REACTION NORMS FOR AGE AND SIZE AT MATURITY IN LASIOMMATA BUTTERFLIES: PREDICTIONS AND TESTS

In recent years there has been a strong increase in the interest in animal life-history plasticity (e.g., Stearns and Koella 1986; Gebhardt and Stearns 1988; West-Eberhard 1989; Kindlmann and Dixon 1992; Newman 1992; Hensley 1993; Reznick and Yang 1993; Via 1993; Bernardo 1994), which has brought into focus the question of how to determine when variation in reaction norms is adaptive (beneficial, and possibly maintained by selection), and when plasticity can even be said to be an adaptation in the strict sense, that is, the origin of the adaptation can be linked to the same selective advantage as the current function. Not all plasticity is beneficial, in the sense that it increases fitness compared with a different reaction norm (Newman 1992). Reaction norms with nonzero slopes can occur because of "constraints" (factors that are external to the system under investigation, including adaptations shared with many populations or species), rather than because of adaptation to the local environment. One example is when a poikilotherm animal speeds up development at high temperatures. This may be incidental in a given case, but it may serve a function if the animal is, for example, an amphibian inhabiting a temporary pond, and higher temperatures signal that the pond will dry up faster (Newman 1992). If these amphibians respond differently to temperature than related species inhabiting more stationary waters, there may be evidence for adaptation. That a given form of plasticity is likely to be an adaptation can be demonstrated in basically two ways: either by showing experimentally that the shape of reaction norms correspond closely to a priori predictions based on optimality criteria, or by using comparative methods to show that differences in reaction norms between species or other categories of individuals (reflecting evolutionary modifications) correspond to predictions (see Gotthard and Nylin 1995). There are few examples in the literature of such attempts at predicting the shape of plasticity. Exceptions include Stearns and Koella's (1986) models predicting reaction norms for age and size at maturity, and Ford and Seigel's (1989) and Reznick and Yang's (1993) predictions regarding how allocation patterns to reproduction should respond to varying food levels in the checkered garter snake, Thamnopsis marcianus, and in the guppy, Poecilia reticulata, respectively. In the present study, we used a combination of the two techniques described above to demonstrate that life-history plasticity observed in butterflies of the genus Lasiommata (Nymphalidae: Satyrinae) may be adaptations to seasonal change in environmental conditions. We focused on effects of photoperiod on the life history, because seasonal constraints act strongly on most insects, and photoperiod is the main cue used by insects to detect the progress of the season (Danilevskii 1965; Beck 1980). Typically a large proportion of the year is unfavorable for insect growth and reproduction, and this season can be survived only in diapause, in a speciesspecific developmental stage or stages. Each individual must complete development up to the diapausing stage before the onset of the unfavorable season, independently of when in the season it started to grow and independently of weather conditions (Reavey and Lawton 1991). This necessity must form a strong selection pressure in favor of life histories promoting this outcome and, hence, some predictions are possible regarding what should constitute the optimal lifehistory response to seasonal cues (Nylin 1994). It has been observed that insects may shorten developmental times in day lengths indicating progressively later dates in the season, in crickets (Masaki 1978) and in butterflies (Nylin et al. 1989, 1995; Nylin 1992). A possible interpretation of such patterns is that the alternative "solution" (i.e., rapid development up to the hibernating stage regardless of time of the season, followed by a waiting period of variable length) may not be feasible in these insects, because rapid growth and development is too costly, or because the hibernating stage is less safe than earlier stages, or both. There are two ways that a butterfly can have a short developmental time: either by pupating at a low weight or by growing fast to the same weight. Life-history theory assumes that there must be some cost associated with a short developmental time, because otherwise organisms should have infinitely short generation times. Small adult size is the most commonly assumed cost (e.g., Pianka 1970; Roff 1983). We have found, however, that short developmental times in butterflies seem to be associated more strongly with high growth rates than with low pupal weight and small adult size (Nylin et al. 1989, 1993, 1995; Nylin 1992, 1994; Wiklund et al. 1991). A theoretical model of this relationship (Abrams et al. 1996) suggests that an organism should, if possible, shorten its developmental time when the time left to the optimal maturation date is shorter and that it should do so more by increasing growth rates than by maturing at a small size, unless the cost of high growth rates (e.g., Gotthard et al. 1994) increases faster than linearly. Lasiommata petropolitana and Lasiommata maera are