time for beetles fed ad libitum (F2,777 = 2.56, P = 0.078).
- Top of page
- Materials and methods
Both initial hypotheses were supported; age-specific maternal effects were evident in progeny development and some effects interacted with larval feeding treatment. Consistent with previous observations on C. maculata (Vargas et al., 2012a), daily fecundity gradually increased for the first 25–30 days and there was a maternal effect on egg size, the fresh mass of eggs increasing over the course of the first 20 days of female oviposition (Fig. 1). The declining trajectory of egg fertility late in the experiment is probably a function of normal female ageing, although the removal of males on day 28 and a lack of male seminal contributions thereafter may have contributed. It should also be noted that all effects of maternal age, per se, include possible effects attributable to oviposition history (number of eggs previously laid) as both increase in parallel over the course of the female reproductive life.
Larval survival was reduced by the restricted food supply (Fig. 2), but not so greatly considering the extreme food deprivation imposed in this treatment. More importantly, larvae from the first cohort had a lower survival than later cohorts, independent of treatment. In general, egg size tends to decline with female age in insects (Fox & Czesak, 2000) and females are normally expected to produce their ‘best’ progeny early in life, if only because the risk of mortality discounts the value of future reproductive effort compared with present (e.g. Mousseau & Dingle, 1991; Tschinkel, 1993). Indeed, Singh and Omkar (2009) found that parental age was positively correlated with developmental time in Cheilomenes sexmaculata (F.) and negatively correlated with survival and adult body weight. However, there are ecological reasons why female C. maculata may benefit by increasing egg size (and offspring quality) as a function of oviposition sequence. Species that exploit aphid outbreaks begin reproduction on an abundant food supply which permits the survival of small, low-quality offspring, but increasingly adverse conditions develop as the outbreak matures, favouring the production of larger, more competitive ones (Vargas et al., 2012a,b). Although the food deprivation treatment extended larval development, there was no main effect of oviposition day on total larval development time. In contrast, H. convergens larvae from later cohorts develop faster than those from earlier ones, largely as a result of faster egg hatching and shorter pupation times (Vargas et al., 2012c). Notably, the final cohort of C. maculata in this study did have shorter pupation times than earlier ones, independent of feeding treatment. The pupal stage is especially vulnerable to cannibalism and intra-guild predation, so a reduction in pupation time is advantageous for later cohorts if mothers produce them as environmental conditions are deteriorating and survival hazards are increasing.
Variation in the number of larval stadia is common in certain insect groups such as Lepidoptera (reviewed by Esperk et al., 2007a). However, coccinellids almost invariably have four instars and this has been assumed to represent a phylogenetic constraint, as fewer instars would presumably afford faster development on ephemeral resources (Dixon, 2000; Nedved & Honek, 2012). Supernumary fifth instars have been reported in Callicaria superba (Mulsant) (Iwata, 1932), Chilocorus nigritus (F.) (Chazeau, 1981), Harmonia axyridis Pallas (Labrie et al. (2006), and C. maculata (Warren & Tadic, 1967), but most cases involve only a small fraction of individuals. Reports of coccinellid pupation after only three instars are even fewer, but examples include Hyperaspis campestris (Herbst) (McKenzie, 1932) and Coccinella undecimnotata L. (Iablokoff-Khnzorian, 1982). In Nephaspis oculatus (Blatchley), a significant fraction of individuals can be induced to pupate after the third instar by subjecting them to high temperature stress (≥ 29 °C) (Ren et al., 2002). In the present study, food deprivation decreased the percentage of individuals undergoing ‘normal’ development with four stadia and increased the percentage undergoing either three or five stadia, most notably the former. Alterations in number of stadia in response to both food and temperature stress are well known in the Lepidoptera. A nitrogen-deficient diet will induce the soybean looper, Pseudoplusia includens (Walker), to extend development and for some larvae to undergo supernumary stadia (Wier & Boethel, 1995). An increased number of stadia can be induced in some larvae of Spodoptera exigua (Hubner) by extremes of temperature and by feeding on a low-quality food plant, Gossypium hirsutum L. (Ali & Gaylor, 1992). Similarly, development on low-quality food plant results in supernumary instars in the tortricid Acleris minuta (Robinson) (Weatherby & Hart, 1986). In coccinellids, maternally-mediated changes in duration of life stages, developmental rate, and number of moults are most likely cued by either genomic imprinting or allohormonal signals in the egg and appear independent of changes in egg size, even although they are sensitive to female condition (Vargas et al., 2012c).
In the context of life history theory, it is often assumed there exists a tradeoff between developmental time and size at maturity (Roff, 1992; Stearns, 1992); attainment of a large body size permits higher fecundity (Honěk, 1993) but requires a longer period of growth to achieve, thus incurring a cost in terms of delayed development. For example, in the tiger moth Eilema depressum (Esper), directly developing larvae pupate at smaller sizes after fewer instars compared with larvae that overwinter (Poykko & Hyvarinen, 2012). However, other species of Lepidoptera pupate upon reaching a critical weight, regardless of instar, such that adult size is unaffected by the number of stadia (Davidowitz et al., 2003; Kingsolver, 2007). It has been argued that the optimal number of instars is that which maximises food acquisition while minimising the costs of moulting (Hutchinson et al., 1997). In general, the addition of stadia is a strategy for attaining greater (or critical) body size, whereas the subtraction of stadia is a means of speeding development or, when resources are limiting, saving the energetic cost of a moult. In some cases, individuals eclosing from smaller eggs are more likely to undergo additional instars (Leonard, 1970; Frago et al., 2009). In insects sexually dimorphic for instar number, females tend to have more instars than males (e.g. Berthiaume et al., 2007), and sexual size dimorphism is often pronounced in these species (Esperk & Tammaru, 2006; Esperk et al., 2007b). For example, females of the grasshopper Heteracris littoralis (Rambur) (Acrididae) require a longer developmental period and pass through 6–7 instars, as opposed to only 5 for males, in order to achieve a significantly larger body size at maturity (Singh & Chaudhary, 1999). Etile and Despland (2008) argued that variation in instar number provides insects flexibility to compensate for poor growing conditions. We interpret the present results (Fig. 3) as representing alternative developmental strategies for dealing with food deprivation that can be characterised as optimistic and pessimistic, respectively. The optimistic strategy is the addition of a fifth instar that will permit attainment of a greater body size through additional feeding time, providing food availability improves towards the end of development. For example, Chen and Ruberson (2008) showed that 2 days of starvation for beet armyworm larvae in the first instar caused a greater proportion to undergo a supernumary sixth instar and that such larvae ultimately achieved comparable pupal weights, albeit after a longer developmental time. In contrast, the pessimistic strategy is to pupate after only three instars and save the energy and risk associated with additional moults, the best option if food conditions do not improve (as in the present experiment). Clearly, pessimists outnumbered optimists in our experiment and were rewarded in the food deprivation treatment; whereas optimists paid a time cost for adding an instar, pessimists that subtracted an instar did not pay a cost in reduced body size.
The maternal effect of oviposition sequence on number of stadia was similar in both treatments; offspring produced later in the life of their mother were more likely to be pessimistic and subtract an instar, consistent with a scenario of increasing hardship for progeny developing later in aphid outbreaks. On the High Plains, females of aphidophagaous species normally encounter only a single reproductive opportunity (aphid outbreak) in either spring or fall (Michaud & Qureshi, 2006) and do not oviposit in the same outbreak that supports their development. Previously (Vargas et al., 2012a,b), we argued that a fixed schedule of increasing offspring quality (and/or developmental rate) as a function of oviposition sequence would maximise maternal fitness when the progeny develop on aphids, populations of which can shift from exponentially increasing to precipitously declining within a period as short as 2–3 weeks. In H. convergens, females accomplish this by shortening progeny developmental time in later clutches, in particular by reducing the duration of the most vulnerable stages, egg, and pupa (Vargas et al., 2012c). In C. maculata, maternal effects are manifest as a developmental polymorphism in which mothers increase the tendency of later progeny to truncate their life history and pupate after only three instars. Although larvae forgoing a fourth instar did not achieve faster development, the normal pattern of change in food supply (abundant to scarce) was not reflected in either feeding treatment. Thus it remains conceivable that the three instar phenotype might yield fitness benefits when development begins on abundant food and ends in deficit. It is also notable that a combination of all three phenotypes was produced in all cohorts (there were even five cases of larvae with six instars, excluded from analysis), suggesting females employ a baseline strategy of producing multiple developmental phenotypes in their progeny, but use maternal effects to modify the relative proportions of each in sequential clutches. To our knowledge, this is the first demonstration of a maternal effect on developmental polymorphism in a holometabolous insect.