Parasitol Res DOI 10.1007/s00436-011-2478-1

ORIGINAL PAPER

Seasonality of gregarine parasitism in the , Nehalennia irene: understanding unimodal patterns

Mark R. Forbes & Julia J. Mlynarek & Jane Allison & Kerry R. Hecker

Received: 30 December 2010 /Accepted: 19 May 2011 # Springer-Verlag 2011

Abstract We studied parasitism by gut protozoans 1971; Zuk 1987; Locklin and Vodopich 2010a). Of course, (Apicomplexa: Eugregarinidae) in the damselfly, Nehalennia seasonality might also occur independent of age differences irene (Hagen) (: ). We tested whether in samples of hosts. An early study by Forbes and Baker there was any seasonal pattern, as has been found for other (1991) showed that parasitic Arrenurus mites decreased on parasites of and which has implications for newly emerged (same-aged) damselflies, as the season selection on emergence and breeding. Using aggregate data advanced. from 12 date-by-site comparisons involving five sites, we Seasonal changes in parasitism are important to the found that both prevalence and intensity of gregarine extent that the parasites or pathogens under study act as parasitism were seasonally unimodal. Parasitism first agents of selection (Altizer et al. 2006). Advantages to early increased and then declined seasonally after peaking mid- breeding could be counteracted by selection imposed by season. This damselfly species has shown seasonal increases parasites, if parasites were most common early in the in density followed by declines at several sites including a site breeding or emergence period (Altizer et al. 2006). Later sampled in this study. Therefore, similar seasonal changes in a emergence might be costly in terms of finding a mate, but directly transmitted parasite were expected and are now beneficial if late emergence was associated with less confirmed. Other factors that might account for seasonal likelihood of becoming parasitized. In comparison, later changes in parasitism by gregarines are either unlikely or can emergence might be disadvantageous if infective stages of be discounted including sampling of older damselflies parasites accumulate through time due to the abundance of mid-season but not late in the season, or sex biases in host individuals that emerged earlier and contribute to an parasitism and overrepresentation of the more parasitized sex increased number of infective stages, as is expected for mid-season. directly transmitted parasites with single obligate hosts (Åbro 1971, 1974). Odonate–gregarine interactions have been studied for Introduction many species of odonates with mixed results or opinions as to the effect of gregarines on their odonate host (Åbro 1971, of many species are expected to show increased 1974, 1976, 1987, 1990, 1996; Locklin and Vodopich 2009, parasitism as they age simply because they have been 2010a). Gregarines are apicomplexan protozoans that have around longer to encounter more infective stages of para- a one-host life cycle. The infective cysts are ingested by sites (e.g., Smith and Cook 1991; Hassall et al. 2010). damselfly hosts either along with prey or through drinking Related to this phenomenon is the fact that samples of (Locklin and Vodopich 2010a). They develop, grow, and insects can show seasonality in patterns of parasitism (Åbro reproduce in the midgut of their host. Once the gametocysts : : : have formed, they are excreted with the feces to infect M. R. Forbes J. J. Mlynarek (*) J. Allison K. R. Hecker another individual (Åbro 1987). Since these parasites are Department of Biology, Carleton University, transmitted trophically, differences in feeding behavior and 209 Nesbitt Bldg, 1125 Colonel By Drive, Ottawa, ON K1S 5B6, Canada habitat use in relation to density have to be considered for e-mail: [email protected] males and females and for individuals at different sites. Parasitol Res

We studied the prevalence and intensity of infection by 76°19′ W). Lindsey Lake Marsh (LLM) is 3 ha in size at septate gregarines in the damselfly Nehalennia irene the southern edge of Lindsey Lake. Barb’s Marsh (BM) is (Hagen). Nehalennia irene is a widespread damselfly approximately 3 ha in size and adjoins a large hayfield. inhabiting a variety of freshwater marshes (Walker 1953). Elgin Pond (EP) is 0.8 ha and has a marshy circumference, Females exhibit two distinct color morphs; andromorphs with a mature oak forest and grassy understory surrounding closely resemble males while heteromorphs are unlike it. Upper Dowsley Lake (UD) is 0.9 ha and approximately males. Andromorphs of N. irene range from 2.1% to 95% round in shape, with a grassy margin (including sedges) of all females (Forbes et al. 1995; Van Gossum et al. 2007). bordered by mixed deciduous forests. Indian Lake Bight Andromorphs appear to experience less male harassment (ILB) is 16 ha in size and part of a reedy lake with shallow and are found in greater densities at pond edges among water and short grassy riparian vegetation. The dominant mate-searching males (Forbes et al. 1995), but more emergent vegetation at all sites is Typha spp. and Carex recently, the density–morph relations have been challenged, spp., and dead trees protrude from the surface of the water, and mixed results have been found in broad surveys (Van providing additional emergence sites for eclosing N. irene. Gossum et al. 2007). As in many other damselfly species, N. irene males spend much of their time mate searching Collection and dissections of individuals and enumeration around natal pond edges, presumably foraging only long of gregarines enough to satisfy energy needs for mate searching (Anholt 1992). In contrast, females maximize foraging time to Reproductively mature adults of N. irene were sweep net mature clutches of eggs. Foraging differences between collected from 16 June–27 July, 1997. Maturity was males and females may lead to differential exposure to determined by body coloration. Adults were sexed, and gregarine infective stages, which are ingested along with females were further classified as andromorph or hetero- food (Åbro 1976). Such sex differences in parasitism have morph (Van Gossum et al. 2007). After decapitation, the been documented for another coenagrionid species (Hecker body was weighed, and the gut was removed by gently et al. 2002). pulling on the posterior abdominal segment with forceps We estimated prevalence and intensity (sensu Bush et al. until the entire digestive tract emerged. The gut was

1997) throughout the season, in different sampling loca- preserved in sugared ethanol (1 L H2O:1 L 95% tions, and for males and two different female morphs. In the EtOH:40 g sucrose) and stored in an Eppendorf micro- first instance, we predict that because males forage in centrifuge tube until dissected. A drop of 2.5% pharma- different areas and forage less than females, males will be ceutical iodine was used to increase contrast between less parasitized by gregarines than same-aged females. We gregarines and tissue. The gut was dissected, and gregarines further predicted that gregarine infection should be highest were separated from the gut wall using size 00 pins. at sites when population densities were highest because The gregarines were prepared using Clopton (1997) high density should increase transmission of infective standard protocols. The gregarines had three distinct body stages. Earlier work by Van Gossum et al. (2007) showed parts and are likely Hoplorhynchus spp. and one as yet that male density first increased at sites seasonally unidentified (Clopton, personal communication). (including sites in this study) and then declined after Estimates of prevalence, defined as the number of peaking mid-season. The purpose of this study was to test individuals infected with at least one gregarine compared for, and explain, any seasonal patterns in parasitism of N. to all individuals collected for a sample, were compared irene by gregarines. between males and females and between female morphs (on a date-by-site basis). Estimates were provided with Clopper–Pearson 95% confidence intervals (Zar 1996). If Materials and methods the 95% confidence intervals overlapped between samples of individuals (controlling for date and site), the prevalence Study sites of gregarine infection for the sexes (or morphs) was statistically indistinguishable. We also compared estimates Five sites were chosen for recurrent sampling of N. irene of prevalence for males and both female morphs across all damselflies and assessment of parasitism levels. This 12 date-by-site comparisons (36 estimates) using a one-way species was studied because much is known about its ANOVA (alpha=0.05). natural history including how male density changes at sites Intensity data were analyzed in two ways. We first seasonally. Males and females were collected from five transformed (Log10) intensity data to satisfy the assump- flooded beaver ponds at those sites. All ponds were within tions of normality (this assumption was tested using a 10-km radius of the Queen’s University Biological Shapiro–Wilk tests). We then calculated mean intensity Station, near Chaffey’s Lock, Ontario, Canada (44°34′ N, based on transformed data for males and females of each Parasitol Res morph for each date-by-site combination. On two occa- parasitized (53.6% prevalence, estimate ranging from sions, no andromorphic females were collected that were 49.2% to 57.9%). The remaining 93 females (15%) were parasitized, and so, no mean intensity values were obtained andromorphs of which 38 were parasitized (40.8% preva- for those females in those samples. Using these values, we lence, estimate ranging from 30.8% to 51.5%). could compare whether males and females of both morphs At first glance, it would appear that andromorphic showed similar or dissimilar mean intensity values. We also females were less likely to be parasitized. However, on a performed a one-way ANOVA (alpha=0.05) to compare date-by-site comparison, this was not the case. No mean intensity between males and females of one or both consistent sex or morph biases in prevalence of infection morphs. In situations where no andromorphic females in a by gregarines were observed (see individual estimates of sample were parasitized, a two-tailed t test (alpha=0.05) prevalence and associated confidence intervals, Table 1). was used for this analysis. We then back transformed the The grand mean prevalence for males across 12 date-by-site mean intensity value and the upper and lower bounds of the samples was 44.7% (range, 20.5–76.2%) compared to standard error (standard errors are asymmetrical around 47.2% (range, 18.5–70%) for heteromorphic females and back-transformed means). We used Bonferonni correction 45.9% for andromorphic females (ranging from 0 based on (0.05/12=0.0042) to ascertain whether any particular single females collected to 73.9% for larger samples). Thus, samples might have shown significant differences in mean there also was no tendency for prevalence estimates to be intensity between morphs and males that could not be higher or lower in males or one of the female morphs explained by sampling error due to multiple tests. (F2, 33=0.044, p>0.95). Sample sizes were much smaller when examining intensity because only damselflies infected by one or more Results gregarines were included and sometimes no parasitized andromorphic females were collected (Table 2). Again, no In total, 1,109 damselflies were collected and dissected for consistent sex or morph biases were observed regarding gregarines. Samples were collected across each date-by-site gregarine intensity (Table 2). Across 12 date-by-site comparison with 76–126 damselflies being collected among comparisons, 11 samples failed to detect differences sites (Table 1). This resulted in adequate representation of between males and females of one or both morphs in mean males and females in order to test for sex biases in either intensity of parasitism (p values ranged from 0.11 to 0.95). prevalence or intensity of parasitism. However, in one sample where males and females of both Of the 1,109 damselflies, 486 (43.8%) were males of morphs were well represented (BM, 3 July, Table 2), which 279 were parasitized (57.4% prevalence, estimate infected heteromorphic females had significantly lower ranging from 52.8% to 61.8%). Of the females, most (530 gregarine numbers than either infected males or infected or 85%) were heteromorphic females, of which 284 were andromorphic females (Table 2). However, the overall

Table 1 Prevalence of infection in male, heteromorphic female, and andromorphic female N. irene (Hagan) over 12 date-by-site samples, for June and July 1997

Site Sample date N ♂,♀ (H, A) Male Heteromorph Andromorph Total

BM 17 June 60, 66 (65, 1) 26.6 (16.1–39.7) 18.5 (9.9–30.0) 0.0 (NA) 22.2 (15.3–30.5) 3 July 21, 67 (44, 23) 76.2 (52.8–91.8) 59.1 (43.2–73.7) 73.9 (51.6–89.8) 67.0 (56.2–76.7) 20 July 42, 34 (33, 1) 38.1 (23.6–54.4) 30.3 (15.6–48.7) 0.0 (NA) 34.2 (23.7–46.0) UD 22 June 44, 56 (46, 10) 34.1 (20.5–49.9) 47.8 (32.9–63.1) 30.0 (6.7–65.2) 40.0 (30.3–50.3) LLM 23 June 44, 57 (47, 10) 20.5 (9.8–35.3) 27.7 (15.6–42.6) 60.0 (26.2–87.8) 27.7 (19.3–37.5) 10 July 38, 48 (45, 3) 47.4 (31.0–64.2) 68.9 (53.4–81.8) 33.3 (0.8–90.6) 58.1 (47.0–68.7) 21 July 40, 59 (55, 4) 35.0 (20.6–51.7) 47.3 (33.7–61.2) 50.0 (6.8–93.2) 42.4 (32.5–52.8) EP 26 June 45, 43 (38, 5) 42.2 (27.7–57.9) 39.5 (24.0–56.6) 60.0 (14.7–94.7) 42.0 (31.6–53.0) 11 July 43, 43 (40, 3) 67.4 (51.5–80.9) 70.0 (53.5–83.4) 66.7 (9.4–99.2) 68.6 (57.7–78.2) 24 July 44, 36 (32, 4) 50.0 (34.6–65.4) 37.5 (21.1–56.3) 50.0 (6.8–93.2) 45.0 (33.8–56.5) ILB 28 June 27, 52 (41, 11) 40.7 (22.4–61.2) 48.8 (32.9–64.9) 54.5 (23.4–83.3) 46.8 (35.5–58.4) 12 July 38, 62 (44, 18) 57.9 (40.8–73.7) 70.5 (54.8–83.2) 72.2 (46.5–90.3) 66.0 (55.8–75.2)

The sample sizes (N) for males (♂) and both female (♀) morphs (H heteromorph, A andromorph) are shown. NA refers to “not applicable” because confidence intervals could not be calculated The site codes are as follows: BM Barb’s Marsh, UD Upper Dowsley Marsh, LLM Lake Lindsey Marsh, EP Elgin Pond, and ILB Indian Lake Bight Parasitol Res

Table 2 Back-transformed mean intensity of gregarine infection in males and female morphs (H heteromorph, A andromorph) of N. irene over 12 date- by-site samples for collecting dates June–July 1997

Site Sample date N ♂,♀ (H, A) Male Mean intensity (lower and upper SE)

Heteromorph Andromorph

BM 17 June 16, 12, 0 5.6 (4.3–7.3) 5.8 (4.2–7.9) NA 3 July 16, 26, 17 9.0 (7.2–11.2) 4.2 (3.5–5.1) 9.3 (7.4–11.7) 20 July 16, 10, 0 4.1 (2.5–6.4) 3.5 (2.6–4.8) NA UD 22 June 15, 22, 3 5.3 (4.1–6.9) 5.3 (4.3–6.6) 1.5 (1–2.2) LLM 23 June 9, 13, 6 2.6 (1.5–4.2) 4.7 (3.6–6.0) 8.6 (6.6–11.1) 10 July 18, 31, 1 3.2 (2.6–3.8) 4.4 (3.5–5.5) 24 (NA) 21 July 14, 26, 2 4.1 (2.9–5.6) 2.6 (2.1–3.2) 13.0 (2–64) EP 26 June 19, 15, 3 9.2 (7.3–11.7) 5.4 (4.1–6.9) 8.0 (5.0–12.5) 11 July 29, 28, 2 5.7 (4.6–6.9) 3.4 (2.6–4.3) 2.9 (2–4) 24 July 22, 12, 2 4.3 (3.0–6.0) 3.0 (2.1–4.1) 1.4 (1–2) ILB 28 June 11, 20, 6 12.3 (9.7–15.6) 9.1 (7.3–11.3) 4.3 (3.0–6.0) 12 July 22, 31, 13 10.7 (8.4–13.5) 6.0 (5.0–7.3) 7.9 (5.7–10.9)

The lower and upper range of the standard error is shown in parentheses (error is asymmetrical about the back-transformed mean). NA refers to not applicable, either because mean intensity could not be calculated (no infected individuals were present) or because standard errors could not be calculated (only one individual was infected) The site codes are as follows: BM Barb’s Marsh, UD Upper Dowsley Marsh, LLM Lake Lindsey Marsh, EP Elgin Pond, and ILB Indian Lake Bight model was not significant following Bonferroni correction and rose by mid-season, dropping later in the season but not (p>0.0042). as low as early-season levels. In comparison, the mean We therefore combined data to test for seasonal changes number of gregarines for infected individuals was already in gregarine parasitism. Here, we found unimodal patterns for both prevalence and intensity of infection (Figs. 1 and 2). Prevalence of infection was relatively low early in the season

Fig. 2 Intensity of gregarine infection in N. irene damselflies in relation to season based on log-10 transformed data (mean±2 SE). Fig. 1 Prevalence of gregarine infection in N. irene damselflies based Both prevalence and intensity peaked mid-season, although intensity on season. Shown are Clopper–Pearson 95% confidence intervals peaked slightly earlier. Note that the back-transformed means range around estimates of prevalence. The trend line is also plotted from a low of 29 gregarines to a high of 194 gregarines Parasitol Res high early in the season, but rose to the highest levels by disproportionate representation of one sex in the middle of the mid-season and dropped to lower levels later in the season. season being the reason for greater prevalence and intensity at Both patterns were repeatable across sites where early-, mid-, that time. In the absence of other information, the most and late-season samples were obtained (e.g., Tables 1 and 2, parsimonious explanation is that density-dependent transmis- Fig. 1). sion explains the unimodal pattern and that males might be more susceptible to gregarines per foraging bout. It is important to reiterate that the unimodal pattern is repeated Discussion across sites with recurrent sampling through the season. Åbro (1971) and Locklin and Vodopich (2010b)haveobserved There are two main findings in our study. The first finding this pattern in the odonate–gregarine system but had not is that, unlike previous work on another coenagrionid tested for them. damselfly (Hecker et al. 2002), females were not more These yearly unimodal patterns are not uncommon in likely than males to be parasitized by gregarines nor did other host–parasite systems, especially in infectious disease they have higher intensities of parasitism, once date and site ecology in both vertebrates and invertebrates (Altizer et al. of sampling were controlled. Female morphs also appeared 2006). They are usually due to yearly fluctuations of similar in their levels of parasitism. Ultimately, parasitism is variables such as temperature, presence of intermediate dictated both by exposure and susceptibility. With respect hosts, or breeding season. Infection in macaques by directly to exposure, we expected that females would forage more transmitted nematodes occurs in a cyclic yearly pattern than males and be more likely to ingest gregarine oocysts. based on breeding season (Macintosh et al. 2010). The We currently have little information that this is the case. It trematode, Echinostephilla patellae, infection peaks during is possible that females die sooner than males, and thus, the summer months because appropriate temperatures allow males in samples tend to be older on average than females. higher densities of the host, common limpets (Prinz et al. However, earlier work (Hecker 1999) shows just the 2010). This unimodal pattern can also be seen in Conopid opposite. When individuals are collected in the field and flies parasitizing wild bumblebee species. Conopid flies deprived of food, males die sooner than females. This parasitize more frequently during the period of the year finding supports the argument that females have more when the bumblebee hosts are at their highest population reserves from having foraged more and are able to survive densities (Gillespie 2010). In this study, density has a better than males during periods of food deprivation (e.g., putative strong role, but it is likely that it is not the sole inclement weather), but it does not resolve why males and determinant of parasitism level. The variation around females have similar levels of parasitism. It is important to prevalence and intensity estimates mid-season is still rather note that the highest densities of individuals occur at the large. As such, site-to-site variation in starting prevalence edges of ponds because ponds are the rendezvous sites, and of gregarines and/or factors favoring transmission of males should be spending more time at the edge of ponds oocysts (prey species) might be important explanatory than females (Van Gossum et al. 2007). Thus, males might variables. In fact, earlier work (Hecker 1999) failed to encounter more infective stages of gregarines per foraging show a strong relation between density and gregarine attempt or drinking bout even though they are expected to parasitism for the same species when density was measured forage less actively than females. Without further data on at the same time of season at four different sites. Within-site foraging behavior and habitat use by the sexes, we cannot relations between density and parasitism should be more know why parasitism levels are not different between males likely to detect because they have the advantage that a and females. relevant baseline is used for comparison. In fact, parasitism Our second main finding was that both prevalence and by gregarines varies substantially between populations intensity of gregarine parasitism showed unimodal distri- (Åbro 1987). butional patterns over the season. One of the underlying That N. irene showed seasonal variation in prevalence reasons for this pattern could be that, like density of hosts and intensity of infection is important. Seasonal patterns which peaks mid-season (Van Gossum et al. 2007), the occur for other insect–parasite associations involving density of infective stages of directly transmitted parasites gregarines (Zuk 1987). As mentioned, gregarines are like gregarines also peaked mid-season. It is therefore not generally accumulated over time owing to recurrent unreasonable to speculate that density-dependent transmis- ingestion of infective cysts and therefore should increase sion is highest in the middle of the season. There is no with advancing age (Åbro 1990). It is likely that both reason to presuppose that older damselflies with higher prevalence and intensity decline later in the season because gregarine loads are more likely to be collected mid-season. heavily infected individuals are lost from samples, newly If anything, they should be more common towards the end emerged individuals with no parasitism are added to of the season. Also, the work described above rules out samples, and the density of infective stages is also Parasitol Res decreased. Individual hosts which emerge early in season are Forbes MRL, Richardson JML, Baker RL (1995) Frequency of female thought to have advantages because of directional selection morphs is related to an index of male density in the damselfly, Nehellenia irene (Hagen). Ecoscience 2:28–33 for breeding onset (cf. Anholt 1992) and because infective Gillespie S (2010) Factors affecting parasite prevalence among wild parasite stages are not yet abundant in the population. Hosts bumblebees. Ecol Entomol 35:737–747 emerging later might be less susceptible to parasitism, but Hassall C, Lowe CD, Harvey IF, Watts PC, Thompson DC (2010) not realize the reproductive advantages of earlier emergence. Phenology determines seasonal variation in ectoparasite loads in a natural insect population. Ecol Entomol 35:514–522 Hecker KR (1999) Testing for sex biases and morph biases in parasitism of Zygopterans (Odonata) by Gregarines (Eugregarinidae). Dissertation, Acknowledgments We would like to thank Frank Phelan for Carleton University assistance in the field and Richard Clopton for help with gregarine Hecker KR, Forbes MR, Léonard NJ (2002) Parasitism of Enallagma identification. This was partly funded by a NSERC discovery grant boreale damselflies by gregarines: sex biases and relations to awarded to MRF. adult survivorship. Can J Zool 80:162–168 Locklin JL, Vodopich DS (2009) Bidirectional gender biases of gregarine parasitism in two coexisting dragonflies (Anisoptera: – References Libellulidae). Odonatologica 38:133 140 Locklin JL, Vodopich DS (2010a) Patterns of gregarine parasitism in dragonflies: host, habitat, and seasonality. Parasitol Res 107: Åbro A (1971) Gregarines: their effects on damselflies (Odonata: 75–87 Zygoptera). Entomol Scand 2:294–300 Locklin JL, Vodopich DS (2010b) Eugregarine parasitism of Erythemis Åbro A (1974) The gregarine infection in different species of Odonata simplicicollis (Say) at a constructed wetland: a fitness cost to from the same habitat. Zoolog Scripta 3:111–120 females? (Anisoptera: Libellulidae). Odonatologica 39:319–331 Åbro A (1976) The mode of gregarine infection in zygoptera MacIntosh AJJ, Hernandez AD, Huffman MA (2010) Host age, sex, (Odonata). Zoolog Scripta 5:265–275 and reproductive seasonality affect nematode parasitism in wild Åbro A (1987) Gregarine infection of Zygoptera in diverse habitats. Japanese macaques. Primates 51:353–364 Odonatologica 16:119–128 Prinz K, Kelly TC, O’Riordan RM, Culloty SC (2010) Temporal Åbro A (1990) The impact of parasites in adult populations of variation in prevalence and cercarial development of Echinoste- zygoptera. Odonatologica 19:223–233 philla patella (Digenea, Philophthalmidae) in the intertidal Åbro A (1996) Gregarine infection of adult Calopteryx virgo L. gastropod Patella vulgata. Acta Parasitologica 55:39–44 (Odonata: Zygoptera). J Nat Hist 30:855–859 Smith BP, Cook WJ (1991) Negative covariance between larval Arrenurus Altizer S, Dobson A, Hossein P, Hudson P, Pascual M, Rohani P species and Limnochares americana (Acari: Hydrachnidia) on male (2006) Seasonality and the dynamics of infectious diseases. Ecol Leucorhinnia frigida (Odonata: Libellulidae) and its relationship to Lett 9:467–484 the host’s age. Can J Zool 69:226–231 Anholt BR (1992) Sex and habitat differences in feeding by an adult Van Gossum H, Beirinckx K, Forbes MR, Sherratt TN (2007) Do damselfly. Oikos 65:428–432 current hypotheses explain continental and seasonal variation in Bush AO, Lafferty KD, Lotz JM, Shostak AW (1997) Parasitology female morph frequencies of the damselfly, Nehallenia irene? meets ecology on its own terms: Margolis et al. re-visited. J Biol J Linn Soc 90:501–508 Parasitol 83:575–583 Walker EM (1953) The Odonata of Canada and Alaska, vol 1. Clopton RE (1997) Hotel intestine gregarine protocol handbook. University of Toronto Press, Toronto http://science.peru.edu/gregarina/html/protocols.html.Accessed Zar JH (1996) Biostatistical analysis, 3rd edn. Prentice-Hall, Upper 15 June 1997 Saddle River Forbes MR, Baker RL (1991) Condition and fecundity of the Zuk M (1987) Seasonal and individual variation in gregarine parasite damselfly, Enallagma ebrium (Hagen): the importance of levels in the field crickets Gryllus veletis and G. pennsylvanicus. ectoparasites. Oecologia 86:335–341 Ecol Entomol 12:341–348