Notes Ecology, 84(1), 2003, pp. 263–268 q 2003 by the Ecological Society of America ANOTHER PERSPECTIVE ON THE SLOW-GROWTH/HIGH-MORTALITY HYPOTHESIS: CHILLING EFFECTS ON SWALLOWTAIL LARVAE JAMES A. FORDYCE1 AND ARTHUR M. SHAPIRO Section of Evolution and Ecology, Center for Population Biology, One Shields Avenue, University of California, Davis, California 95616 USA Abstract. The slower-growth/higher-mortality hypothesis proposes that reduced herbivore growth rates benefit plants because slower growing herbivores remain vulnerable to predator attack for an extended time period, resulting in a lower reduction in plant fitness attributed to herbivory. We propose cooling events as an alternative mechanism leading to higher mortality for slower growing larvae. We observed that, in the absence of predators, slower growing pipevine swallowtail (Battus philenor) larvae had higher mortality compared to faster growing larvae during the unseasonably cool spring temperatures of 1998. However, this was not true for the warmer spring of 1999. Laboratory experiments showed that the probability of surviving a chill coma increased with larval mass. We propose that smaller larvae are susceptible to chilling events because they have less energy reserved for metabolism during a chill coma, and that ‘‘warm’’ chill events near the activity threshold may be more lethal than ‘‘cold’’ chill events. Weather station data since 1951 and field data collected from 1976 to 2000 suggest that exposure to chill events below the chill coma threshold is not uncommon for these larvae. We propose that faster growth may be important for larvae because slower growing larvae remain at a chill-susceptible size for an extended period of time. Key words: Battus philenor; butterfly larvae; chill coma; growth rate; herbivore; Lepidoptera; mortality; pipevine swallowtail; plant–insect interactions; slower-growth/higher-mortality hypothesis; temperature; weather. INTRODUCTION Insect growth rate, size, and predation rate specific to the life history stage have been used to develop adaptive hypotheses of plant defense theory. Specifically, those plant defenses that effectively reduce herbivore growth rates allow the plant to exploit the suite of herbivore natural enemies that are most effective against insect larvae in early developmental stages (Feeny 1976). This idea, termed the slower-growth/ higher-mortality hypothesis (Clancy and Price 1987), posits that slower growing insect larvae remain in a ‘‘window of vulnerability’’ for an extended period of time. The support for this hypothesis is mixed, varying among taxa and natural enemy guilds (Williams 1999). Like many theories of plant defense, the slower-growth/ higher-mortality hypothesis champions the importance of higher trophic levels for regulation of herbivore populations (Hairston et al. 1960). However, less attention has been given to how slow growth may also interact Manuscript received 5 November 2001; revised and accepted 11 June 2002. Corresponding Author: S. J. Simpson. 1 E-mail: jafordyce@ucdavis.edu with abiotic factors such as weather (Andrewartha and Birch 1954, Morris 1964, 1969, Reavey 1993). The effect of weather, especially temperature, on the ecology of insect larvae has largely focused on how temperature affects phenology, foraging, and growth rates (Taylor 1981, Casey et al. 1988, Kukal and Dawson 1989, Sømme 1989, Casey 1993, Stamp 1993, Neal et al. 1997, Kingsolver 2000), or directly on mortality through freezing (Klok and Chown 1997, 1998). All insects have an optimum temperature at which growth rate is maximized, and deviations from this optimum result in decreased growth rates (Taylor 1981, Danks 1987). Recent models have suggested that temperature and food quality can interact in ways that influence insect growth rate and fecundity (Kindlmann et al. 2001). Suboptimal temperatures have been shown to interact with host-plant quality in ecologically significant ways (Stamp 1993). For example, Morris (1964, 1967) showed that in cold years, fall webworm larvae (Hyphantria cunea; Arctiidae) develop at a decreased rate and, as a result, must feed on older, lower quality foliage. As a result of the combined effects of temperature and food quality, many larvae are unable to 263 264 NOTES Ecology, Vol. 84, No. 1 PLATE. 1. An aggregation of first-instar Battus philenor larvae on their host plant, Aristolochia californica. In California, larvae feed in aggregation until late in the third instar when they usually disperse into smaller groups or feed singly. Photograph by James A. Fordyce. pupate before winter and they die. He observed that outbreaks of the fall webworm occur following unseasonably warm years and attributed this to an increase in the number of larvae that successfully pupate because of more optimal developmental temperatures and host-plant quality. Less attention has been paid, however, to temperature-dependent mortality that occurs when larvae are in a chill-induced coma, also referred to as a cold stupor or critical thermal minimum (Taylor 1981, Block 1990). Chill coma occurs at temperatures above the lethal chilling threshold and below the range of temperatures at which insects can move and feed. The size of larvae when entering a chill coma may be important if larvae remain in this state under temperatures that still require a significant amount of maintenance metabolism and, subsequently, some metabolic cost. Thus, warmer chill comas may be more lethal than cold chill comas because of an increased metabolic demand. Slower growing, smaller larvae may be more sensitive to operating at an energy deficit because they have lower stored energy reserves. The ecological relevance of operating in such an energy deficit has been described in some systems. For example, Kukal and Dawson (1989) suggested that arctic woolly-bear larvae (Gynaephora groenlandica; Lymantriidae) may undergo voluntary hypothermia to avoid an energy deficit when the energy demands of maintenance metabolism exceed the energy available through food assimilation. Mild winter temperatures have been shown to increase overwintering mortality and decrease fecundity of the goldenrod gall fly (Eurosta solidaginis; Tephritidae), presumably because of the increased metabolic demands imposed by higher temperatures (Irwin and Lee 2000). In the spring of 1998, California experienced periods of unseasonably cold temperatures due to persistent overcast associated with an El Niño. This meteorological event provided an opportunity to conduct a natural experiment on the effects of temperature on survivorship of pipevine swallowtail (Battus philenor; Papilionidae) larvae. In California, B. philenor adults emerge from diapaused pupae in early March. Females lay clusters of eggs (mean ;13 eggs) on their only available host plant, Aristolochia californica (Aristolochiacae). Upon hatching, larvae feed in aggregations, and the number of individuals in a feeding group is positively correlated with their growth rate (Fordyce and Agrawal 2001; see Plate 1). This afforded us the opportunity to observe the effect of larval growth rate on larval mortality between an unusually cold spring (1998) and a ‘‘normal’’ spring (1999). Here, we investigated the following: Was mortality different between cold and normal years for larvae developing at different rates? At what ambient temperature do B. philenor enter chill coma? Does the size NOTES January 2003 of larvae entering chill coma impact their mortality? How commonly are California B. philenor larvae exposed to chill coma temperatures? MATERIALS AND METHODS Growth and survivorship differences between 1998 and 1999 To demonstrate that group size can affect growth rate, we conducted an experiment at Stebbins Cold Canyon Ecological Reserve (University of California Natural Reserve System) in Solano County, California, USA, located in the inner Coast Range. Newly hatched larvae were assigned to groups of two (N 5 10) or 12 (N 5 11) individuals and were permitted to feed for 48 h. Larvae were confined to spun polyester mesh bags (Kleen Test Products, Brown Deer, Wisconsin, USA) to exclude predators. After two days of feeding, larvae were returned to the lab and weighed as a measure of growth rate. Mean larval mass was compared between the two treatments using a t test. To assess whether mortality rates between cold and normal years differed for larvae developing at different rates, we observed survivorship of larvae on plants in the field. Experiments in both years were conducted at Stebbins Cold Canyon. In 1998, larvae were placed on plants in feeding groups of 1, 4, 8, or 16 individuals; replicates for each group size treatment were 48, 34, 29, and 23, respectively. In 1999, larvae were placed in groups of 1, 2, 4, 6, 8, 10, 12, 14, and 16 individuals, and each group size was replicated 10 times. Predators were excluded in both years using spun polyester mesh bags. These larvae were observed each day through the first molt, and mortality was noted. Threshold temperature for larval activity To determine the threshold temperature at which larvae begin to recover from a chill coma, we conducted a laboratory experiment in which larvae were observed over a temperature gradient. Forty 2-d-old larvae that had previously fed were chilled to 88C in a growth chamber and each was placed on an individual A. californica sprig. The growth chamber was warmed slowly, ;0.048C/min, over the next 240 min. Every 10 min, the larvae were observed to determine the temperature at which they begin to show movement and feeding. Temperature in the growth chamber was recorded using a data logger. Larval mass and mortality in chill coma Here, we were interested in the relationship between larval mass and survival while larvae were in a chill coma. We used larvae that had fed in the field for 1– 4 d in groups of two individuals (slower growth rate) and 12 individuals (faster growth rate). In total, 243 265 larvae were individually weighed and placed in 1.5mL plastic tubes. These larvae were placed in an incubator maintained between 108 and 118C. After 60 h elapsed, we recorded whether each individual had survived this chilling period. We used logistic regression to assess how larval mass affected the probability of surviving this chill period. The phenology of B. philenor and the history of chilling events To estimate the average date of emergence of B. philenor in the focal area of this study, we used 25 yr of field data collected by one of us (A. M. Shapiro) from 1976 to 2000 at Gates Canyon (Solano County, California, USA). Gates Canyon is ;12 km due south of Cold Canyon. The census of adult butterflies was conducted using a fixed-course walk along a road in the canyon during periods of time amenable to butterfly flight (i.e., sunny and low wind) approximately every 14 d throughout the year. From 1998 to 2001, the emergence of B. philenor in Gates Canyon occurred within 5 d of the emergence at Cold Canyon. Based upon observations from 1998 to 2001, the peak of egg-laying activity begins ;3–4 wk after the first sighting and continues for 3 wk, after which it quickly tapers off. We were interested in determining how frequently potentially important chilling events occur in this area of California. To do this, we used weather station records that date to 1951 from the city of Winters (Yolo County, California), the nearest weather station to Stebbins Cold Canyon (;6.5 km due east). Based on these data, we estimated how frequently B. philenor larvae may be exposed to days with temperatures continuously below the chill coma threshold. RESULTS AND DISCUSSION The number of individuals in a feeding aggregation affects the growth rate of first-instar Battus philenor larvae. During a 2-d period with a mean daily maximum temperature of 248C, larvae feeding in groups of two individuals had significantly lower body mass (1736 6 194 mg, mean 6 1 SE) than larvae feeding in aggregations of 12 individuals (2455 6 186 mg; df 5 19, t 5 2.67, P 5 0.015). The reduced growth rate associated with smaller groups of first-instar larvae is similarly reflected in the time that it takes individuals to molt into the second instar. For example, in 1998 single individuals remained in the first instar 2 d longer than groups of 16, and groups with four and eight individuals remained in the first instar 1 d longer than groups of 16 (F3, 130 5 3.20, P 5 0.014; ANOVA on days in first instar). First-instar larvae are more susceptible to many predators than are later instars (J. A. Fordyce, personal observation). Thus, the importance of slower growth/on the vulnerability to predators pre- 266 NOTES dicted by the slower-growth/higher-mortality hypothesis may operate in this system (Fordyce and Agrawal 2001). However, differences in mortality associated with different growth rates cannot be explained by increased predator exposure when predators are excluded. Survivorship of first-instar larvae in the absence of predators was lower for slower developing larvae during the El Niño event in 1998 (F3, 164 5 3.20, P 5 0.025). Survivorship for the slowest growing larvae (single individuals) was 59%. Survivorship for faster growing larvae (groups of 4, 8, and 16 individuals) was 77%, 80%, and 90%, respectively. We observed that many of the larvae did not move for nearly an entire week during the coldest periods of 1998. In 1999, there was no effect of growth rate on mortality (data are group size and survivorship: 1, 90%; 2, 83%; 4, 90%; 6, 95%; 8, 100%; 10, 98%; 12, 100%; 14, 90%; 16, 93%; F8,79 5 0.76, P 5 0.64). One explanation for the difference in survivorship between these two years is that the cooler conditions during the El Niño year compared to the warm temperatures of the following year played a role in larval mortality and that growth rate was an important factor interacting with temperature in 1998. To investigate the plausibility of this idea, we conducted laboratory experiments to assess the importance of larval mass on mortality during chill coma and to establish the chill coma temperature threshold. The experiment in which the incubator temperature was slowly increased was conducted to establish the approximate threshold temperature for the chill coma. The first activity was observed when the temperature in the incubator reached 158C. However feeding was not observed and the larva remained stationary. The first larva was observed feeding at 16.28C, and when the temperature reached 16.58C, nearly all of the larvae were observed feeding. The amount of feeding and movement increased as temperature continued to rise throughout the remainder of the experiment. Based on these observations, we approximate the chill coma threshold for B. philenor first-instar larvae to be between 16.28C and 16.58C. This temperature is higher than that described for many other lepidoptera (Klok and Chown 1997, 1998, Fitzgerald and Underwood 2000). However, it should be noted that the genus Battus is largely tropical and subtropical (Racheli and Pariset 1992). Thus, the higher chill coma threshold for B. philenor may reflect physiological constraints imposed by a tropical evolutionary history. The body mass of a larva entering a chill coma had a significant effect on the probability of its survival through the chill coma period of 60 h. The probability of surviving through the chill coma increased as larval mass increased (df 5 1, x2 5 58.40, P , 0.0001, r2 5 Ecology, Vol. 84, No. 1 FIG. 1. The probability of surviving a 60-h chill coma at 108–118C over a range in body mass of first-instar larvae of the pipevine swallowtail, Battus philenor. 0.20; Fig. 1). Larvae with mass ,1600 mg, roughly double the mass of a newly hatched larva, had ,50% chance of surviving through this chill coma. However, we often store larvae of various ages, including newly hatched, for up to 72 h at 28C with little or no resulting mortality. This suggests that larvae are more vulnerable when in a ‘‘warm’’ chill coma than in a ‘‘cold’’ chill coma. The importance of size in warmer chill comas may reflect an increase in metabolic demand for stored resources as the temperature approaches the chill coma threshold. Using the weather station data from Winters, we considered days with a maximum temperature ,16.28C to be days below the chill coma threshold. This is probably a conservative estimate of chilling events for Cold Canyon and other nearby canyons because, unlike the canyons, Winters is exposed to full sunlight for most of the day. For example, temperature data from 1998 to 2000 collected using a data logger at Cold Canyon indicated that the maximum daily temperatures at Cold Canyon are normally 28–48C cooler than those recorded at the Winters weather station. Fig. 2 shows the average emergence date of B. philenor from 1976 to 2000 and indicates the days when the maximum temperature was below 16.18C since 1951. To account for sampling error due to the biweekly (once every two weaks) sampling regime, we estimate that the average emergence of B. philenor occurs roughly between 22 February and 8 March (which also corresponds with the 99% CI around the mean observed emergence date of 2 March) and that peak egg laying begins 3–4 wk later (roughly between 23 March and January 2003 NOTES 267 28.78C). This cooler temperature is reflected in the developmental rate of larvae, with larvae remaining in the first instar 5–6 d longer in 1998 than in 1999. The slower growth rate associated with smaller groups of larvae, coupled with the cooler temperatures in 1998, may have prevented these larvae from attaining a size less susceptible to extended chill comas. However, the cold period in early April of 1999 may similarly have had a lethal effect on early clutches of larvae during that year. CONCLUSION FIG. 2. History of spring chilling events recorded at the Winters weather station, Yolo County, California, USA. Blocked regions indicate days below the chill threshold of 16.18C for the pipevine swallowtail. Mean emergence of Battus philenor at Gates Canyon, 13.5 km southwest of Winters, is based upon field observations from 1976 to 2000. Peak egg laying and first-instar presence are based upon field observations from 1998 to 2001 at Stebbins Cold Canyon, 6.5 km east of Winters. 30 March). Eggs usually hatch within 5–10 d, depending on the temperature. In total, 39 of the 50 yr of recorded weather data had periods of at least two consecutive days below the chill coma threshold that occurred after the mean emergence time interval. Only four years lacked chill coma temperatures after 1 April. These phenological and temperature data suggest that chill coma periods can occur when developing larvae are most vulnerable to cold temperatures. The difference in survivorship that we observed for first-instar larvae between 1998 and 1999 was likely due to the cold periods observed in May of 1998. Our experiments where group size, and subsequently growth rate, was manipulated began during the first week of May for both years. Based on the Winters weather station data, May of 1998 was much cooler (average daily maximum temperature 21.68C) compared to 1999 (average daily maximum temperature The slower-growth/higher-mortality hypothesis was developed to explain why plants should invest in defense strategies that reduce herbivore developmental rates, and, indeed, there are some demonstrations that herbivore natural enemies exploit a prolonged exposure of herbivores in a vulnerable stage (Williams 1999). Here, we propose another ecological consequence of slower growth. Specifically, there exists a growth-ratedependent window of vulnerability to chilling events for developing larvae, and slower growing larvae remain in this vulnerable state for an increased amount of time. Plants may benefit if herbivore growth rates are reduced because slower growing larvae may be more vulnerable to cold periods. Interestingly, warmer chilling events may be more costly because of an increased metabolic demand. Thus, the ecological consequences are the same as those initially predicted by the slowergrowth/higher-mortality hypothesis, but the underlying mechanism is different. To truly test this hypothesis and assess the importance of late-spring cold events, such as those observed we in 1998, extensive time series data on temperature, larval density, and herbivore-inflicted plant damage are required. To our knowledge, no such data exist for this system, and the required labor and time effort involved in such a study makes it unlikely that they will exist. However, our data suggest that cool weather may, at times, interact with herbivore developmental rates in ecologically significant ways. ACKNOWLEDGMENTS We thank R. Karban for helpful comments and discussion, and one anonymous reviewer. Thanks to S. Strauss for use of her micro-balance, and V. Boucher and D. Tolson of the University of California Natural Reserve System for providing logistical support for work at Stebbins Cold Canyon. This study was supported by Public Service Research Program and the Putah-Cache Bioregion Project (UC Davis), a Mildred E. Mathias Graduate Student Research Grant (University of California Natural Reserve System), Graduate Group in Ecology and Jastro Shields Awards Program (UC Davis), Center for Population Biology (UC Davis), Sigma-Xi Grants-in-Aid of Research, and the U.S. National Science Foundation (DEB9306721). NOTES 268 LITERATURE CITED Andrewartha, H. G., and L. C. Birch. 1954. The distribution and abundance of animals. University of Chicago Press, Chicago, Illinois, USA. Block, W. 1990. Cold tolerance of insects and other arthropods. Philosophical Transactions of the Royal Society of London B, Biological Sciences 326:613–633. Casey, T. M. 1993. Effects of temperature on foraging caterpillars. Pages 5–28 in N. E. Stamp and T. M. Casey, editors. Caterpillars: ecological and evolutionary constraints on foraging. Chapman and Hall, New York, New York, USA. Casey, T. M., B. Joos, T. D. Fitzgerald, M. E. Yurlina, and P. A. Young. 1988. Synchronized group foraging, thermoregulation, and growth of Eastern tent caterpillars in relation to microclimate. Physiological Zoology 61:372– 377. Clancy, K. M., and P. W. Price. 1987. Rapid herbivore growth enhances enemy attack: sublethal plant defenses remain a paradox. Ecology 68:733–737. Danks, H. V. 1987. Insect dormancy: an ecological perspective. Biological Survey of Canada Monograph Series Number 1, Ottawa, Canada. Feeny, P. 1976. Plant apparency and chemical defense. Pages 1–40 in J. W. Wallace and R. L. Mansell, editors. Biochemical interaction between plants and insects. Plenum Press, New York, New York, USA. Fitzgerald, T. D., and D. L. A. Underwood. 2000. Winter foraging patterns and voluntary hypothermia in the social caterpillar Eucheira socialis. Ecological Entomology 25: 35–44. Fordyce, J. A., and A. A. Agrawal. 2001. The role of plant trichomes and caterpillar group size on growth and defence of the pipevine swallowtail, Battus philenor. Journal of Animal Ecology 70:997–1005. Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure, population control, and competition. American Naturalist 44:421–425. Irwin, J. T., and R. E. Lee, Jr. 2000. Mild winter temperatures reduce survival and potential fecundity of the goldenrod gall fly, Eurosta solidaginis (Diptera: Tephritidae). Journal of Insect Physiology 46:655–661. Kindlmann, P., A. F. G. Dixon, and I. Dostálková. 2001. Role of ageing and temperature in shaping reaction norms and fecundity functions in insects. Journal of Evolutionary Biology 14:835–840. Ecology, Vol. 84, No. 1 Kingsolver, J. G. 2000. Feeding, growth, and the thermal environment of cabbage white caterpillars, Pieris rapae L. Physiological and Biochemical Zoology 73:621–628. Klok, C. J., and S. L. Chown. 1997. Critical thermal limits, temperature tolerance and water balance of a sub-antarctic caterpillar, Pringleophaga marioni (Lepidoptera: Tineidae). Journal of Insect Physiology 43:685–694. Klok, C. J., and S. L. Chown. 1998. Interactions between desiccation resistance, host-plant contact and the thermal biology of a leaf-dwelling sub-antarctic caterpillar, Embryonopsis halticella (Lepidoptera: Yponomeutidae). Journal of Insect Physiology 44:615–628. Kukal, O., and T. E. Dawson. 1989. Temperature and food quality influences feeding behavior, assimilation efficiency and growth rate of arctic woolly-bear caterpillars. Oecologia 79:526–532. Morris, R. F. 1964. The value of historical data in population research, with particular reference to Hyphantria cunea Drury. Canadian Entomologist 96:356–368. Morris, R. F. 1969. Approaches to the study of population dynamics. Pages 9–28 in W. E. Waters, editor. Forest insect population dynamics. USDA Forest Service Research Paper NE-125, U.S. Government Printing Office, Washington, D.C. Neal, J. W., J. L. Chittams, and J. Bentz. 1997. Spring emergence by larvae of the eastern tent caterpillar (Lepidoptera: Lasiocampidae): a hedge against high-risk conditions. Annals of the Entomological Society of America 90:596–603. Racheli, T., and L. Pariset. 1992. II genere Battus tassonomia e storia naturale. Fragmenta Entomologica (Roma) Supplemento 23:1–163. Reavey, D. 1993. Why body size matters to caterpillars. Pages 248–279 in N. E. Stamp and T. M. Casey, editors. Caterpillars: ecological and evolutionary constraints on foraging. Chapman and Hall, New York, New York, USA. Sømme, L. 1989. Adaptations of terrestrial arthropods to the alpine environment. Biological Reviews of the Cambridge Philosophical Society 64:367–408. Stamp, N. E. 1993. A temperate region view of the interaction of temperature, food quality, and predators on caterpillar foraging. Pages 478–508 in N. E. Stamp and T. M. Casey, editors. Caterpillars: ecological and evolutionary constraints on foraging. Chapman and Hall, New York, New York, USA. Taylor, F. 1981. Ecology and evolution of physiological time in insects. American Naturalist 117:1–23. Williams, I. S. 1999. Slow-growth, high-mortality—a general hypothesis, or is it? Ecological Entomology 24:490–495.