Size-mediated Performance of a Generalist Herbivore Feeding on Mixed Diets

Mixed diets can allow generalist insect herbivores to obtain nutritionally balanced resources or dilute toxins from specific foods, but also present the generalist with greater challenges in decision-making and require a greater ability to detoxify a wide range of plant defensive compounds. Young and small generalist larvae can have different nutritional requirements, ability to detoxify compounds, and mobility compared to older and larger larvae. In this field study, I asked how larval size affected performance of the woolly bear (Platyprepia virginalis), a generalist herbivore, on a uniform diet of bush lupine (Lupinus arboreus) compared to a mixed diet including bush lupine. Large larvae had greater survival on the mixed diet treatment compared to the lupine-only diet, but survival of small larvae did not vary with diet. Larval size also influenced growth on each diet, but this effect varied with year. In 1997, large larvae had higher growth on a lupine-only diet compared to a mixed diet, whereas small larvae had equivalent growth on both diets. In 1998, larvae of each size did not differ in their response to diet treatments. In the field, large larvae apparently eat a more diverse diet than small larvae, which contrasts with the growth result for 1997. This suggests that factors other than growth, such as parasitism or predation, might influence choice of host plants. RESUMEN Una dieta mixta puede permitir a los herbı́voros obtener recursos nutritivamente balanceados o diluir toxinas de alimentos especı́ficos, pero también presentan más retos para el herbı́voro al tomar decisiones y requieren de una mayor capacidad para neutralizar una mayor cantidad de compuestos defensivos de las plantas. Larvas generalistas jóvenes y pequeñas pueden tener diferentes requerimientos nutricionales, capacidades para neutralizar compuestos, y movilidad en comparación con larvas más viejas y grandes. En este estudio de campo, investigué cómo el tamaño de la larva afecta la función de la oruga Platyprepia virginalis, un herbı́voro generalista, con una dieta uniforme de lupino arbusto, Lupinus arboreus, comparada con una dieta mixta que incluye lupino. Larvas grandes sobrevivieron mejor con la dieta mixta que la dieta de solamente lupino, aunque la supervivencia de larvas pequeñas no varió con dieta. El tamaño de la larva también influyó el crecimiento con cada dieta, pero esto dependió en el año. En 1997, larvas grandes tuvieron más crecimiento con la dieta de sólo lupino comparada con la dieta mixta, mientras que la dieta no afectó el crecimiento de larvas pequeñas. En 1998, la dieta no tuvo efecto sobre las larvas de ninguno de los tamaños. Bajo condiciones naturales parece que las larvas grandes tienen una dieta más diversa que las larvas pequeñas, lo que difiere con los resultados de crecimiento de 1997. Esto sugiere que existen otros factores aparte del crecimiento, tales como el parasitismo o la depredación, que pueden influir en la elección de la planta hospedera. One might expect an organism with a broad diet to be at a selective advantage to one with a narrow diet simply due to increased resource availability (Futuyma and Moreno, 1988). For generalist herbivores, consuming a mixed diet rather than a uniform diet also can provide more balanced nutrition (Pulliam, 1975; Westoby, 1978; Rapport, 1980) or a means for diluting toxins from specific plant species (Freeland and Janzen, 1974). For an insect, being a generalist also can contribute to escape from predators or parasites by having a less predictable odor (De Moraes et al., 1998) or location. Despite the benefits of a broad diet, most insect herbivores are specialists that use only a few species of host plants for feeding and oviposition (Bernays and Graham, 1988; Jaenike, 1990). The relative scarcity of generalist species suggests that the costs of being a generalist might outweigh the potential benefits. Some 190 vol. 49, no. 2 The Southwestern Naturalist generalists incur metabolic costs associated with switching foods (Schoonhoven and Meerman, 1978; Scriber, 1979). Generalists can lack the cues that specialists use to find food and take longer to locate food, be less decisive once feeding, switch plants more frequently, or incur greater ecological risks (Bernays, 1999a, 1999b; Bernays and Funk, 1999; Bernays, 2001). Although there are many proposed benefits of a mixed diet rather than uniform diet, no overall pattern emerges from studies. Orthopterans benefit from a mixed diet compared to a uniform diet (reviewed in Bernays and Minkenberg, 1997), as do many non-insect herbivores (Pennings et al., 1993). However, other insect taxa, such as lepidopterans and hemipterans, generally perform as well or better with a uniform diet of a preferred food compared to a mixed diet (Stoyenoff et al., 1994; Bernays and Minkenberg, 1997; Ballabeni and Rahier, 2000; Singer, 2001). Larval age can affect the costs and benefits of being a generalist. Larval age can influence performance when eating mixtures of plant defensive compounds (Yang et al., 1996) or performance on previously damaged hosts (Krause and Raffa, 1995). In several lepidopteran larvae, the activity of mixed-function oxidases increases with age (reviewed in Brattsten, 1979). In gypsy moths, another broad generalist, midgut pH increases with larval age, providing greater ability to dissociate tanninprotein complexes (Schultz and Lechowicz, 1986). Thus, older larvae might have a greater ability to cope with a mixed diet due to increased ability to detoxify secondary compounds. Many microlepidopteran larvae change feeding habit as they grow, so they can incur both costs and benefits of a wider range of environmental and nutritional conditions (Gaston et al., 1991). Larger larvae also can be more mobile and encounter a greater range of potential host species (Scriber and Slansky, 1981). Fox and Morrow (1981) pointed out that even generalists at the species level are not always generalists at the population or individual level. It is possible that even within an individual, costs and benefits of being a generalist might vary with the age or size of the larvae. The woolly bear Platyprepia virginalis (Lepidoptera: Arctiidae) is a suitable species to study how age affects performance on mixed diets. This species is a broad generalist known to feed on a wide diversity of taxa (English-Loeb et al., 1993). However, young larvae at the Bodega Bay Marine Laboratory feed primarily upon a single host, bush lupine (Lupinus arboreus: Fabaceae). As larvae become larger and more mobile, they also feed on other plants, including L. nanus (Fabaceae), Amsinckia menziesii (Boraginaceae), Conium maculatum (Apiaceae), Plantago lanceolata (Plantaginaceae), Iris (Iridaceae), and various grass species (Poaceae) (English-Loeb et al., 1993). Thus, I hypothesized that larger larvae would perform better than small larvae on a mixed diet compared to a uniform diet of bush lupine. In this study I asked how diet and larval size affect performance (measured as survival and growth) of the woolly bear P. virginalis. The question of how larval size affects performance on mixed diets compared to uniform diets has not been previously addressed in a field study. Because age and size covary, these experiments are a test of larval age as much as size. For simplicity, I will use the term ‘‘size’’ hereafter with the understanding that age is confounded with size. METHODS Adult moths of P. virginalis are univoltine at Bodega Bay and oviposit in May or June. Eggs hatch in the summer and remain as first or second instar larvae until late winter, when they begin to feed rapidly (English-Loeb et al., 1993). Early instars at Bodega Bay live at the base of bush lupines and move little. In the spring, larvae crawl into the foliage and are evident feeding on bush lupines, and a diversity of other foliage, until pupation in April or May (Karban, 1998). I conducted experiments over 2 years at Horseshoe Cove and Mussel Point at the Bodega Bay Marine Laboratory in Sonoma County, California (388199N, 1238049W). In both years, I used enclosures to create diet treatments. Enclosures consisted of chicken-wire cylinders approximately 10 cm high and 20 to 25 cm in diameter, with the wide end of a conical spun polyester bag wrapped underneath. I secured chicken-wire cylinders to the ground with 4 metal U-shaped stakes and tied the top narrow end of the conical bag with plastic flagging. I placed enclosures either around a small bush lupine with all other vegetation removed from the enclosure (lupine only treatment) or around a diverse mix of vegetation that included 1 bush lupine (mixed diet treatment). The exact mixture of plants in the mixed diet treatment varied among enclosures, but estimates from a pilot study at Horseshoe Cove (AdJune 2004 191 Adler—Performance of a generalist herbivore ler, unpubl. data) indicated that the most common species, comprising over 88% of the ground cover, were various grasses (Poaceae), Plantago lanceolata (Plantaginaceae), Rumex acetosella (Polygonaceae), and Lupinus arboreus and L. nanus (Fabaceae). Other taxa were Eschscholzia californica and Platystemon californicus (Papaveraceae), Triphysaria pusilla (Scrophulariaceae), Anagallis arvensis (Primulaceae), Acaena pinnatifida and Rubus (Rosaceae), Erodium cicutarium and Geranium molle (Geraniaceae), Claytonia perfoliata (Portulacaceae), Stachys rigida (Lamiaceae), Nemophila menziesii and Phacelia distans (Hydrophyllaceae), Achillea millefolium and Hypochoeris radicata (Asteraceae), Trifolium (Fabaceae), Carex (Cyperaceae), and Daucus carota (Apiaceae). Enclosures from the pilot study (n 5 60) had an average of 4.6 6 0.2 species and a range of 2 to 8 species. I collected woolly bear larvae on 23 and 24 March 1997. I recorded host plant during collection and stored all larvae in a refrigerator in paper cups with bush lupine leaves until the beginning of the experiment. On 25 March, all larvae were individually weighed and placed in separate enclosures. Seven replicates were used for each larval size (small or large) and treatment (mixed diet or lupine-only diet) combination, for a total of 28 enclosures. While this is a small sample size (n 5 7), I found statistically significant differences even when considering this year of data alone, suggesting that samples sizes were sufficient to detect treatment effects. Larvae under 0.40 g were considered small and were typically second or third instars, judging by their color patterns (R. Karban, pers. comm.). Larvae above 0.75 g were considered large and were typically fourth or fifth instars, although this category might have included some large third instar larvae. Larvae weighing 0.40 to 0.75 g were not used. Larvae were collected, weighed, and returned to their enclosures on 30 March, 4 April, 9 April, 14 April, and 27 April. The status of each larva (alive, dead, missing, or pupated) was recorded on each date. On 9 April, damage to all plants in the mixed diet treatment was recorded to determine whether larvae were truly consuming a mixed diet. The same methods were repeated in 1998. Bush lupines of the appropriate size were rare at Horseshoe Cove during this year due to efforts to control bush lupine seedlings in the grasslands at Bodega Bay (P. G. Connors, pers. comm.). The experiment was therefore conducted at a nearby area on the opposite side of the station with similar vegetation (on the Mussel Point side of the station, approximately 0.28 km north of the 1997 site). Woolly bears were weighed and placed in enclosures on 3 March and then weighed and returned to their enclosures on 7 March, 12 March, 21 March, 28 March, and 4 April. Larvae under 0.27 g (second-third instar) were considered small, and larvae above 0.60 g (fourth-fifth instar) were considered large. Fifteen replicates were used per larval size and treatment combination, for a total of 60 enclosures. The effect of diet treatment, larval size, and year on larval weight was determined using repeated measures 3-way ANOVA, with initial weight as a covariate and treatment, year, and larval size as main effects. All interaction terms were included in the model. Using initial weight as a covariate and final weight as a response is essentially analyzing relative growth rate but without using ratios; I will refer to this response as growth. Due to missing cells caused by larval escape, pupation, or death, only data from the first 2 weighing dates (5-d and 10-d periods in 1997 and 4-d and 9-d periods in 1998) were used in this analysis. The effects of diet treatment, larval size, and year on survival were tested using a maximum likelihood analysis of variance with the categorical modeling procedure of SAS, version 8.02 (SAS Institute, Cary, North Carolina). Missing larvae were included with dead larvae, and pupated larvae were included with larvae that lived but did not pupate. Due to small sample size, it was not possible to test a full 3-way model with interaction terms. The interaction between larval size and treatment and the main effects of treatment, larval size, and year were included in the model. To determine whether lumping data across years obscured qualitative differences between years, separate G-tests with Yates correction for continuity (Zar, 1996:503) were used for each year and larval size combination. RESULTS Eight plant taxa from 5 families were observed with damage within the enclosures in the 1997 study (Lupinus arboreus, L. nanus, grasses, Plantago lanceolata, Rumex acetosella, Acaena pinnatifida, Trifolium, and Rubus). Two to 5 plant species were observed with damage within each enclosure (mean 6 SD: 2.9 6 0.2; large larvae 3.0 6 0.3; small larvae 2.7 6 0.3), demonstrating that individual small and large woolly bears in the mixed diet treatment consumed a relatively mixed diet. Lupinus arboreus received damage in every enclosure. Rumex acetosella, grass, and Plantago lanceolata also were damaged frequently (64%, 57%, and 29% of enclosures, respectively). Growth (final weight using initial weight as a covariate) of large larvae compared to small larvae differed between diet treatments and years of the experiment (size 3 diet and size 3 year interactions; Table 1, Fig. 1) in a repeated measures analysis. In 1997, large larvae had higher growth on a lupine-only diet compared to mixed diets, but small larvae had 192 vol. 49, no. 2 The Southwestern Naturalist TABLE 1—Effect of diet type, larval size, and year on larval weight using initial weight as a covariate for wooly bear larvae (Platyprepia virginalis). Repeated measures analysis with Type III sums of squares of the between-subjects effect of initial larval size, treatment, and year on subsequent larval weight.