# Environmental Entomology

### Editor-in-Chief

E. Alan Cameron

1.315
1.597
36 out of 94

Published on behalf of

## Predation of Gypsy Moth (Lepidoptera: Lymantriidae) Pupae in Three Ecosystems Along the Southern Edge of Infestation

F. L. Hastings, F. P. Hain, H. R. Smith, S. P. Cook, J. F. Monahan
668-675 First published online: 1 August 2002

## Abstract

The predation potential of small mammals, in particular mice, Peromyscus spp., and invertebrates, was evaluated from 1992 to 1995 near the leading edge of gypsy moth, Lymantria dispar (L.), spread into the southeastern United States. Two study sites were established in each of three geographic areas: the coastal plain, Piedmont, and mountains. All sites were mixed hardwood stands with varying amounts of oak, Quercus spp., and all were classified for gypsy moth susceptibility. Small mammal density was estimated using Sherman live-traps and pitfall traps within these 4.68-ha sites in early and late summer. Each site contained 75 trapping stations located on a 25-m grid. Predation was measured by offering freeze-dried gypsy moth pupae near trapping stations at four heights (0, 0.25, 1.0, and 2.0 m) on different tree boles. Pupal predation was monitored for three consecutive nights. Vertebrate predation was positively correlated with good mast production in the previous autumn. Predation data showed that when mice were at high densities they were the major source of pupal predation. However, within these southern sites, when densities of Peromyscus spp. were low, predation by invertebrates was occasionally greater than predation by vertebrates. These data suggest that in some years invertebrates may retard gypsy moth buildup when small mammals are scarce due to mast crop failures.

• Lymantria dispar
• Peromyscus
• predation
• mast

Small mammals were first recognized as important predators of gypsy moth by Bess et al. 1947; later studies (Bess 1961; Campbell and Sloan 1976,1977) indicated that predation by the white-footed mouse (Peromyscus leucopus Rafinesque) was the dominant cause of pupal mortality at low population densities. More recent studies in northeastern United States have reaffirmed the white-footed mouse as the most important predator of endemic populations of gypsy moth (Smith 1985,1989; Elkinton et al. 1989; Yahner and Smith 1991; Elkinton and Liebhold 1992). In a 10-yr study where gypsy moth was at low population densities (Elkinton et al. 1996), increases in gypsy moth density were directly linked to declines in mouse density and mouse density was positively correlated with acorn density. The removal of the white-footed mouse from experimental grids near Millbrook, NY, the year following a large red-oak acorn crop, caused an endemic population of gypsy moth to increase; within control grids, no increase in gypsy moth density occurred (Jones et al. 1998). Predation of female pupae was measured for the native gypsy moth population and this was compared with predation of freeze-dried female gypsy moth pupae. Data showed a close correlation between these two types of offerings with respect to time and level of mouse predation.

The movement of the gypsy moth into southern forests creates visions of massive tree defoliation and unhappy homeowners, campers, picnickers and forest managers. Whether this outlook is warranted will depend on the response of host plants and reaction of local fauna (predators, parasitoids and pathogens) toward this potential new food resource. The abundance of hardwood species, in particular oaks (Quercus spp.), in these forests will generally sustain moderate to high densities of small mammals; however, reductions in mast crops may limit small mammal populations. Mast crop losses may result from a combination of weather, attacks by insects or diseases (Beck 1993) or from the inheritable tendencies of oaks to produce acorns with extreme inter-annual variation (Silvertown 1980; Sork et al.1993). Changes in the impact of natural enemies associated with gypsy moths should be key to understanding whether modification of forest management practices are necessary as the insect moves into climatically moderate southern ecosystems. The objective of this study was to evaluate the potential roles of predators within selected forest ecosystems along the southern edge of gypsy moth infestation.

## Materials and Methods

Six permanent sites were chosen in 1991, two each in the coastal plain (Currituck County, NC [W 76° 00′ 00″, N 36° 24′ 00″] and Northampton County, VA [W 75° 49′ 36″, N 37° 30′ 21″]), Piedmont, Lake Anna State Park, Spotsylvania County, VA (W 77° 49′ 13″, N 38° 07′ 30; W 77° 50′ 00″, N 38° 07′ 09″) and mountain, George Washington National Forest, Amherst County, VA (W 79° 09′ 29″, N 37° 45′ 56″; W 79° 11′ 23″, N 37° 45′ 24″). The two sites in the Piedmont and mountains were segregated according to altitude (upper and lower slopes). Differences were ≈7.6 m in the Piedmont and ≈61 m in the mountains. The Northampton County site and the two sites in George Washington National Forest (GWNF) had sustained gypsy moth infestations during the last 3 yr; the other sites had no known gypsy moth infestation (Cook et al. 1995). The infested sites sustained partial defoliation on a few trees. The six 4.68-ha sites (75 by 625 m) were grids of three parallel lines with each line having 25 trapping stations. Trap stations were 25 m apart and lines were 25 m apart. The center of each trapping station was an open pitfall trap (plastic buckets ≈20 cm in diameter by ≈20 cm deep) with a 20.3 by 152.4-cm piece of stainless steel flashing "drift-fencing" on either side and perpendicular to the slope of the land. Small mammal abundance was determined twice each year, early (May-June) and late (August-September) in 1992, 1993, and 1994 with the exception of Lake Anna that was surveyed only once in 1994. The single survey was due to a lack of manpower and this was true for 1995 in which we again only surveyed once at each site. When two surveys were done the first was done by opening 75 unbaited pitfall traps and 75 Sherman live-traps (H.B. Sherman Traps, Tallahassee, FL., 7.5 by 7.5 by 40 cm) near the center of each station. We only opened 50 pitfall traps and 50 live-traps for the second survey. Traps were monitored for three consecutive nights. Live-traps were baited with a mixture of peanut butter and oatmeal, with cotton provided for bedding. All captured small mammals were identified to species. Mice were uniquely marked by toe clipping and released at the point of capture; other small mammals were recorded and released unmarked. Trapping data were expressed following Brooks et al. (1998) on a 100 trap night basis (one trap set for one night = one trap night).

In areas where populations of two or more species of the genus Peromyscus overlap they are difficult to identify to species (Webster et al. 1985). Populations of P. leucopus and P. maniculatus (Wagner) are known to overlap in the Appalachian Mountains and we judged all to be P. leucopus. Because Peromyscus spp. are characterized as ground foraging generalists (Eisenberg 1968), any error in misidentification that might have occurred would not affect the results of this study.

Between small mammal surveys or within 1 wk of the single surveys, we measured potential gypsy moth predation by offering freeze-dried female gypsy moth pupae (USDA APHIS, Otis Method Development Center, Otis ANGB, MA) near the center of each trapping station. In 1994, we repeated the pupal offering in the GWNF due to the large mast crop which remained on the forest floor. Freeze-dried pupae were used because these studies were within quarantined zones that prevented the use of live pupae. This is a proven technique for measuring predation by vertebrates and invertebrates (Smith et al. 1993, Jones et al. 1998). Single pupa were attached to small squares of burlap using melted beeswax (Smith 1985) and stapled to different trees at four heights (0, 0.25, 1.0, and 2.0 m). Pupal predation was monitored each morning for three consecutive days. The identity of predators was based on the pattern of damage to pupae. Small mammals either consume or remove the entire pupa or leave large pupal fragments with ragged edges caused by their incisors. Invertebrate feeding sign was distinguished by the finely serrated edges left in the pupal integument (Smith and Lautenschlager 1981). Daily monitoring is necessary to identify the predator and to eliminate the problem of scavengers from feeding on pupal remains.

In the southeastern United States, 1/5 of the trapping stations at each site (15 of 75) was classified for susceptibility to gypsy moth defoliation in 1991 using discriminant variables (Houston and Valentine 1985). Calculations were made according to the Houston-Valentine equation: $$mathtex$$Y = 14.799 - 0.4245x_1 /x_2 - 0.004962x_3 - 0.005376x_4$$mathtex$$

Positive values for Y predict resistance and negative values susceptibility. The terms; x1 = sum of diameters of preferred host trees per 0.4 ha; x2 = basal area of preferred host trees per 0.4 ha; x3 = sum of diameters per 0.4 ha of preferred host trees with one or more deep fissures; x4 = sum of diameters per 0.4 ha of preferred host trees with one or more bark flaps.

Vertebrate and invertebrate predation were compared between the southeastern and northeastern U.S. sites. Northeastern sites (Vermont) were also classified for gypsy moth susceptibility (Houston and Valentine 1985). Because these forests are generally infested with gypsy moth, live pupae rather than freeze-dried pupae were offered at the base of the tree bole and 2 m above the base (on the bole or within a bark flap). Pupae were observed daily until they were eaten or adults emerged (≈14 d).

Additional descriptions of the understory and overstory vegetation were obtained in 1993 by using the Line-Point Interception Method (Canfield 1941) and the point-centered quarter method (Cottam and Curtis 1956, respectively. The understory near every fifth trapping station was sampled on 15.24-m lines from randomly chosen compass headings. The number and species of all vegetation were recorded that touched a 1-m stick that was placed vertically at 10 cm intervals along the line. The overstory was determined by measuring the distance of the nearest tree in each quadrant from the centers of every fifth trapping station, in each site. The species and basal area were recorded and the relative density and relative dominance determined by the formulae:

Relative density = number of individuals of the species/number of individuals of all species × 100; Relative dominance = total basal area of the species/total basal area of all species × 100

Because small mammal population density is dependent on the availability of mast, we measured nut and acorn production within all sites during this 4-yr study. Within each site we choose ≈42 nut-producing trees (Quercus and Carya) based on the evidence of earlier nut production, and placed two 18-liter buckets under their canopies. Buckets were suspended on steel rods at heights that prevented white-tailed deer (Odocoileus virginianus Boddaert) from removing the nuts. Buckets were fitted with a piece of wire-mesh that served to retain the falling nuts by reducing their impact velocity. Buckets were placed in the forest before nut-drop and removed in December. Sound nuts were tallied and the productivity of each site was compared on a scale of 1-5 (1 = lowest, 5 = highest nut production). Mouse density was then related to mast production.

Linear regression analyses were used to determine the probability of predation at different heights along the tree bole. If we assume the probability of a pupae being preyed upon overnight is a constant across three nights, at a given elevation, location, year, and height, then the maximum likelihood estimates, f, for this probability were calculated after the formula: $$mathtex$$f = (n1 + n2 + n3)/n1 + 2*n2 + 3*n3 + 3*n4)$$mathtex$$ where n1 = number taken on day 1, n2 = number taken on day 2, n3 = number taken on day 3. Analysis of the nightly predation probability f allows for comparisons to other studies and other sampling methods. Linear regression analyses were also used to examine the relationships between vertebrate predation and nut catch, mast crop (current and following year) and vertebrate populations.

## Results and Discussion

The Houston-Valentine classification of stand susceptibility showed few negative scores suggesting that all stands were resistant to defoliation (Table 1). Trapping stations having the most negative scores (Currituck County [6 of 15] and the upper slope at Lake Anna [3 of 15]) were not infested, but Northampton County and the two mountain sites each of which had only one negative score, did sustain some partial defoliation during the study. The summary data used to calculate stand susceptibility (Houston and Valentine 1985) indicated that trees in the mountains and Piedmont had almost no bark fissures or bark flaps, and few gypsy moth refuges (snags, logs and stumps) whereas these were common in the coastal plain. Litter depth increased from higher to lower elevations (1.9 cm, mountain; 3.3 cm Piedmont; 5.5 cm coastal plain) reflecting differences between these ecosystems. These results suggest that the Houston-Valentine classification, which was developed within the northeastern United States, may be applicable in the southeastern United States.

View this table:

Each region had a characteristic species that was one of the dominant components of the understory providing cover and/or food for mice. (Table 2). These were Clethra alnifolia in the coastal plain, Vaccinium vacillans in the Piedmont, Dennstaedtia punctilobula in the mountains. This survey revealed a number of berries, (Gaylussiacia and Vaccinium spp.) within the coastal plain and Piedmont, respectively, which could serve as food resources for the white-footed mouse (Webster et al. 1985). The overstory vegetation survey indicated that all sites were mixed hardwoods with varying amounts of oak; oak dominance was ≈20% in the coastal plain and ≈40% in the Piedmont (Table 3). However, the upper slope of the mountains had ≈30% oak dominance but oak was not represented within the top five of dominant species in the lower slope. This information supports the Houston-Valentine scores (Table 1) which indicated that oak was not dominant at any site (negative scores being generally related to oak density).

View this table:
View this table:

Eight hundred and thirty-three Peromyscus spp. were trapped during this 4-yr study, 113 were recaptured the second night and 207 were recaptured the third night. The number of newly captured mice decreased 24% on night 2 and 36% on night 3. This information suggests that within our trapping sites we had a good measure of the vertebrates preying on freeze-dried pupae. Table 4 expresses mouse catch in terms of catch per 100 trap nights (one set trap for one night = one trap night). It is obvious that the Currituck County site never had substantial populations of Peromyscus spp. and at times mice were quite scarce in other sites (Lake Anna sites 1993). We also trapped 29 other small mammals that are believed to feed on gypsy moth. Fifteen were northern short-tailed shrews (Blarina brevicauda Say), one masked shrew (Sorex cinerus Kerr), eight southern red-backed voles (Clethrionmys gapperi Vigors), three woodland jumping mice (Napaeozapus insignis Miller), two eastern harvest mice (Reithrodontomys humuli s Audubon and Bachman). Although each of these animals contribute to gypsy moth consumption, they constitute <3.5% of the captured small mammals. We also trapped one opossum, Didelphis virginiana Kerr., and two striped skunks, Mephitis mephitis Schreber. Shrews were the only mammals caught in pitfall traps.

View this table:

The percentage of freeze-dried gypsy moth pupae that were preyed upon by vertebrates and invertebrates at each site during the 3 d exposures for each of the 4 yr is shown in Table 5. In the coastal plain small mammals remained at low population densities during the 4-yr study (Table 4). The lower oak component in these coastal sites (Table 3) could account for the lack of small mammals (McCracken et al. 1999). Within these sites predation of freeze-dried pupae was almost equally divided between small mammals and invertebrates (≈30% for each) in each year (Table 5). By contrast, within the Piedmont and mountains, predation was more variable. Highest mouse populations were found in the mountain (upper slope) in 1992 and 1994, this site had the highest component of mast producing trees (oak and hickory). When vertebrates were the dominant predators (1992, 1995), invertebrates accounted for only a fraction of the predation. However, invertebrates took more prey than vertebrates during one-third of the pupal exposure periods. The 1994 Lake Anna data underscore the importance of invertebrate predation because when combined with vertebrate predation, the percentages of uneaten prey were only 0.7 and 3.7, possibly enough to keep gypsy moth populations in check. The contribution made by invertebrates is noteworthy when considering the short exposure period (3 d). Data show that gypsy moth emergence in northeastern forests of Vermont approached zero when vertebrates preyed on ≈ 90% of the pupae (Table 6). These data are typical of more northerly forests where gypsy moth predation is primarily by vertebrates whereas invertebrates have been shown to be insignificant as predators (Smith 1989). By contrast, data from the southeastern United States are unique in that invertebrates are capable of much higher predation across diverse ecosystems. Although this predation was variable, the invertebrate attack rates were higher than expected. Ants were the most important invertebrate predators observed during these 3-d exposures. The magnitude of invertebrate predation in these diverse southern forests may suggest a faunal richness greater than in more northern climes.

View this table:
View this table:

A composite of all predation curves (n = 24) for freeze-dried gypsy moth pupae which were placed on tree boles at 0.0, 0.25, 1.0, and 2.0 m is shown in Fig. 1A. The predation curves of sites where vertebrates were the dominant predators (Lake Anna, lower slope, 1995) and where invertebrates were the dominant predators (Lake Anna, lower slope, 1993) are shown in Fig. 1B and C. When vertebrates are at moderate to high density the rate of predation is high, particularly near the ground. Even in years when small mammals were sparse, or when small mammals fed on alternative food sources such as berries, it appears that the activity of invertebrates did keep predation rates relatively high. In Fig. 1 it is shown that location of prey affects the amount of predation. Food selection by small mammals is strongly effected by the inherent risk associated with the capture of the prey item. The lack of cover at 2 m (much more open) would increase risk thus increasing survivorship of the prey. Note that there is significant differences in predation probability across nights since the curves are not straight lines; this predation probability varied only from 0.30 to 0.39 ± 0.03.

Fig. 1.

Predation curves for the consumption of freeze-dried pupae during this 4-yr study. (A) A composite of all predation curves n = 24. (B) A site where vertebrates were the predominate predator (Lake Anna, lower slope, 1995). (C) A site where invertebrates were the predominate predator (Lake Anna, lower slope, 1993).

Regressing the independent variables (elevation [mountain, piedmont, coastal plain], upper and lower slopes, year, and four heights [0, 0.25, 1.0, and 2.0 m]) against f (daily predation probability) indicate that pupal height on the tree bole was highly significant (F = 111.46 Pr > F = 0.0001, n = 96. The probabilities of predation per day were 0.657 ± 0.044, 0.381 ± 0.047, 0.278 ± 0.037 and 0.247 ± 0.037 at heights of 0, 0.25, 1.0, and 2.0 m, respectively. These calculations did not distinguish between vertebrate and invertebrate predation. The effect of mouse catch density is difficult to assess since the mouse catch is also strongly affected by elevation and year (Table 4). When added to the regression above as an independent variable, catch alone (df = 1) accounts for >2/3 of the variation in predation rate explained by catch, elevation and year (df = 12) together. Dropping elevation and year, and fitting a quadratic model in height did not explain as much variation in predation rate (R2 = 0.62 versus R2 = 0.91 with catch, elevation, year), but gave a reasonable decreasing rate of predation with height and a positive coefficient with vertebrate catch (coefficient = 0.0072 ± 0.00097). The dependent variable nightly vertebrate predation (number of pupae eaten by a vertebrate/total number eaten) was then regressed against height, time (night), catch, elevation and year. This calculation resulted in negative coefficient estimates for height and time (height, coefficient = -0.051 ± 0.0088; time, coefficient = -0.081 ± 0.0092, n = 4554); thus, as time of pupal exposure increases and as height of pupa increases, there is less likelihood that predation was by a mouse. The implication of these findings would be that invertebrates are more likely to feed on pupae which are placed higher on the tree bole, and later predation would be more likely attributed to invertebrates. This pattern of predation is in agreement with that observed for vertebrates and invertebrates in Connecticut (Smith 1989).

To establish the importance of mast on predation, vertebrate predation was again treated as the dependent variable and regressed against height, time, catch and mast score (1-5 scale) from either the current or the previous year's mast. The probability of being eaten by a vertebrate was reduced if mast production was high within the same year (mast score coefficient = -0.149 ± 0.0069, n = 3217, t = -21.74). By contrast, the same probability increased when the mast crop was high for the previous year (mast score coefficient = 0.1005 ± 0.0096, n = 3519, t = 10.41). These data are in agreement with others (Batzli 1977, Hansen and Batzli 1979, Wolff 1985, Elkinton et al. 1996, Ostfeld et al. 1996, McCracken et al. 1999) who demonstrated the importance of autumnal mast production in maintaining mouse density in the next year.

In 1994, pupal predation (early June) in the GWNF was extremely low even though trapping data indicated high populations of small mammals (Table 4). At that time mast, a preferred food (acorns and hickory nuts) was noticeably abundant on the ground. To examine the effect of this mast on predation of gypsy moth pupae, we repeated the study in early August when mast was no longer evident. The results were ambiguous because although total predation increased dramatically (from 36 to 73% on the lower slope and from 30 to 81% on the upper slope) vertebrate predation remained the same within the lower slope, but increased three-fold within the upper slope. The confounding factor appeared to be the prominent increase in predation by invertebrates (four-fold increase within the lower slope and three-fold within the upper slope). Thus, the influence of an over-abundance of mast on food selection by Peromyscus spp. may not be a straightforward proposition. These results suggest that understanding the variation in both invertebrate and vertebrate predation could contribute to a better understanding of forest pest management strategies in similar sites.

We initiated this study to gather information regarding the role of predators of the gypsy moth along the leading edge of its southward movement. Information accruing over the last 20 yr has shown that small mammal predators (in particular, Peromyscus leucopus) limit populations of gypsy moth at low population densities within northeastern United States. The technique of offering freeze-dried female gypsy moth pupae (Smith 1985, Jones et al. 1998) allowed us to measure the respective contributions of vertebrate and invertebrate predators in areas where we could not introduce live pupae. In New York, it was found that the magnitude of predation was similar when freeze-dried and live pupae were compared (Jones et al. 1998). Our data in these six southern sites show predation by invertebrates was equal to predation by vertebrates at 10 of 24 locations from 1992 to 1995 (Table 5). During 1994, within the Piedmont, invertebrates accounted for ≈58% of the total predation resulting in 99.3 and 96.3% predation.

In summary, as the gypsy moth invades southern forests it appears that mortality by ground-foraging generalist small mammals will remain the key to effective predation maintenance within sparse stable populations. However, the predatory role and potential of invertebrates in southern forests also appear to be much greater than in the northeastern United States. We do not know whether this is a result of greater species diversity, greater density of the same species found in more northern climes, or perhaps more aggressive species. Regardless of the nature of this predation, the likelihood that gypsy moth will reach outbreak status in southern forests could be influenced by the combination of vertebrate and invertebrate predation. These data suggest that within certain years and localities of low vertebrate density, invertebrates may contribute enough mortality to delay an outbreak of gypsy moth. Obviously, this is transitory and unless small mammal predators recover quickly, gypsy moth outbreaks may occur. Our results suggest that continued research is necessary to ascertain the effect of invertebrate predation on gypsy moth in these southeastern forests. Additional studies on the relationship of other natural enemies, such as parasitoids would give a more complete understanding of the factors contributing to the regulation of this exotic moth in southern ecosystems.

## Acknowledgments

We thank W. Hobbs, C. F. Johnson, J. Ludwick, and R. Ross for technical support; C. E. Franklin for suggestions regarding mast sampling; and D. Orr, D. Robison, and R. Stinner for reviewing drafts of the manuscript. This study was partially supported by a Cooperative Research Agreement Supplement No. 196 to Contract No. A8fs-20, 147. USDA Forest Service Southeastern Forest Experiment Station and by the Gypsy Moth Research and Development Program, USDA Forest Service Northeastern Forest Experiment Station.