EFFECTS OF BOT FLY (CUTEREBRA FONTINELLA) PARASITISM ON A POPULATION OF WHITE-FOOTED MICE (PEROMYSCUS LEUCOPUS)

Transcription

1 Journal of Mammalogy, 87(6): , 2006 EFFECTS OF BOT FLY (CUTEREBRA FONTINELLA) PARASITISM ON A POPULATION OF WHITE-FOOTED MICE (PEROMYSCUS LEUCOPUS) MICHAEL J. CRAMER* AND GUY N. CAMERON Department of Biological Sciences, University of Cincinnati, Cincinnati, OH , USA Bot flies are common parasites of Peromyscus leucopus, although determination of a cost to the host has been elusive. The goal of this study was to further explore the potential costs of bot fly parasites for a population of P. leucopus. We investigated the effects of parasitism on host condition (mass after controlling for parasite mass and host body length) and survivorship (the number of days animals persisted on trapping grids). Parasitism was quantified by prevalence (proportion of the population infected), intensity (the number of parasites per infected host), and dispersion of parasites within hosts (clumped, regular, or random). In addition, we searched for spatial and temporal patterns in infection. Finally, we analyzed the relationship between population demography and parasitism. Contrary to expectations, we found that infected mice persisted longer on trapping grids and were in better condition than uninfected mice. Also, we discovered that when considering overall infection levels, parasites were clumped within hosts, but when considering the number of simultaneous infections, parasites were randomly distributed among hosts. Although most animals had single infections, there was a high incidence of reinfections, leading to bimodal patterns of parasitism. Prevalence was not correlated with host density, sex ratio, or proportion reproductive, but there were significant relationships between intensity and density and sex ratio in 1 year. In addition, prevalence and proportion of reproductively active animals were asynchronous. These results suggest that bot flies do not impose an obvious cost to their hosts, and hosts may express some degree of tolerance for bot fly parasitism. Key words: bot fly, condition, parasite dispersion, parasitism, Peromyscus leucopus Parasitism is a pervasive and important interaction between 2 species, in which one species benefits at the cost to the other. Because of the close association of parasites and their hosts in space and time, parasites potentially affect many aspects of host biology, including individual physiology, behavior, and population demography. The bot fly Cuterebra fontinella is an obligate parasite of the white-footed mouse (Peromyscus leucopus), a common inhabitant of eastern deciduous forests, that relies on mouse hosts to complete its life cycle (Catts 1982). Gravid bot flies lay eggs in host habitat (Dalmat 1943), and egg hatching is triggered by an increase in temperature and humidity, usually indicating proximity to a potential host (Catts 1982). The larva enters the host through the nostril, mouth, or an open wound, and migrates to the inguinal region of the host (Cogley 1991), where it forms a large swelling, or warble (Miller and Getz 1969). The period of development within the host lasts approximately 1 month (Cogley 1991). * Correspondent: cramerm@ .uc.edu Ó 2006 American Society of Mammalogists Because a developing larva can represent up to 5% of host weight, many researchers have assumed that bot flies impose a cost on their hosts. However, such a cost has been difficult to document. For example, studies of reproductive physiology of infected animals failed to detect a reproductive cost. Although reproductive organs of infected P. leucopus had less mass (Smith 1977b), infected males had similar seminal vesicles and epididymides and infected females had the same number of corpora lutea as those of uninfected animals (Timm and Cook 1979). In addition, bot flies use a small percentage of the energy budget of the host, indicating minor costs to host metabolism (Munger and Karasov 1994). Studies of host survival have provided counterintuitive results. Removal of parasites had no effect on host survival (Munger and Karasov 1991), whereas Clark and Kaufman (1990) found that parasitized P. leucopus had longer residence (i.e., survival) times. However, these studies included both transients and residents in analyses of survivorship (Clark and Kaufman 1990; Munger and Karasov 1991). For accurate assessment of survivorship, only residents should be considered because they are more likely to be infected compared to transients (Hunter et al. 1972; Jaffe et al. 2005). In a recent study, Burns et al. (2005) confirmed that infected animals survived 1103

2 1104 JOURNAL OF MAMMALOGY Vol. 87, No. 6 longer than uninfected animals, after correcting for residence bias. Although few costs to individuals have been revealed, adverse effects of bot fly parasitism may be manifest at the population level. There is some evidence that bot fly infections negatively affect population growth rates (Burns et al. 2005). In addition, the proportion of reproductive individuals was reduced in some populations subject to bot fly infection (Brown 1965; Dunaway et al. 1967; Sealander 1961; Wecker 1962), but not in others (Burns et al. 2005). Nevertheless, infestation does not affect length of the breeding season (Munger and Karasov 1991). One shortcoming of many studies is that they have only used prevalence, or the proportion of the host population that is infected at a given point in time, to measure parasitism (Brown 1965; Burns et al. 2005; Clark and Kaufman 1990; Dunaway et al. 1967; Goertz 1966; Hunter et al. 1972; Miller and Getz 1969; Smith 1977a; Xia and Millar 1990). Other measures of parasitism, such as intensity (the number of parasites divided by the number of hosts Margolis et al. 1982) and the distribution of parasites among hosts provide a broader understanding of parasite host relationships. Information on the distribution of parasites can be used to identify patterns in infection, such as the frequency of multiple simultaneous infections, or how many individuals escape parasitism. Certain demographic subsets of the population, such as young males or reproductive females, could be more susceptible to parasites, which would be reflected in intensity values. Although the distribution of parasites among hosts has been reported (Hensley 1976; Smith 1977a; Wecker 1962), this pattern has not been statistically assessed to determine whether parasites are clumped in a small subset of the population or whether parasites are distributed randomly among hosts. These differential effects could alter population sex ratio, survival, or fitness. In combination with prevalence, these measures provide a more complete picture of population-level effects of bot flies. The objectives of this study were to expand the approach of other studies by assessing the effects of prevalence and intensity of bot fly parasitism on host population measures such as density, sex ratio, reproduction, survival, physical condition, and the distribution of parasites among hosts. If bot flies affect population structure, their distribution pattern among mice would be expected to be nonrandom. Moreover, significant relationships between parasitism and host density, sex ratio, and proportion reproductively active could indicate an effect that parasites have on the host population. In addition, infected mice would be expected to be in worse physical condition and have shorter survival times than uninfected mice if there was a cost of bot fly parasitism. MATERIALS AND METHODS Study site. This study was conducted at East Fork Wildlife Area, Williamsburg, Ohio (39819N, 84849W), a 1,095-ha area reserved for recreational hiking, hunting, and fishing. This area is located at the northeastern edge of East Fork Lake, near the confluence with the east fork of the Little Miami River. The habitat is 2nd-growth eastern deciduous forest, interspersed with agricultural fields and old fields of varying successional stages. Common tree species include sugar maple (Acer saccharum), beech (Fagus grandifolia), red oak (Quercus rubra), white oak (Q. alba), shagbark hickory (Carya ovata), Ohio buckeye (Aesculus glabra), and wild black cherry (Prunus serotina). Undergrowth was dominated by poison ivy (Toxicodendron radicans), Virginia creeper (Parthenocissus quinquefolia), garlic mustard (Alliaria petiolata), multiflora rose (Rosa multiflora), and Amur honeysuckle (Lonicera maackii). Rodent trapping. Three m grids were established in forested areas during May 2002, and a 4th grid was added in July Average distance between grids was 0.81 km (range: km). Mice were never observed moving among grids. Each grid consisted of 36 aluminum Sherman collapsible live traps ( cm; H. B. Sherman Traps, Inc., Tallahassee, Florida), placed in a 6 6 configuration, with a spacing interval of 10 m. Each site was trapped once weekly from May to November In 2003, all 4 sites were trapped simultaneously over a 3-night period each month from April to November. In addition, during the peak of bot fly infection (July September), each site was trapped once weekly. Generally, 2 sites were trapped in a given night, except during the 3-night trapping periods, when all grids were trapped simultaneously. Traps were removed after each night of trapping, except during the 3-night trapping in 2003, when traps were maintained on the grids during the trapping period. Vandalism from northern raccoons (Procyon lotor), striped skunks (Mephitis mephitis), and long-tailed weasels (Mustela frenata) was rare, and any damaged traps were replaced. Traps were set at least 1 h before sunset (1900 h) and checked the following morning ( h). During summer (July September), traps were checked approximately 4 h after sunset, encompassing the period of maximal activity for white-footed mice (Bruseo and Barry 1995). Time of trap check had minimal effect on capture success: the same proportion of captures was obtained by checking traps in the morning (mean 6 SE: ) as during the night ( ; t ¼ 0.474, d.f. ¼ 24, P ¼ 0.640). Traps were baited with a mixture of rolled oats and sunflower seeds. Traps were not prebaited. Data recorded for each capture of P. leucopus were sex, age class (adult or juvenile; based on weight and pelage [sensu Layne 1968]), body mass, body length, tail length, length of hind foot, number of bot fly infections, diameter of respiratory pore of bot fly, and number and condition (recent or healed) of bot fly wounds. Body length, tail length, and length of hind foot were measured only for the 1st capture of an individual; other measurements were taken for each capture of an individual. Diameter of respiratory pore, which increases with the age of the infection, was measured with a dial caliper and was used to determine the approximate number of days that a host had been infected (Cogley 1991). Body mass of mice was corrected to account for mass of bot fly larvae (Smith 1977a). Reproductive condition also was recorded, based on the location of testes (abdominal or scrotal) for males or the condition of the vagina (open or closed) and mammae (ascended or descended) for females. Each animal was uniquely marked with a metal ear tag (#1005 Size 1 Monel; National Band and Tag Co., Newport, Kentucky). Animals were released at the site of capture. Other small mammal species captured were immediately released without processing. Animals were trapped and handled following the animal care and use guidelines adopted by the American Society of Mammalogists (Animal Care and Use Committee 1998), and trapping protocol was approved by the University of Cincinnati Institutional Animal Care and Use Committee. Statistical analysis. Animals that were captured at least 5 times over the entire trapping period were considered to be residents (Anderson 1989); only residents were used in analyses. Transient mice were excluded from analyses because they were less likely to be infected by bot fly larvae (Hunter et al. 1972).

3 December 2006 CRAMER AND CAMERON BOT FLY PARASITISM OF WHITE FOOTED MICE 1105 Bot flies were present during summer (July September) and autumn (October November; seasons determined by differences in mean daily temperature and precipitation). The effect of season on prevalence (proportion of population infected with bot flies) was analyzed with a paired t-test because of the inherent dependence of the data (the same individuals captured at the same sites in different seasons). If there was a significant seasonal effect, data were separated by season and the potential relationship between an infected individual and its sex, site, year, and reproductive activity was analyzed with logistic regression. Intensity (number of parasites per infected host over entire bot fly season) was analyzed with a general linear model. Season was entered into the model as a repeated measure, and sex and year were used as independent variables. Site was entered into the model as a blocking factor, because of limited degrees of freedom. Some groups were excluded from this analysis because they contained no infected animals, and intensity would therefore be undefined. Population size was calculated as the minimum number of individuals known to be alive (Krebs 1999). Although this index of abundance may underestimate population size because it relies on capture data while ignoring differences in capture probabilities (Pocock et al. 2004), it was the best index for our purposes. Although statistical methods of population estimation likely would remove this capture bias (Slade and Blair 2000), they cannot accurately estimate the number of infected individuals independent of capture data. Mean density (minimum number of individuals known to be alive per hectare) across all grids was calculated for each week during the trapping season. In addition, weekly sex ratio was determined by calculating the proportion of males known to be alive in the population. Proportion of the population that was reproductively active each week was determined based on trapping records. The relationship between parasitism (prevalence and intensity) and these parameters was determined using Pearson s correlation analysis. Two measures of infection were used to ascertain the distribution of parasites within hosts: the total number of bot fly larvae observed in a host over the entire bot fly season, and the maximum number of simultaneous bot fly larvae observed in a host. It was assumed that bot fly infection was a rare event, and therefore would follow a Poisson distribution. Goodness of fit of observed data to a Poisson distribution was determined using a chi-square test. Distribution of bot flies among all individuals was assessed, and then these data were decomposed into groups, based on sex, year, season, and reproductive status. For these tests, a Bonferroni sequential adjustment (Rice 1989) was used to maintain the experimentwise error rate at a ¼ 0.05 (Sokal and Rohlf 1981). Persistence time of white-footed mice was measured as the minimum number of days known alive, which was the number of days between the 1st and last capture of each individual. These data were analyzed using a general linear model, with sex, year, and infection status as independent variables, and minimum number of days known alive as the dependent variable. Because of limited degrees of freedom, site was included in the model as a blocking factor. In addition, a 1-way analysis of variance was used to assess whether persistence time of white-footed mice with a single infection differed from that of white-footed mice with multiple infections. Physical condition was assessed by measuring body mass because it could be measured without unduly stressing animals (Jakob et al. 1996). Host mass was adjusted to account for the mass of the developing parasite. First, the diameter of the parasite s respiratory pore was used to estimate the number of days the host had been infected, based on the regression equation of Cogley (1991). Second, the age of infection was used to estimate the approximate mass of the developing parasite, based on the growth curve in Smith (1977a). However, one problem with estimating condition based on body mass was the high correlation between mass and body length. To control for the effect of body length on body mass, condition was assessed using an analysis of covariance with body length as a covariate (García- Berthou 2001). Because infection could not be manipulated, weight of individuals before infection was compared to weight during infection, requiring a repeated-measures approach. In the analysis, site was entered as a blocking variable, year and sex were entered as independent variables, and body length was entered as a covariate. Mice weighing less than 15 g and with gray pelage were classified as juveniles (Layne 1968) and were excluded from this analysis. In addition, pregnant females were excluded from analyses. General climatic data were obtained from the National Climatic Data Center, National Oceanic and Atmospheric Administration, from the Milford, Ohio, weather station, located approximately 32 km northwest from the study site. These data were used to investigate any relationships between mean daily temperature and mean daily precipitation and prevalence of infection. These data were used because adult bot flies are sensitive to changes in ambient temperature (Shiffer 1983). Although changes in soil temperature would affect pupation time, changes in ambient temperature would affect flight behavior, which would have ramifications for successful reproduction of bot flies. RESULTS Total captures were biased heavily toward P. leucopus (1,542 captures of 266 individuals). Other species captured included eastern chipmunks (Tamias striatus; 9 captures), northern short-tailed shrews (Blarina brevicauda; 22 captures), a southern flying squirrel (Glaucomys volans), and a long-tailed weasel (M. frenata). Infected P. leucopus were not observed before July during either year. Because of a relatively small sample size, variation in prevalence and intensity was high. Nevertheless, there was a significant seasonal pattern. Prevalence decreased significantly in September in both 2002 and 2003 (Fig. 1A). Although there was an increase in prevalence after this decline, it was more pronounced in 2002 than in 2003 (Fig. 1A). Intensity also showed distinct summer and autumn peaks in both years (Fig. 1B). Prevalence and intensity were significantly correlated in 2002 (r ¼ 0.775, P, 0.001), but not 2003 (r ¼ 0.279, P ¼ 0.247). A higher proportion of animals were infected in summer (mean 6 SE: ) than in autumn ( ; t ¼ 2.99, d.f. ¼ 15, P ¼ 0.009), thus we further separated analysis of prevalence by season. For both summer and autumn, there was a difference in prevalence between years (Table 1). In addition, reproductively active individuals were more likely to harbor infections than reproductively inactive individuals during summer (Table 1), although this result is suspect because of the discrepancy in sample sizes between reproductively active (n ¼ 85) and inactive (n ¼ 15) individuals. Intensity of infections was marginally greater in summer ( bots/ individual) than in autumn ( bots/individual; F ¼ 6.15, d.f. ¼ 1, 4, P ¼ 0.068, n ¼ 11; Fig. 1B). However, there was no effect of site (F ¼ 0.71, d.f. ¼ 3, 4, P ¼ 0.594, n ¼ 11), sex (F ¼ 0.84, d.f. ¼ 1, 4, P ¼ 0.412, n ¼ 11), year (F ¼ 1.14, d.f. ¼ 1, 4, P ¼ 0.347, n ¼ 11), or the interaction between sex and year (F ¼ 0.09, d.f. ¼ 1, 4, P ¼ 0.775, n ¼ 11). This result

4 1106 JOURNAL OF MAMMALOGY Vol. 87, No. 6 FIG. 1. Weekly A) prevalence (proportion of population infected with bot flies) and B) intensity (number of parasites per infected host) for populations of Peromyscus leucopus in Ohio in 2002 (closed diamonds) and 2003 (open squares), shown as mean 6 SE. could reflect the fact that intensity is mathematically undefined for groups with no infected animals, which makes it more difficult to detect differences because variation increases with a decrease in sample size. Host population density was low in spring (May June), reached a peak during summer (July September), and declined into autumn (October November; Fig. 2A). Weekly densities of P. leucopus did not vary between years during the bot fly season (July November; paired t-test: t ¼ 0.92, d.f. ¼ 18, P ¼ 0.364; Fig. 2A). Density of host populations was not correlated with prevalence for either year (Table 2), but was significantly correlated with intensity in 2002 (Table 2). There was no difference between years for proportion of males in the population (paired t-test: t ¼ 1.395, d.f. ¼ 18, P ¼ 0.180; Fig 2B). There was no significant relationship between sex ratio (proportion of males) of the population and prevalence in either year (Table 2), but there was a significant relationship between proportion of males and intensity in 2002 (Table 2). The proportion of reproductive individuals was higher in 2003 than 2002 (paired t-test: t ¼ 9.822, d.f. ¼ 18, P, 0.001; Fig 2C), but there was no relationship between proportion reproductively active and either measure of parasitism (Table 2). The relationship between prevalence and proportion reproductively active was nonlinear (2002: r 2 ¼ 0.762, F ¼ 98.42, d.f. ¼ 3, 17, P, 0.001; 2003: r 2 ¼ 0.480, F ¼ 24.72, d.f. ¼ 3, 17, P, 0.001; Fig. 3). Accordingly, an alternative analysis of prevalence and proportion of reproductively active individuals indicated that these variables were significantly asynchronous in 2003 (Kologorov Smirnov test: 2002: d max ¼ 0.20, P ¼ 0.749; 2003: d max ¼ 0.74, P, 0.001; Figs. 1 and 2C). The total number of bot flies observed per host over the entire bot fly season ranged from 0 to 9, with a mean of (n ¼ 90 hosts). Total number of bot fly larvae observed per individual was significantly clumped for all data (Fig. 4A) and for both sexes, summer infections, and both reproductively active and inactive individuals (Table 3). In addition, a high percentage of individuals that developed infections during autumn also were infected the previous summer in both years (2002: 85% reinfected, v 2 ¼ 6.231, d.f. ¼ 1, n ¼ 13, P ¼ 0.013; 2003: 88% reinfected, v 2 ¼ 4.500, d.f. ¼ 1, n ¼ 8, P ¼ TABLE 1. Bot fly prevalence (proportion of population infected) in Peromyscus leucopus in Ohio analyzed with logistic regression by sex, site, year, and reproductive condition for summer and autumn. Factor Summer Autumn Prevalence n Wald v 2 d.f. P Prevalence n Wald v 2 d.f. P Sex Male Female Site Site Site Site Site Year Reproductive condition Active Inactive

5 December 2006 CRAMER AND CAMERON BOT FLY PARASITISM OF WHITE FOOTED MICE 1107 TABLE 2. Comparison of weekly prevalence (proportion of population infected) and intensity (number of parasites per infected host) of parasitism of Peromyscus leucopus in Ohio with weekly measures of population parameters as shown by Spearman correlation coefficients (r). Intensity was log-transformed before analyses. Population measures Density Proportion male Proportion reproductive FIG. 2. Weekly A) population density (minimum number known alive per hectare), B) sex ratio (proportion of males), and C) proportion of population reproductively active for 2002 (closed diamonds) and 2003 (open squares) for Peromyscus leucopus in Ohio, shown as mean 6 SE ). The percentage of reinfections did not differ between years (v 2 ¼ 0.034, d.f. ¼ 1, P ¼ 0.854). The distribution of parasites in hosts was random for total number of parasites per host in both years (regardless of season) and autumn (Table 3). The random distribution in autumn could have resulted from fewer infections and reduced degrees of freedom for this test. On the other hand, the maximum number of simultaneous bot fly infections per host ranged from 0 to 6 infections, with X ¼ The dispersion of parasites within hosts was random (v 2 ¼ 4.83, d.f. ¼ 4, P ¼ 0.305; Fig. 4B). The same distribution pattern was evident for sex, year, season, and reproductive condition (Table 3). Overall, most hosts (66%) harbored a single bot fly infection at a time. Parasitism Year r P r P r P Prevalence Intensity Infected animals persisted on the study site an average of 42.7 days longer than uninfected animals (Table 4), but there was no effect of site, sex, or year (Table 4). In addition, there were no significant interactions between year, sex, and infection (year sex: F ¼ 1.038, d.f. ¼ 1, 91, P ¼ 0.311, n ¼ 102; year infection: F ¼ 1.067, d.f. ¼ 1, 91, P ¼ 0.304, n ¼ 102; sex infection: F ¼ 0.135, d.f. ¼ 1, 91, P ¼ 0.714, n ¼ 102; year sex infection: F ¼ 0.096, d.f. ¼ 1, 91, P ¼ 0.758, n ¼ 102). Animals with multiple infections persisted longer than uninfected animals or animals with single infections (Table 4). Individuals of P. leucopus weighed significantly more when they were infected (Table 5); this result was consistent regardless of body length (F ¼ 6.037, d.f. ¼ 1, 63, P ¼ 0.017; Fig. 5). The difference between weight before infection and during infection was significantly greater for males than females, and weight gains with infection were greater in 2003 than in 2002 (Table 5). DISCUSSION Several of the results of this study were counterintuitive to the assumption that there is a cost to bot fly parasitism. First, physical condition increased after animals were infected by bot flies; that is, body mass corrected for mass of the bot fly increased after onset of infection. Laboratory studies have FIG. 3. Relationship between proportion reproductive and prevalence (proportion infected) for 2002 (closed diamonds) and 2003 (open diamonds). Best-fit lines assume a 2nd-order polynomial relationship.

6 1108 JOURNAL OF MAMMALOGY Vol. 87, No. 6 FIG. 4. Observed dispersion of parasites (bars) within hosts compared to that expected assuming a Poisson distribution (dark line) for A) total number of bot fly larvae observed per individual over the entire season, and B) maximum number of simultaneous infections observed per individual. demonstrated an increase in food intake by infected whitefooted mice, presumably to replace proteins lost to the parasite (Hunter and Webster 1974). In addition, laboratory and field studies demonstrated that uninfected white-footed mice preferred high-energy, low-protein resources (Lewis et al. 2001). Therefore, unless mice switch preference to high-protein resources during infection, they would have to consume more high-energy resources to replace proteins lost to the parasite. Switching to high-energy resources also would increase their stores of fat, leading to an increase in weight. The 2nd counterintuitive result was that resident animals infected with bot flies had longer residence times than those who escaped infection. Animals infected with bot flies persisted longer on the trapping grids, even after accounting for bias due to residency status (Hunter et al. 1972). In addition, animals with multiple simultaneous bot fly infections survived longer than those with a single infection. This finding does not necessarily indicate a causal relationship between infection and survivorship because animals that survive longer may be exposed to parasites for a longer period, and therefore may develop more infections. Further study, using experimental manipulation of parasite loads, would be required to determine if a causal relationship exists (Munger and Karasov 1994). In addition, Burns et al. (2005) suggested that there may be a trade-off between survival and reproductive output for females. They also found that mice with infections survived longer, but these mice produced fewer litters per year than those that escaped infection. Another objective of this study was to determine whether bot flies had a negative effect on population demography. Analysis of the distribution of infections within hosts indicated that the distribution of total bot flies observed per host was clumped; that is, some hosts harbored sequential infections of up to 9 bot flies, whereas others avoided infection completely. Analysis of reinfection data supported the idea that previously infected individuals were more likely to become reinfected. However, no demographic group had higher prevalence, suggesting that the clumped distribution of bot flies had no negative effects on demography. On the other hand, the distribution of simultaneous bot fly infections was random. These results suggested that multiple infections were less frequent than single infections, and most infected individuals (66%) harbored a single bot fly infection at a time, in concordance with other studies of bot fly parasitism (Dunaway et al. 1967; Hensley 1976; Miller and Getz 1969; Munger and Karasov 1991; Scott and Snead 1942; Sealander 1961; Smith 1977a; Test and Test 1943; Wecker 1962). Such a result was not expected because a single female bot fly can lay as many as 2,500 eggs in host habitat (Catts 1982). Female bot flies do not lay their eggs directly on the host, so gravid flies cannot avoid hosts that are already infected. Competition may occur between larvae, which is supported by the fact that larvae collected from hosts with multiple infections are of lower mass than those that are collected from hosts with a single parasite (Smith 1977a). Another possibility is that once infected, hosts alter their behavior to avoid contact with other bot fly eggs. Types of behavior that may achieve this goal include changes in use of space, movement, or grooming. Smith (1978a) suggested that infected mice may spend more time grooming than uninfected mice. There was no observed relationship between prevalence and density. In many parasite host systems, prevalence is positively related to density because parasites in crowded populations move between hosts more easily (Altizer et al. 2003; Arneberg et al. 1998). However, bot fly larvae are not passed between hosts, which may explain the lack of a relationship between density and prevalence. On the other hand, there was a significant positive relationship between intensity and density. This result could be explained by concomitant seasonal trends in both density and parasitism. Intensity was highest in summer, coincident with a peak in population density. It is also possible that female bot flies alter oviposition rate to track

7 December 2006 CRAMER AND CAMERON BOT FLY PARASITISM OF WHITE FOOTED MICE 1109 TABLE 3. Dispersion of bot flies within Peromyscus leucopus in Ohio; observed numbers of parasites per host were compared to expected values assuming a Poisson distribution. For significant deviations from expected values, analysis of coefficient of dispersion (CD) was used to determine dispersion pattern. P-values less than 0.01 were considered to be statistically significant, based on Bonferroni s sequential adjustment (Rice 1989). Factor Total number of parasites changes in host density, but this is unlikely given the short life span of adult bot flies (Catts 1982). Alternatively, intensity could increase if infected animals were more susceptible to new parasites, which would occur if individuals in high density populations had reduced immune defenses. Intensity also was correlated with the proportion of males in the population; that is, there were more males in the population when intensity of parasitism was high. This result implies that females may be more affected by infection than males. However, there was no other evidence of a differential sex effect of parasitism. The same proportion of females were infected as males, females had the same mean intensity as males, and females had the same persistence times as males. The only difference between the sexes was that males gained more weight with infection than did females, implying that females could be less affected by parasitism than males, not more affected. One possible explanation for the inverse relationship between proportion of females and intensity is that females moved less when infected than did males, which would lead to male-biased capture rates. Territory sizes are generally smaller for females compared to males because females defend their nest sites from other females (Wolff 1989). Therefore, given that intensity of parasitism was equivalent for males and females, females may compensate for infection by reducing their movements, without losing the advantage of high-quality nesting sites. Preliminary analyses of our unpublished data on movements of infected males and females indicated that females moved less. The analysis of patterns in reproduction and infection raises some interesting questions. In both years, the peak of reproduction occurred in late summer, coincident with a decline in prevalence and intensity. This asynchrony was significant in 2003, but not This result reflected higher reproduction and lower prevalence in 2003, leading to a greater difference. In addition, prevalence was highest in 2002, the same year that reproduction was lowest. Although such correlations do not imply causality, this result is similar to that of Burns et al. (2005), who also reported a negative effect of parasitism on reproduction in P. leucopus. To verify an inverse relationship between parasitism and reproduction, which would indicate a clear negative effect of bot fly parasitism, quantification of the Maximum number of simultaneous infections v 2 d.f. P n CD Dispersion v 2 d.f. P n CD Dispersion Overall , Clumped Random Male , Clumped Random Female Clumped Random Random Random Random Random Summer , Clumped Random Autumn a Random Reproductively active , Clumped Random Reproductively inactive Clumped Random a Values for total number of infections and maximum number of simultaneous infections were the same in autumn. number of offspring produced would be necessary. Nevertheless, the results of this study and that of Burns et al. (2005) indicate that negative effect of bot flies on reproduction may exist, and that it merits further attention. There was significant temporal variation in infection. As in many studies, bot fly parasitism was highly seasonal (reviewed by Catts [1982]), with the peak of both prevalence and intensity occurring in summer (July September), with a smaller peak in autumn (October November). The higher prevalence observed in 2002 than in 2003 may reflect climatic factors that affect bot flies. Mean daily temperature was higher during the bot fly season in 2002 than in 2003 (X 2002 ¼ 21.58C; X 2003 ¼ 20.08C, n ¼ 120; paired t-test: t ¼ 3.208, d.f. ¼ 119, P ¼ 0.002). In addition, there was a nonsignificant increase in mean daily precipitation in 2003 (X 2002 ¼ 3.1 mm; X 2003 ¼ 5.5 mm, n ¼ 115; Wilcoxon signed ranks test: Z ¼ 1.937, P ¼ 0.053). Thus, on average, the bot fly season was wetter and cooler in 2003 than in Field observations of C. fontinella indicate that TABLE 4. Mean persistence time, measured as the minimum number of days known alive, for Peromyscus leucopus of differing infection status. Factor X SE n F d.f. P Sex Male , , Female Year , Site Site , Site Site Site Infection status Uninfected , Infected Single infections , Multiple infections

8 1110 JOURNAL OF MAMMALOGY Vol. 87, No. 6 TABLE 5. Physical condition (mean weight in grams) of individual Peromyscus leucopus before and during bot fly infection. Paired data were analyzed with a repeated-measures analysis of covariance. Weight of developing bot fly larvae was removed before analyses. Weight before infection Weight after infection Factor X SE X SE n F P Overall Sex Male Female Year adult bot flies required a temperature of approximately 238C for initiation of flight activity (Shiffer 1983). Flight is essential for bot fly mating, because males spend most of their time displaying and defending small territories to attract females (Shiffer 1983). In this study, 57% of the days in the bot fly season of 2002 had a mean daily temperature greater than 238C, whereas only 31% of the days in the bot fly season of 2003 had temperatures over 238C (Fisher s exact test: v 2 ¼ , d.f. ¼ 1, P, 0.001). Therefore, there were more days in the bot fly season in 2002 during which temperatures were high enough to allow bot fly flight activity, which may have affected parasite reproductive behavior and hence prevalence levels in the host. Climatic factors thus may have had more of an influence on bot fly populations than the number of suitable hosts would suggest. The bimodality observed in prevalence and intensity may be linked to the life cycle of the parasite. Most hosts in the population are parasitized by midsummer, usually with a single bot at a time. During this period, additional parasites may infect an already parasitized host because suitable uninfected hosts are rare, which leads to an increase in intensity. It is possible for C. fontinella to infect other host species, such as eastern chipmunks (T. striatus), but such infections should be rare because bot flies tend to be very host-specific, and T. striatus is parasitized by another bot fly species, C. emasculator (Sabrosky 1986). As larvae complete their development and leave their hosts, prevalence drops accordingly, followed by a drop in intensity. After a late summer lull in parasitism, nearly all individual hosts are reinfected (86% over both years), giving rise to the autumn increase in prevalence and intensity. Lower temperatures in autumn may affect hatching rates and larval movement, thus making it more difficult for fly larvae to reinfect hosts, leading to a smaller peak in prevalence and intensity in autumn. Prevalence and intensity were significantly correlated in 2002 but not in 2003, which could be due to the fact that prevalence was lower that year; that is, host saturation was not as ubiquitous as it was in Results of this study indicated that bot flies do not negatively affect survival (persistence time) or physical condition of their hosts. The results also indicated that, although bot fly parasitism FIG. 5. Individual body weights (g) of Peromyscus leucopus as a function of body length (mm). Closed diamonds indicate weights before infection and the open squares the weights of same mice during infection. Lines are best fit lines before infection (solid line) and during infection (dashed line). was prevalent in this population, hosts generally were infected by a single parasite at a time. Even though parasites are clumped within few individuals over the season, there is no evidence of a negative demographic effect. This supports the idea that these species have been in close association for enough time that each species has evolved a tolerance for the other (Timm and Cook 1979). Some evidence indicates that multiple infections do cause a potential survival cost to the host in terms of increased predation rates (Miller and Getz 1969; Smith 1978b), although this idea is not supported by the longer persistence times of parasitized mice in this study. The negative relationship between the proportion reproductively active and prevalence may belie a cost, but more data are necessary to document the trend reported here. The fact that a definitive cost to the host cannot be determined brings into question the validity of defining the interaction between bot flies and white-footed mice as parasitism. It is also possible that bot flies may impose a cost to white-footed mice in other aspects of their biology, such as behavior. Further research on the effects of bot fly parasitism on white-footed mouse behavior would help to resolve the nature of this relationship. ACKNOWLEDGMENTS Many people helped with the conceptual and physical implementation of this study. T. Kane, K. Petren, M. Polak, N. Solomon, and G. Uetz helped greatly with the development of this project. This study could not have been done without the assistance of the Ohio Department of Natural Resources, Division of Wildlife, especially R. Morgan and T. Haynes. Field assistants A. Allemang, C. Christopher, G. Klein, C. Hennessy, A. Mattingly, D. McCubbin, T. Pharr, K. Roberts, and A. Shelton facilitated data collection. F. Bennett helped with the logistic regression analysis. Financial support was provided by the Department of Biological Sciences at the University of Cincinnati through several Weiman Summer Research Grants and Graduate Assistantships. C. Christopher, M. Hopton, C. Hennessy, and 2 anonymous reviewers provided thoughtful comments on earlier versions of this manuscript.

9 December 2006 CRAMER AND CAMERON BOT FLY PARASITISM OF WHITE FOOTED MICE 1111 LITERATURE CITED ALTIZER, S., ET AL Social organization and parasite risk in mammals: integrating theory and empirical studies. Annual Review of Ecology, Evolution and Systematics 34: ANDERSON, P. K Dispersal in rodents: a resident fitness hypothesis. Special Publication 9, The American Society of Mammalogists. ANIMAL CARE AND USE COMMITTEE Guidelines for the capture, handling, and care of mammals as approved by the American Society of Mammalogists. Journal of Mammalogy 79: ARNEBERG, P., A. SKORPING, B.GRENFELL, AND A. F. READ Host densities as determinants of abundance in parasite communities. Proceedings of the Royal Society of London, B. Biological Sciences 265: BROWN, L. N Botfly parasitism in the brush mouse and whitefooted mouse in the Ozarks. Journal of Parasitology 51: BRUSEO, J. A., AND R. E. BARRY, JR Temporal activity of syntopic Peromyscus in the central Appalachians. Journal of Mammalogy 76: BURNS, C. E., B. J. GOODWIN, AND R. S. OSTFELD A prescription for longer life? Bot fly parasitism of the white-footed mouse. Ecology 86: CATTS, E. P Biology of new world bot flies: Cuterebridae. Annual Review of Entomology 27: CLARK, B. K., AND D. W. KAUFMAN Prevalence of botfly (Cuterebra sp.) parasitism in populations of small mammals in eastern Kansas. American Midland Naturalist 124: COGLEY, T. P Warble development by the rodent bot Cuterebra fontinella (Diptera: Cuterebridae) in the deer mouse. Veterinary Parasitology 38: DALMAT, H. T A contribution to the knowledge of the rodent warble flies (Cuterebridae). Journal of Parasitology 29: DUNAWAY, P. B., J. A. PAYNE, L.L.LEWIS, AND J. D. STORY Incidence and effects of Cuterebra in Peromyscus. Journal of Mammalogy 48: GARCÍA-BERTHOU, E On the misuse of residuals in ecology: testing regression residuals vs. the analysis of covariance. Journal of Animal Ecology 70: GOERTZ, J. W Incidence of warbles in some Oklahoma rodents. American Midland Naturalist 75: HENSLEY, M. S Prevalence of cuterebrid parasitism among woodmice in Virginia. Journal of Wildlife Diseases 12: HUNTER, D. M., R. M. F. S. SADLEIR, AND J. M. WEBSTER Studies on the ecology of cuterebrid parasitism in deermice. Canadian Journal of Zoology 50: HUNTER, D. M., AND J. M. WEBSTER Effects of cuterebrid larval parasitism on deer-mouse metabolism. Canadian Journal of Zoology 52: JAFFE, G., D. A. ZEGERS,M.A.STEELE, AND J. F. MERRITT Longterm patterns of botfly parasitism in Peromyscus maniculatus, P. leucopus, and Tamias striatus. Journal of Mammalogy 86: JAKOB, E. M., S. D. MARSHALL, AND G. W. UETZ Estimating fitness: a comparison of body condition indices. Oikos 77: KREBS, C. J Ecological methodology. 2nd ed. Benjamin Cummings, Menlo Park, California. LAYNE, J. N Ontogeny. Pp in Biology of Peromyscus (Rodentia) (J. A. King, ed.). Special Publication 2, The American Society of Mammalogists. LEWIS, C. E., T. W. CLARK, AND T. L. DERTING Food selection by the white-footed mouse (Peromyscus leucopus) on the basis of energy and protein contents. Canadian Journal of Zoology 79: MARGOLIS, L., G. W. ESCH, J.C.HOLMES, A.M.KURIS, AND G. A. SCHAD The use of ecological terms in parasitology (report of an ad hoc committee of the American Society of Parasitologists). Journal of Parasitology 68: MILLER, D. H., AND L. L. GETZ Botfly infections in a population of Peromyscus leucopus. Journal of Mammalogy 50: MUNGER, J. C., AND W. H. KARASOV Sublethal parasites in white-footed mice: impact on survival and reproduction. Canadian Journal of Zoology 69: MUNGER, J. C., AND W. H. KARASOV Costs of bot fly infection in white-footed mice: energy and mass flow. Canadian Journal of Zoology 72: POCOCK, M. J. O., A. C. FRANTZ,D.P.COWAN,P.C.L.WHITE, AND J. B. SEARLE Tapering bias inherent in minimum number alive (MNA) population indices. Journal of Mammalogy 85: RICE, W. R Analyzing tables of statistical tests. Evolution 43: SABROSKY, C. W North American species of Cuterebra, the rabbit and rodent bot flies (Diptera: Cuterebridae). Entomological Society of America, College Park, Maryland. SCOTT, T. G., AND E. SNEAD Warbles in Peromyscus leucopus noveboracensis. Journal of Mammalogy 23: SEALANDER, J. A Hematological values in deer mice in relation to botfly infection. Journal of Mammalogy 42: SHIFFER, C. N Aggregation behavior of adult Cuterebra fontinella (Diptera: Cuterebridae) in Pennsylvania, USA. Journal of Medical Entomology 20: SLADE, N. A., AND S. M. BLAIR An empirical test of using counts of individuals captured as indices of population size. Journal of Mammalogy 81: SMITH, D. H. 1977a. The natural history and development of Cuterebra approximata (Diptera: Cuterebridae) in its natural host, Peromyscus maniculatus (Rodentia: Cricetidae), in western Montana. Journal of Medical Entomology 14: SMITH, D. H. 1977b. Effects of experimental bot fly parasitism on gonad weights of Peromyscus maniculatus. Journal of Mammalogy 58: SMITH, D. H. 1978a. Effects of bot fly (Cuterebra) parasitism on activity patterns of Peromyscus maniculatus in the laboratory. Journal of Wildlife Diseases 14: SMITH, D. H. 1978b. Vulnerability of bot fly (Cuterebra) infected Peromyscus maniculatus to shorttail weasel predation in the laboratory. Journal of Wildlife Diseases 14: SOKAL, R. R., AND F. J. ROHLF Biometry. 2nd ed. W. H. Freeman and Company, New York. TEST, F. H., AND A. R. TEST Incidence of dipteran parasitosis in populations of small mammals. Journal of Mammalogy 24: TIMM, R. M., AND E. F. COOK The effect of bot fly larvae on reproduction in white-footed mice, Peromyscus leucopus. American Midland Naturalist 101: WECKER, S. C The effects of bot fly parasitism on a local population of the white-footed mouse. Ecology 43: WOLFF, J. O Social behavior. Pp in Advances in the study of Peromyscus (Rodentia) (G. L. Kirkland, Jr., and J. N. Layne, eds). Texas Tech University Press, Lubbock. XIA, X., AND J. S. MILLAR Infestations of wild Peromyscus leucopus by bot fly larvae. Journal of Mammalogy 71: Submitted 23 November Accepted 2 June Associate Editor was Rodrigo Medellín.