Oecologia (1990) 85 : 101-107 Oecologia 9 Springer-Verlag 1990 The adaptive significance of an environmentally-cued developmental switch in an anuran tadpole David Pfennig* Department of Zoology, University of Texas, Austin, TX 78712, USA ReceivedJanuary 2, 1990 / Accepted in revised form June 25, 1990 Summary. This study investigated the proximate basis of bimodally-distributed, environmentally-induced vari- ation that occurs in natural populations of spade:foot toad tadpoles (Scaphiopus multiplicatus). Most individ- uals in most populations occur as a small, slowly-devel- oping omnivore morph. In some of these same popula- tions, a varying number of individuals occur as a large, rapidly-developing carnivore morph (Pfennig 1989). Censuses of 37 different natural ponds revealed that the frequency of the faster-developing carnivore morph cor- related significantly positively with fairy shrimp density (their chief prey) and pond drying rate. By simultaneous- ly varying two diet components and pond drying regime in artificial pools I found that only fairy shrimp density significantly affected the proportion of carnivores. Sepa- rate experiments established that the extent to which tadpoles developed the carnivore morphology correlated with shrimp density, and that morph determination de- pended on the ingestion of shrimp, not simply their pres- ence. If a critical number of shrimp were ingested, the tadpole developed into a carnivore; if not, the tadpole developed by default into an omnivore. Thus a single cue shrimp ingestion - triggers alternative ontogenetic trajectories. Using shrimp density to induce morph dif- ferentiation enables tadpoles to respond to their environ- ment adaptively as shrimp are most abundant in highly ephemeral ponds, where the faster developing carnivores are favored. Key words: Causes of polyphenism - Adaptive plasticity Trophic polymorphism - Developmental strategies Phenotypic plasticity A fundamental issue in the study of phenotypic plasticity is resolving the extent to which the plastic response is due to selection and the extent to which it is due to * Present address and address for offprint requests: Department of Zoology, Arizona State University, Tempe, AZ 85287, USA nonadaptive processes. Phenotypic plasticity often oc- curs among organisms in variable environments (Sha- piro 1976), which suggests to some that it provides an optimal response to environmental heterogeneity (Brad- shaw 1965; Levins 1968; Lloyd 1984; Lively 1986a). This conclusion is reinforced by classical population ge- netic models, which show that alternative forms are fa- vored in spatially varying environments (Levene 1953). However, not all plasticity need be adaptive. Any devel- opmental system that is not buffered against changes in the environment to which critical chemical reactions in development are sensitive will be phenotypically plas- tic (Stearns 1982). To understand the evolutionary signif- icance of phenotypic plasticity, one must discern the ex- tent to which plasticity has evolved as an adaptation and the extent to which it is an unavoidable side effect of environmental heterogeneity. Among the most striking examples of presumably adaptive phenotypic plasticity are all-or-nothing re- sponses, which are induced by an environmentally-me- diated developmental switch (polyphenism; sensu Mayr 1963). Levins (1968) suggested that natural selection should favor such developmental switches in "a fine- grained environment where the range of the environment exceeds the tolerance of the individual phenotype." En- vironmentally-induced developmental switches have been found in bryozoans (Harvell 1984), Daphnia (Grant and Bayly 1981), gastropods (Crowl and Covich 1990), barnacles (Lively 1986b), locusts (Kennedy 1956) cater- pillars (Bernays 1986; Greene 1989) and various other insects (reviewed in Roff 1986), fish (Kornfield and Tay- lor 1983; Meyer 1987), salamanders (Collins and Cheek 1983), and anurans (Pfennig 1989). Theoretical studies of developmental switches have indicated that one environmental cue may affect the de- velopment of two different morphs. Lively (1986a) re- cently used a game theoretic model to illustrate that in a two patch system, diphenism may persist even though the cue for one type of patch is poor, if the cue for the other type of patch leads to the "correct" developmental choice greater than 50% of the time. The