Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 317 2017 NORTHEASTERN NATURALIST 24(3):317–330 Overwinter Parasitism of Callosamia promethea (Drury) (Saturniidae) (Promethea Moth) in a Northern Hardwood Forest Stephanie M. Juice1,* and Bernd Heinrich2 Abstract - The giant silk moth Callosamia promethea (Promethea Moth) overwinters as pupae in cocoons spun in hanging leaves. Even though they are camouflaged, moths at this developmental stage are vulnerable to parasitism and predation, both of which are relevant to local population abundance and persistence. Our study documented the fates of overwintering Promethea Moths near Weld, ME, during 7 winters from 1997 to 2017. Our collection and dissection of 923 cocoons revealed that moth emergence declined from 47% to 10% over the study period. Parasitism by 2 native species of ichneumonid wasp (Gambrus nuncius and Enicospilus americanus) was the dominant cause of mortality, accounting for 59% of Promethea Moth pupal death. Predation accounted for less than 4% of mortality. Our results provide evidence of parasitism as a major contributor to mortality of Promethea Moth pupae. Introduction Callosamia promethea (Drury) (Promethea Moth or Spicebush Silkmoth) belongs to the group of giant silk moths (Lepidoptera: Saturniidae), which contains some of the largest and best-known moths in the US (Tuskes et al. 1996). Population declines of giant silk moth species formerly common in the Northeast, including the Promethea Moth, have been of concern since the mid-20th century (Ferguson 1971, Hessel 1976, Muller 1979, Schweitzer 1988, Tuskes et al. 1996, Wagner 2012), with some reported declines occurring as early as 1919 (Culver 1919). Various anthropogenic causes for the decline have been posited, including the effects of outdoor lighting (Pyle et al. 1981, Worth and Muller 1979), spraying of DDT (Holden 1992), and the introduction of a polyphagous biocontrol agent (Compsilura concinnata [Meigen], Diptera: Tachinidae) intended to target nonnative moth pests (Boettner et al. 2000, Culver 1919). The decline has also been attributed to natural changes, such as increased predation on cocoons or ecological succession, which may reduce habitat availability for the Promethea Moth because it is considered to be an early successional species (Wagner 2012, Wagner et al. 1981). Yet others have pointed to combinations of causes, for example increased predation by birds and bats due to the presence of outdoor lighting (Frank 1988). Although possible causes of the Promethea Moth’s overall decline in the Northeast have been proposed, observations of local population fluctuations in this species are largely unexplained. The Promethea Moth inhabits the deciduous forest 1University of Vermont, Rubenstein School of Environment and Natural Resources, Burlington, VT 05401. 2University of Vermont, Department of Biology, Burlington, VT 05401. *Corresponding author- stephanie.juice@uvm.edu. Manuscript Editor: Christopher M. Heckscher Northeastern Naturalist 318 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 region of eastern North America, ranging from southern Ontario and Quebec, west to the eastern edge of the Great Plains, and south to northern Florida (Ferguson 1971). Its range is expansive, but local presence and abundance vary greatly (Ferguson 1971, Riotte 1992). It has, in fact, been reported to have disappeared where it was previously abundant, only to reappear years later (Hessel 1976, Riotte 1992, Tuskes et al. 1996). The source of this fluctuation is unknown, but suggested causes include predation, parasitism, disease, and migration within the population’s range (Tuskes et al. 1996). Our study took place on York Hill, in the northern hardwood forest near Weld, ME. We made observations of overwintering Promethea Moths at this site over a 20-year period that catalog rates of moth eclosion, parasitism by 2 native wasps, and on-tree predation. Here, we synthesize these data to shed light into the local overwintering dynamics of the Promethea Moth near the northern limit of its range. In northern New England, adult Promethea Moths emerge from their cocoons in June. Eclosion takes place in the morning, and they mate in the late afternoon the same day. Females attract males through pheromone release (Ferguson 1971), and the pair stays together until dusk, at which time the female begins oviposition (Tuskes et al. 1996). For several nights, the female lays eggs on larval food plants, which encompass a large variety of species across the moth’s range (Ferguson 1971). After hatching, the caterpillars initially feed in groups (Tuskes et al. 1996), which may increase their vulnerability to parasitism. Indeed, much of the parasitism that ultimately kills the pupa is initiated by attack on the caterpillar. This pattern is particularly true for one of the parasitic wasps native to York Hill: Enicospilus americanus (Christ) (Hymenoptera: Ichneumonidae), a large wasp that produces 1 adult from each host it infects (Peigler 1977). It specializes on saturniid moths and has been documented to parasitize 14 species (Peigler 1994). This wasp is distributed across the Americas, from northern Argentina to Eastern Canada (Ontario and Quebec), and from coast to coast across the US (Peigler 1994). Enicospilus americanus attacks caterpillars during the early to mid-larval stage by inserting an egg into the Promethea Moth caterpillar that will hatch but stay small until triggered to grow later in the moth’s development (Peigler 1994). By late July or August, Promethea Moth caterpillars stop feeding, disperse, and spin cocoons by rolling a leaf along its midvein and spinning a silk lining which is firmly attached to the twig of the host plant by a long, silk peduncle (Fig. 1c; Waldbauer and Sternburg 1982). This strand secures the cocoon to the tree throughout the winter, and perhaps for many years (Waldbauer and Sternburg 1982). After firmly attaching its leaf to the tree, the caterpillar spins a double-walled, compact cocoon inside (Tuskes et al. 1996), leaving an exit hole at the top from which the adult moth, or an adult parasitoid, can emerge. Before the cocoon hardens, the moth larva is vulnerable to attack by parasitoids that can insert eggs through the soft silk. The scent of freshly spun silk may attract ichneumons; they have been observed flying upwind toward new cocoons (Marsh 1937). Gambrus nuncius (Say) (Hymenoptera : Ichneumonidae), the 2nd parasitic wasp native to York Hill, employs this strategy to attack the Promethea Moth larva (Peigler 1994). Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 319 Gambrus nuncius is a small wasp that stings its host as it is spinning its cocoon (Peigler 1977). The wasp larvae consume the host as ectoparasites (Peigler 1977) and form a tightly packed mass of cocoons inside the host cocoon (Peigler 1994). At York Hill, 20–40 adults hatch from each parasitized host. Gambrus nuncius ranges across Canada, from New Brunswick to the Northwest Territories, and as far south as Alabama (Peigler 1994). Like E. americanus, it is a specialist on Saturniids, and has been documented to parasitize 5 species, including all 3 species in the genus Callosamia (Peigler 1994). After spinning its cocoon, the Promethea Moth larva undergoes a hormonal change that initates pupation. If the moth larva was previously attacked by E. americanus, this hormonal change causes the wasp larva to grow quickly, consume its host, and pupate inside the host’s cocoon (Peigler 1994). Enicospilus americanus adults exit through the valve made at the top of the cocoon by the caterpillar, while G. nuncius chews 1 or 2 exit holes in the side of the host cocoon (Peigler 1977). Both wasp species leave behind the exuviate of the host, and their own cocoon which is hard and dark (Fig. 2e), or a bundle of numerous white cocoons (Fig. 2f) for E. americanus and G. nuncius, respectively (Peigler 1977). The Promethea Moth’s life-history strategy leaves behind a record of its fate. Its tough cocoon securely anchored to a tree, when dissected, contains evidence of successful moth eclosion, parasitization, or predation that occurs on the tree. In this study, we synthesized data on cocoons that we collected over the course of 20 years on York Hill, in Maine. In all years, we used the same protocol to collect and dissect Promethea Moth cocoons. We identified and catalogued moth pupae or pupal cuticle, parasites or their remnants, and mortality due to on-tree predation, as well as unexplained death of the caterpillar or pupa. Field-site Description We conducted this study at York Hill (44°23'60''N, 70°13'12''E), a 121-ha (300- ac) property in the western Maine foothills near the town of Weld. This site is near the northernmost documented occurrence of the Promethea Moth, which is in Fredericton, NB, Canada (45°57'48''N, 66°38'35''W). Temperatures in Weld range from the mean minimum of -14 °C in January to the mean maximums of 22 °C and 25 °C in June and July, respectively. The frost-free period ranges from 110 to 140 days. Dominant canopy trees at the site include Acer rubrum L. (Red Maple), A. saccharum Marsh (Sugar Maple), and Fraxinus americana L. (White Ash). Fagus grandifolia Ehrh. (American Beech), Prunus serotina Ehrh. (Black Cherry), and Corylus cornuta Marshall (Beaked Hazelnut) are predominant in the understory. Methods We collected cocoons during all months after leaf-fall, but most intensively for 1 week in January of each collection year (1997, 2005, 2007, 2012, 2014, 2015, and 2017) as part of a class on Winter Ecology offered by the University of Vermont. Course instruction ensured uniformity of search area (121 ha), criteria, and Northeastern Naturalist 320 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 methodology across years. Promethea Moth cocoons are unusual in their strong attachment, typically on low tree branches, where they persist for several years (Fig. 1). They become highly visible after leaf-fall, permitting easy access. Promethea Moths utilize a wide variety of host plants from 6–10 families across their range, which contributes to their extensive geographical distribution (Peigler 1976, Scriber et al. 1991). At the York Hill site, preferred larval food plants are White Ash and Black Cherry, and searchers found cocoons within close proximity to these species. We regularly found cocoons on non-food plants, including those of conifers (Abies sp. [fir] and Picea sp. [spruce]), provided they were near the larval food source. After collection, we dissected the cocoons and categorized their contents as one of the following: successful Promethea Moth (live pupa or pupal shell indicating prior moth eclosion), successful G. nuncius (live pupa or empty cocoons), successful E. americanus (live pupa or empty cocoon), cocoon predated on-tree, or unknown cause of larval or pupal death (Fig. 2). There have been no parasitoids documented in the Promethea Moth pupae on York Hill other than the 2 species cataloged here. We identified empty cocoons with a hole in the side as predated cocoons (Fig. 2h). Importantly, this dataset does not capture predation from predators, such as squirrels, that remove cocoons from the branch before consuming them (see Discussion); we refer to on-tree predation only. Finally, we counted dead caterpillars or pupae in 1 group classified as unknown (Figs. 2c and 2d) . In addition to documenting the contents, we determined the age of the cocoons from the 5 most-recent collection years (2007, 2012, 2014, 2015, and 2017) according to 2 criteria: (1) cocoons with live specimens were considered to be from the previous summer, whereas pupal cuticle or wasp remnants were indicative of older cocoons; and (2) comparison of the cocoon with vouchers of known age. This was not done for the first 2 collection years (1997, 2005), which are, therefore, an integration Figure 1. Examples of Promethea Moth cocoons collected at a 121-ha study site on York Hill, ME, in (a) Beaked Hazelnut and (b) Red Maple; (c) shows the scale of Promethea Moth cocoons in inches. Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 321 of the previous several broods because each year’s collection includes current- and previous-year specimens. Nevertheless, the length of the study period (20 y) allows for insight into population trends because the time span is longer than any potential short-term overlap in specimens collected from year to year. We performed statistical analyses by collection year so that we could utilize the entire dataset. We examined yearly success rates for both the Promethea Moth and its parasites with binary logistic regression. The dataset meets the assumptions of having a binary- coded outcome variable, independence of observations, and a large sample size (923 cocoons). In all models, the independent variable was time and the dependent (binary) variable was success (1) or death (0) of the Promethea Moth, and presence (1) or absence (0) of parasites according to cocoon contents. On-tree predation accounted for a small percentage of samples; thus, we did not include this category in our analyses. We treated time, expressed as collection year, as a categorical variable to allow for comparison between years. We set statistical significance below an a priori alpha level of 0.05, and considered P-values between 0.05–0.1 indicative of a trend but not significant. We compared data from all years to both 1997 and 2017 Figure 2. Examples of contents of Promethea Moth cocoons collected at a 121-ha study site on York Hill, ME: (a) Promethea Moth pupa, (b) Promethea Moth pupal cuticle shed during metamorphosis to adult moth, (c) and (d) Promethea Moths that died due to unknown causes, (e) E. americanus pupa, (f) G. nuncius larvae in cocoons (g) empty G. nuncius cocoons indicating successful adult emergence, and (h) Promethea Moth cocoon predated by a vertebrate. Northeastern Naturalist 322 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 to examine patterns in moth survival and parasitization across the study period. The data from 1997 differed significantly from the rest of the dataset; thus, we excluded it from the set of logistic regressions that compared all years with 2017 to examine trends without this apparent outlier year. We computed odds ratios to determine the likelihood of success for both the moth and its parasites in any given year compared to both 1997 and 2017 (1997 data excluded). We performed logistic regression and odds ratio analysis in SAS software (SAS Institute 2012) and linear regression of yearly percentages in JMP Pro (version 12.1.0, SAS Institute Inc., Cary, NC) to illustrate the overall trends across the study period. The dependent variables in this analysis were percent successful cocoons and percent parasitized cocoons each year, and the independent variable was time. Results We collected a total of 923 cocoons over the course of 7 sampling seasons spanning a period of 20 years (Table 1). We categorized 27% of the cocoons as successful for the Promethea Moth, with the large majority of these containing remnants of pupal cuticle (as opposed to live pupae) indicating prior moth emergence. Data from the cocoons for which we determined age (Table 2) reinforce this point; in any given year, we found many more cocoons that contained pupal cuticle—indicating successful moth eclosion—than those that contained live moth pupae. We found only 41 live pupae (4.4% of the sample) over the course of this study (1997 data not available). Causes of death for the Promethea Moth included, in order of prevalence: total parasitization by G. nuncius and E. americanus (59% of cocoons), death due to unknown diseases (11%), and on-tree predation (3.9%). Over the entire period, G. nuncius and E. americanus parasitized 50% and 9% of the cocoons collected, respectively. Linear regression results (Fig. 3) showed a significant decrease in the rate of successful Promethea Moth pupation over the study period both with 1997 included (R2 = 0.88, F1,5 = 35.44, P = 0.0019) and excluded (R2 = 0.69, F1,4 = 8.72, P = 0.04). Conversely, percent of cocoons parasitized significantly increased across the Table 1. Yearly categorization of Promethea Moth cocoon contents collected on York Hill, a 121-ha property in Maine, from 1997 to 2017. - indicate data that were not collected. Values are counts with percentages in parentheses. Successful cocoons = live pupae or eclosed. Parasitized Parasitized Unknown Total Live Successful by by cause of Year cocoons pupae cocoons G. nuncius E. americanus Predated death 1997 138 - 65 (47) 43 (31) 9 (7) - 21 (15) 2005 181 30 (17) 44 (24) 108 (60) 0 (0) 14 (8) 15 (8) 2007 269 0 (0) 72 (27) 154 (57) 18 (7) 7 (3) 18 (7) 2012 113 9 (8) 26 (23) 43 (38) 21 (19) 8 (7) 15 (13) 2014 95 1 (1) 17 (18) 40 (42) 12 (13) 3 (3) 23 (24) 2015 79 1 (1) 16 (20) 43 (54) 12 (15) 1 (1) 7 (9) 2017 48 0 (0) 5 (10) 31 (65) 7 (15) 3 (6) 2 (4) Total: 923 41 (4.4) 245 (26.5) 462 (50.1) 79 (8.6) 36 (3.9) 101 (10.9) Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 323 Table 2. Contents of Promethea Moth cocoons from 5 of the 7 years of this study according to age determination at time of collection from a 121-ha property on York Hill, ME. Values in parentheses indicate the number of wasp larvae out of the total that were dead or dying (this number was particularly relevant for G. nuncius in 2007, when the great majority of its larvae were killed by an undiagnosed pathogen). Asterisks (*) indicate grouped data for current- and previous-year cocoons (2007 collection only). Unknown pupation year indicates cocoons older than the previous year that could not reliably be aged. Successful Promethea Moth = live pupa or pupal shell indicating prior moth eclosion. Pupation Collection Year year Cocoon contents 2007 2012 2014 2015 2017 Current Successful Promethea Moth pupa 0 9 1 1 0 Parasitized by Gambrus nuncius 14 34 21 3 8 Parasitized by Enicospilus americanus 9 13 9 1 1 Cocoon predated on tree - 4 0 0 3 Dead pupa or larva - unkown cause - 5 5 2 1 Previous Successful Promethea Moth pupa 18 14 16 8 2 Parasitized by Gambrus nuncius 85* (85) 3 19 23 9 Parasitized by Enicospilus americanus 5 (3*) 6 3 11 (5) 2 Coccoon predated on tree 7* 3 0 1 0 Dead pupa or larva- unkown cause 13* 8 6 5 0 Unknown Successful Promethea Moth pupa 54 3 0 7 3 Parasitized by Gambrus nuncius 55 (55) 6 0 17 14 Parasitized by Enicospilus americanus 4 (2) 2 0 0 4 Cocoon predated on tree 0 1 3 0 0 Dead pupa or larva - unkown cause 5 2 12 0 1 Figure 3. Results of linear regression of percent successful Promethea Moth cocoons (closed circles, solid line, R2 = 0.88, F1,5 = 35.44, P = 0.0019) and total percent cocoons parasitized by G. nuncius and E. americanus (open circles, dashed line, R2 = 0.64, F1,5 = 8.73, P = 0.03) at a 121-ha study site on York Hill, ME. Removing 1997 only slightly changed the statistics for percent successful Promethea Moths (R2 = 0.69, F1,4 = 8.72, P = 0.04), but without 1997 the trend in percent cocoons parasitized was no longer significant (R2 = 0.24, F1,4 = 1.23, P = 0.33). Northeastern Naturalist 324 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 study period when 1997 data were included (R2 = 0.64, F1,5 = 8.73, P = 0.03), but no significant trend was found in the dataset that excluded 1997 (R2 = 0.24, F1,4 = 1.23, P = 0.33). Similarly, logistic regression models that included 1997 data indicated significant differences in the populations of both the Promethea Moth and its ichneumonid parasites from the beginning to the end of the study period (Table 3). Table 3. Results of logistic regression of successful Promethea Moth cocoons and parasitized cocoons collected from a 121-ha study site on York Hill, ME. Left column (All data) compares each year of data to 1997 as the beginning of the record. Right (1997 excluded) removes the 1997 data and compares each year to 2017, the end of the record. Model results are presented for (a and e) successful cocoons (live pupae and eclosed moths); (b and f) cocoons parasitized b y both G. nuncius and E. americanus; (c and g) cocoons parasitized by only G. nuncius; and (d and h) cocoons parasitized by only E. americanus. * indicate statistical significance (P < 0.05). Reported R2 values are the Nagelkerke R2, a pseudo R2 for logistic regression. All data 1997 excluded Variable Estimate SE P Estimate SE P Successful cocoons (a) (R2 = 0.56) (e) (R2 = 0.25) Intercept -0.1161 0.1705 0.4961 -2.1516 0.4725 less than 0.0001* 2005 -1.0197 0.2431 less than 0.0001* 1.0158 0.5032 0.0435* 2007 -0.8905 0.2192 less than 0.0001* 1.1451 0.4921 0.0200* 2012 -1.0917 0.2811 0.0001* 0.9438 0.5227 0.0710 2014 -1.4074 0.3174 less than 0.0001* 0.6281 0.5430 0.2474 2015 -1.2545 0.3278 0.0001* 0.7811 0.5492 0.1550 2017 -2.0353 0.5023 less than 0.0001* Total parasitized (b) (R2 = 0.55) (f) (R2 = 0.30) Intercept -0.5031 0.1757 0.0042* 1.3350 0.3554 0.0002* 2005 0.8948 0.2320 0.0001* -0.9433 0.3864 0.0146* 2007 1.0759 0.2168 less than 0.0001* -0.7622 0.3774 0.0434* 2012 0.7702 0.2586 0.0029* -1.0679 0.4029 0.0080* 2014 0.6931 0.2708 0.0105* -1.1450 0.4109 0.0053* 2015 1.3324 0.3012 less than 0.0001* -0.5057 0.4315 0.2412 2017 1.8381 0.3965 less than 0.0001* Parasitized by G. nuncius (c) (R2 = 0.57) (g) (R2 = 0.43) Intercept -0.7924 0.1838 less than 0.0001* 0.6007 0.3018 0.0465* 2005 1.1841 0.2382 less than 0.0001* -0.2091 0.3377 0.5358 2007 1.0844 0.2213 less than 0.0001* -0.3087 0.3260 0.3436 2012 0.3051 0.2671 0.2532 -1.0880 0.3586 0.0024* 2014 0.4739 0.2774 0.0876 -0.9192 0.3664 0.0121* 2015 0.9701 0.2912 0.0009* -0.4231 0.3770 0.2618 2017 1.3931 0.3534 less than 0.0001* Parasitized by E. americanus (d) (R2 = 0.40) (h) (R2 = 0.39) Intercept -2.6626 0.3448 less than 0.0001* -1.7677 0.4090 less than 0.0001* 2007 0.0275 0.4224 0.9481 -0.8674 0.4762 0.0685 2012 1.1853 0.4211 0.0049* 0.2904 0.4751 0.5411 2014 0.7286 0.4629 0.1154 -0.1663 0.5125 0.7456 2015 0.9428 0.4660 0.0430* 0.0479 0.5153 0.9260 2017 0.8949 0.5349 0.0943 Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 325 The highest incidence of cocoon success occurred in 1997, with all the following years significantly less successful for the Promethea Moth (Table 3a). Odds-ratio analysis showed that the likelihood of a successful cocoon was greatly reduced in years 2005–2017 when compared to 1997 (Fig. 4a). The total number of parasitized cocoons by G. nuncius and E. americanus combined exhibited the opposite trend, with 1997 having significantly less parasitization of cocoons than all following years included in this study (Table 3b). Odds-ratio analysis confirmed this, showing the likelihood of parasitization of cocoons to be greater in all years after 1997 when compared to the rates in 1997 (Fig. 5a). When examined separately, G. nuncius and E. americanus parasitization rates were somewhat more varied through time. Figure 4. Odds ratio of Promethea Moth success in cocoons collected from a 121-ha study site on York Hill, ME, in each year compared (a) to the first year of the study (1997) and (b) the last year of the study (2017) with 1997 data excluded. An odds ratio of 1 (dashed line) indicates equal likelihood of an event occurring in the year in question or in the comparison year (1997 or 2017 for panel a or b, respectively). An odds ratio < 1 indicates reduced likelihood of the event happening. Note the different axis ranges. Figure 5. Odds ratio of total parasitization by both G. nuncius and E. americanus in cocoons collected from a 121-ha study site on York Hill, ME, in each year compared (a) to the first year of the study (1997) and (b) the last year of the study (2017) with 1997 data excluded. An odds ratio of 1 (dashed line) indicates equal likelihood of an event occurring in the year in question or in the comparison year (1997 or 2017 for panel a or b, respectively). An odds ratio < 1 indicates reduced likelihood of the event happening. Note the different axis ranges. Northeastern Naturalist 326 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 The years 2005, 2007, 2015, and 2017 had significantly more parasitization by G. nuncius than occurred in 1997, but 2012 and 2014 were not statistically different than 1997, although 2014 exhibited a trend toward significantly more parasitization compared to 1997 (Table 3c). Enicospilus americanus exhibited significantly higher parasitization than in 1997 only in 2012 and 2015, with 2017 showing a trend toward higher parasitization (Table 3d). Logistic regressions that compared each year of the study to 2017, and excluded 1997 data, showed results similar to those that included 1997 data. Compared to 2017, 2005 and 2007 had a significantly greater percentage of successful cocoons, and 2012 exhibited the same trend but was not statistically significant (Table 3e). These findings were confirmed by odds-ratio analysis, which demonstrated that a moth was more likely to successfully pupate in 2005 and 2007 than in 2017 (Fig. 4b). Compared to all years except 2015, 2017 had significantly more parasitization of cocoons by both wasp species (Table 3f). That result was confirmed by odds-ratio analysis, which found that a cocoon was more likely to be parasitized in 2017 than in any of the study years from 2005 to 2014 (Fig. 5b). As with the dataset that included 1997, the trends were less clear-cut for G. nuncius and E. americanus separately. Parasitization by G. nuncius was significantly less in 2012 and 2014 than in 2017 (Table 3g), and there were no significant differences in parasitization between years by E. americanus, although 2007 exhibited a trend toward less parasitization (Table 3h). Discussion The results of this study identify parasitism as the predominant cause of mortality for overwintering Promethea Moth pupae at York Hill, ME. Across all the years of this study, we found a low success rate for the Promethea Moth (26.5%) as compared to prior studies that found over 80% survival for the moth (Waldbauer and Sternburg 1982). Our analysis found that 1997 had by far the highest rate of success for the Promethea Moth, and the lowest rate of parasitization. When we excluded 1997 data from the analysis, the earlier part of the record still had significantly higher success for the Promethea Moth and lower rates of parasitization than the latter years. The overwintering York Hill Promethea Moth population has been of interest since the mid-1980s when about half of several hundred cocoons examined contained live Promethea Moth pupae (Heinrich 2009). By 2006, there had been a steep decrease in the number of live moth pupae found, to less than 1% (Heinrich 2009). Unfortunately, the number of winter-collected cocoons containing pupal exoskeleton remains (evidence of previous-year moth eclosion) had not been counted, thus underestimating the percentage of live moths from previous years. During the time record reported here, where we included total eclosion of all collected pupae, moth success has continued to significantly decrease and total parasitization rates have increased. The rates of parasitization on York Hill differ greatly from previously collected comparable data. In east-central Illinois and northwestern Indiana, only 5.8% of examined cocoons were parasitized, as reported by Waldbauer and Sternburg (1982). Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 327 Even in 1997, when Promethea Moth success was near 50% on York Hill, the total parasitization rate was much higher (38%). It may therefore be the case that high rates of parasitization by these native parasitoid wasps is normal for the York Hill study area. It is also worth noting that although we found a signficant decline in the percent of successful cocoons collected throughout the study period, both the Promethea Moth and the 2 parasitic wasps appear to be persisting at this site near the northern limit of the moth’s range. The data also reveal what appears to be the collapse and recovery of a parasitoid population. In 2007, 86% of G. nuncius larvae examined were dead or dying of an unknown cause (Table 2). Interestingly, the 2 following collection years (2012 and 2014) saw reduced rates of parasitism by this species compared to prior and subsequent years (Table 1). Furthermore, the parasitization rates these years (2012 and 2014) did not differ signficantly from 1997 (Table 3). On-tree predation accounted for the smallest percentage (3.9%) of Promethea Moth death over the study period. These results are consistent with findings by Waldbauer and Sternburg (1982) showing that 3.2% of cocoons were predated by woodpeckers and 0.5% by mice, for a total of 3.7%. They attribute these low rates to the high placement and flexible attachment of the cocoons, which provide protection from mice and birds, respectively (Waldbauer and Sternburg 1982). Due to our collection methodology, our results account for predation by birds because they either extract the pupa vertically through the cocoon valve, or peck holes into the side; in both instances the cocoon is left hanging on the tree (Nielsen 1977, Waldbauer et al. 1970). On the contrary, squirrels have been observed chewing cocoons off the branch (Tamiasciurus hudsonicus (Erxleben) [Red Squirrel; Heinrich 2016]) or knocking them to the ground (Sciurus carolinensis Gmelin [Eastern Gray Squirrel; Young 1982]) before consuming the contents. Due to our methodology, such off-tree predation was not accounted for, and so actual cocoon predation rates at York Hill could therefore be significantly higher than our dataset illustrates. Interestingly, woodpeckers have been observed to select healthy Promethea Moth pupae to predate while leaving behind sick, damaged, or parasitized cocoons, and Eastern Gray Squirrels have been observed to do the same with other species of giant silk moths (Nielsen 1977, Waldbauer et al. 1970, Young 1982). The high rates of parasitization and low incidence of successful eclosion and live pupae at our study site may therefore explain the low incidence of predation by birds in our results. Furthermore, the effects of selective predation of healthy pupae may be magnified when combined with an already heavily parasitized moth population. Examination of aged cocoons revealed a consistent trend toward finding many more examples of moth pupal cuticle than live pupae in any given year (Table 2). This finding has several possible explanations. First, we may be more effective at finding cocoons from previous years because they lose their camouflage over time. However, if there were such a sampling bias, the pattern should be consistent regardless of the contents of the cocoon (moth pupa, wasp pupa, etc.), and this is not the case (Table 2). An alternate but related possibility is that successful cocoons for Northeastern Naturalist 328 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 the Promethea Moth have better camouflage, and are thus less likely to be found the year of pupation than those that are parasitized by G. nuncius, which inserts its eggs through the soft silk of the cocoon. However, if G. nuncius follows olfactory cues to find the cocoons (Marsh 1937), this explanation, too, seems unlikely. Finally, the high incidence of pupal cuticles could indicate that York Hill Promethea Moths are bivoltine, or partially so. Understood to be univoltine in the northern part of its range, the Promethea Moth has responded to the longer, warmer summers caused by climate change by adopting a facultative second generation in Massachussetts (Wagner 2012). Historically, partial double broods have been reported as far north as Meaford, ON, Canada (44°36'21''N, 80°35'37''W; Riotte 1992), which is similar to York Hill in its latitude and northern position within the Promethea Moth’s range. We collected our data under the assumption that the York Hill Promethea Moth population is univoltine; thus, we considered presence of pupal cuticle in a cocoon to be diagnostic of a previous year specimen. Further study would be necessary to provide evidence of a second generation at York Hill. The high rates of parasitism and low rates of Promethea Moth eclosion documented at York Hill point to a population that is persisting on a knife’s edge. Parasite–host interactions are of great importance to the persistence of Lepidoptera. For this Order of insects, overall mortality rates from hatching until adulthood normally range from 95% to 99% (Wagner 2012), leaving a tight margin for fluctuations if a population is to remain viable. In a population such as that at York Hill, the introduction of new parasitoids could stress the moth population beyond sustainable levels. For example, silkmoth declines have been attributed to introduction of a non-native generalist parasitoid fly, Compsilura concinnata, that attacks larvae. It was repeatedly introduced to North America between 1906 and 1986 as a biological control agent to target Lymantria dispar dispar (L.) (Gypsy Moth; Boettner et al. 2000). Precipitous declines in Promethea Moth populations were observed as early as 1919 in fly-release areas (Culver 1919), and field trials have shown up to 70% larval mortality due to C. concinnata in only 1 week, a much shorter period than that needed for larval development (Boettner et al. 2000). The addition of such a parasitoid at York Hill could decimate the Promethea Moth population, which is currently persisting at very high levels of native parasitism. However, saturniid populations can endure at relatively low densities because of their life-history strategies (Tuskes et al. 1996). For example, Promethea Moth males can detect females and are capable of dispersing over substantial distances (Weast 1959), with flights up to 32.2 km (20 mi) recorded under favorable wind conditions (Toliver and Jeffords 1981). Additionally, females lay eggs over several nights in different locations thereby distributing the population over a wider area. These life-history traits combined with the observed rates of parasitism may help explain Promethea Moth population dynamics at York Hill. Acknowledgments We thank all of the students who participated in this project through their enrollment in Winter Ecology at the University of Vermont. In particular, we acknowledge the Northeastern Naturalist Vol. 24, No. 3 S.M. Juice and B. Heinrich 2017 329 contributions of Sarah Lynch, Mark Ward, Michelle Brown, and Alexandra Kosiba. We greatly appreciate the assistance of Alan Howard at the Statistical Consulting Clinic at the University of Vermont. We also thank an anonymous reviewer for helpful comments on an earlier version of this manuscript. Literature Cited Boettner, G.H., J.S. Elkinton, and C.J. Boettner. 2000. Effects of a biological control introduction on three nontarget native species of saturniid moths. Conservation Biology 14:1798–1806. Culver, J. 1919. A study of Compsilura concinnata, an imported tachinid parasite of the Gypsy Moth and the Brown-tail Moth. Bulletin 766. US Department of Agriculture, Washington, DC. Ferguson, D.C. 1971. The Moths of America North of Mexico, fasc. 20.2, Bombycoidea (Saturniidae). E.W. Classey Limited and R.B.D. Publications, Inc., London, UK. 277 pp. Frank, K.D. 1988. Impact of outdoor lighting on moths: An assessment. Journal of the Lepidopterists’ Society 42:63–93. Heinrich, B. 2009. Summer World: A Season of Bounty. Harper Collins, New York, NY. 253 pp. Heinrich, B. 2016. Squirrel smarts: Tapping into sought-after food supplies. Natural History 124:12–15. Hessel, S. 1976. A preliminary scan of rare and endangered nearctic moths. Atala 4:19–20. Holden, C. 1992. Random samples: Thousands of insects enroll at Yale. Science 256:313. Marsh, F.L. 1937. Ecological observations upon the enemies of cecropia, with particular reference to its hymenopterous parasites. Ecology 18:106–112. Muller, J. 1979. Fourth addition to the supplemental list of Macrolepidoptera of New Jersey. Journal of the Lepidopterists’ Society 33:174–178. Nielsen, M. 1977. Woodpecker feeding on Callosamia promethea (Saturniidae) cocoon. Journal of the Lepidopterists’ Society 31:148–149. Peigler, R. 1976. Observations on host-plant relationships and larval nutrition in Callosamia (Saturniidae). Journal of the Lepidopterists’ Society 30:184–187. Peigler, R. 1977. Parasitism of Callosamia. Journal of the Georgia Entomological Society 12:111–114. Peigler, R. 1994. Catalogue of parasitoids of Saturniidae of the world. Journal of Research on the Lepidoptera 33:1–121. Pyle, R., M. Bentzien, and P. Opler. 1981. Insect conservation. Annual Review of Entomology 26:233–258. Riotte, J. 1992. Annotated List of Ontario Lepidoptera. Miscellaneous Publications in the Life Sciences. Royal Ontario Museum, Toronto, ON, Canada. SAS Institute. 2012. SAS, Version 9.4. Cary, NC. Schweitzer, D. 1988. Status of Saturniidae in the northeastern USA: A quick review. News of the Lepidopterists’ Society 1:4–5. Scriber, J.M., J. Potter, and K. Johnson. 1991. Lack of physiological improvement in performance of Callosamia promethea larvae on local host-plant favorites. Oecologia 86:232–235. Toliver, M.E., and M.R. Jeffords. 1981. Long-distance dipsersal by Callosamia promethea (Saturniidae). Journal of the Lepidopterists’ Society 35:76. Tuskes, P.M., J.P. Tuttle, and M.M. Collins. 1996. The Wild Silk Moths of North America: A Natural History of the Saturniidae of the United States and Canada. Cornell University Press, Ithaca, N.Y. 250 pp. Northeastern Naturalist 330 S.M. Juice and B. Heinrich 2017 Vol. 24, No. 3 Wagner, D. 2012. Moth decline in the northeastern United States. News of the Lepidopterists’ Society 54:52–56. Wagner, W.H., M.K. Hansen, and M.R. Mayfield. 1981. True and false foodplants of Callosamia promethea (Lepidoptera: Saturniidae) in Southern Michigan. Great Lakes Entomologist 14:159–165. Waldbauer, G., and J. Sternburg. 1982. Cocoons of Callosamia promethea (Saturniidae): Adaptive significance of differences in mode of attachment to the host tree. Journal of the Lepidopterists’ Society 36:192–199. Waldbauer, G., J. Sternburg, W. George, and A. Scarbrough. 1970. Hairy and Downy Woodpecker attacks on cocoons of urban Hyalophora cecropia and other saturniids (Lepidoptera). Annals of the Entomological Society of America 63:1366–1369. Weast, R. 1959. Isolation mechanisms in populations of Hyalophora (Saturniidae). Journal of the Lepidopterists’ Society 13:213–216. Worth, C., and J. Muller. 1979. Captures of large moths by an ultraviolet light trap. Journal of the Lepidopterists’ Society 33:261–264. Young, A.M. 1982. Predation on the pupae of Saturniidae (Lepidoptera) by Gray Squirrels in Wisconsin. Great Lakes Entomologist 15:145–145.