INTRODUCTION
The persistence of any vector-borne disease requires the co-existence of the pathogen, vector, and host. Arthropod vectors associated with transmission of human disease normally are not limited to single host species (Ostfeld and Keesing, Reference Ostfeld and Keesing2000), and feeding success, growth, and development of vectors can vary with host species. For example, engorged larval black-legged ticks (Ixodes scapularis) recovered from the striped skunk, red and eastern gray squirrel, and white-tailed deer had higher moulting success than ticks foraging on birds and the northern raccoon (LoGiudice et al. Reference LoGiudice, Ostfeld, Schmidt and Keesing2003). Unfed larval I. scapularis placed on the white-footed mouse had much higher feeding success than those placed on the Virginia opossum (Keesing et al. Reference Keesing, Brunner, Duerr, Killilea, LoGiudice, Schmidt, Vuong and Ostfeld2009). Assassin bugs (Triatoma infestans) were heavier and more engorged when feeding on domestic dogs than on cats or chickens (Gürtler et al. Reference Gürtler, Ceballos, Ordóñez-Krasnowski, Lanati, Stariolo and Kitron2009), and bloodmeal size and egg production by fleas (Parapulex chephrenis and Xenopsylla spp.) differed with hosts (Krasnov et al. Reference Krasnov, Khokhlova, Burdelova, Mirzoyan and Degen2004; Khokhlova et al. Reference Khokhlova, Ghazaryan, Krasnov and Degen2008). Finally, survival and reproduction of many mosquito species also vary with vertebrate hosts (Lyimo and Ferguson, Reference Lyimo and Ferguson2009).
Given these observations, assessing host suitability has important implications for prevention of vector-borne diseases, as it allows disease control efforts to more clearly target appropriate hosts. Moreover, vectors on unsuitable hosts may have much lower chances of surviving, moulting, or laying eggs so that the inclusion of such hosts in the community may reduce vector abundance (Keesing et al. Reference Keesing, Brunner, Duerr, Killilea, LoGiudice, Schmidt, Vuong and Ostfeld2009), a phenomenon resembling the ‘dilution effect’, in which the presence of incompetent hosts can reduce the prevalence of disease among vectors (Ostfeld and Keesing, Reference Ostfeld and Keesing2000; LoGiudice et al. Reference LoGiudice, Ostfeld, Schmidt and Keesing2003).
The observation that vectors feed on unsuitable hosts is, nevertheless, inconsistent with the evolutionary perspective that vectors should endeavour to maximize fitness. The Ideal Free Distribution (IFD; Fretwell and Lucas, Reference Fretwell and Lucas1970) predicts that competition for resources will impel animals to disperse themselves among resource patches so that at equilibrium, per capita fitness should be similar across patches. The IFD recently has been proposed to explain the marked heterogeneity of vectors distributed among hosts, and it predicts that vectors should disperse themselves such that per capita fitness is similar across hosts (Kelly et al. Reference Kelly, Mustafa and Dye1996; Kelly and Thompson, Reference Kelly and Thompson2000; Kelly, Reference Kelly2001). However, most studies on variation in host suitability have constrained access to only one or a few selected hosts (Kelly, Reference Kelly2001; but see LoGiudice et al. Reference LoGiudice, Ostfeld, Schmidt and Keesing2003; Gürtler et al. Reference Gürtler, Ceballos, Ordóñez-Krasnowski, Lanati, Stariolo and Kitron2009). As such, these studies were unable to demonstrate if unrestrained vectors would feed on unsuitable hosts and distribute themselves according to the IFD. This is notable because the IFD assumes that the cost of assessing the quality of a resource patch is negligible (Fretwell and Lucas, Reference Fretwell and Lucas1970). The IFD may thus hold for vectors with greater mobility (e.g., mosquitoes), but not for less vagile vectors (e.g., ticks, mites) due to the high costs associated with host switching.
Scrub typhus (Orientia tsutsugamushi ; OT) is an acute human infectious disease prevalent mainly in the western Pacific (Kawamura et al. Reference Kawamura, Tanaka and Takamura1995). Scrub typhus is vectored by larval trombiculid mites (i.e., chiggers), which are very limited in mobility (<45 cm, Philip et al. Reference Philip, Traub and Smadel1949), normally waiting on debris or twigs for hosts passing by (Traub and Wisseman, Reference Traub and Wisseman1974). Leptotrombidium chiggers are the primary vectors of OT, and murine rodents, especially Rattus species, are the predominant hosts of chiggers in regions with endemic scrub typhus (Traub and Wisseman, Reference Traub and Wisseman1974; Kawamura et al. Reference Kawamura, Tanaka and Takamura1995). Trombiculid mites are the only reservoirs of OT, while vertebrate hosts provide chiggers with food resources, but play little role in transmitting OT (Kawamura et al. Reference Kawamura, Tanaka and Takamura1995).
Between 2007 and 2008, we trapped rodents and collected trombiculid chiggers in a region of Taiwan with prevalent scrub typhus (Kuo, Reference Kuo2010). Apodemus agrarius was the primary host (47·82% of recovered), followed by Rattus losea (40·77%) and Bandicota indica (10·42%). Leptotrombidium imphalum dominated both among (84·47% of identified) and within host species (>80%, except Mus musculus), and is the potential vector in transmitting local inhabitants scrub typhus (Kuo, Reference Kuo2010). We quantified and compared the relative degree of engorgement (RDE) of L. imphalum recovered from different host species and different individuals of the same host species. We addressed 3 questions. First, we asked whether RDE of L. imphalum differed among host species across the study region under natural conditions. Second, because feeding success of vectors can be related to some intrinsic factors of hosts, such as age, gender, and body condition (Krasnov et al. Reference Krasnov, Khokhlova, Arakelyan and Degen2005; Hawlena et al. Reference Hawlena, Abramsky and Krasnov2007a; Khokhlova et al. Reference Khokhlova, Serobyan, Krasnov and Degen2009), we evaluated whether RDE of L. imphalum was affected by host gender, as well as host body and reproductive condition. We also evaluated whether negative density-dependent feeding success occurred under natural conditions. Finally, we tested whether RDE of L. imphalum was similar within and among host species in each trapping plot, and discussed the compliance with the IFD of this disease vector.
MATERIALS AND METHODS
Study area
We studied chiggers and their hosts in abandoned agricultural fields in the lowlands of central Hua-lien County of eastern Taiwan (23°47′N, 121°29′E). Although one of Taiwan's least populated counties, Hua-lien had the country's second highest number of human cases of scrub typhus between 1998 and 2007 (after Kin-men Island; CDC, URL: http://nidss.cdc.gov.tw/).
Small mammal trapping and collection of chiggers
Small mammals were trapped monthly from January to March 2007, and from August 2007 to March 2008 (except November 2007). In the former period, we sampled 14 plots, deploying 2 parallel transect lines containing 10 Sherman traps (26·5×10×8·5 cm) at 10 m intervals, and 2 hand-made live traps (27×16× 13 cm) at 50 m intervals. Hand-made traps were used to target the less abundant, but larger B. indica. Adjacent transect lines were separated by 10 m. In the latter period, we increased trapping effort, sampling 35 additional fields with 3 parallel transect lines containing 10 Sherman traps at 10 m intervals, and 3 hand-made live traps at 30 m intervals in each plot; adjacent transect lines were separated by 10 m. Six plots were surveyed in both periods. Plots were separated by at least 500 m. Traps were opened and baited in the evening and checked for captures early in the morning. Sites were surveyed for 3 consecutive nights throughout the course of this study.
Trapped small mammals (rodents and shrews) were transferred to a clean nylon mesh bag; bags were carefully examined to ensure that no arthropod vectors remained from earlier captures. Rodents were anaesthetized with Zoletil 50 (Fa. Virbac. Carros, France) and examined for gender and reproductive status. Reproductive activity was inferred for males if testes were scrotal, and for females if vaginae were perforate and/or nipples were swollen; otherwise, animals were considered reproductively inactive. Animals were weighed to the nearest 0·5 g. Head+ body length (snout to anus) and tail length (anus to tip of tail) were measured (mm), and the length of the ear and hind foot were recorded (0·1 mm). Morphological measurements were implemented by the same researcher to minimize individual differences in measurement bias. Rodents were examined for ectoparasites by combing their fur carefully. Skin with attached chiggers was detached with minimal injury to the animals and preserved in vials; chiggers released themselves from the skin and were transferred to 70% ethanol after 2 days. We counted the number of chiggers recovered from each host animal. Rodents were identified by fur clips and released at least 5 km away from the study areas. Shrews (Crocidura attenuata, C. suaveolens, and Suncus murinus) were screened for ectoparasites; those with chiggers were euthanized with an overdose of Zoletil 50, and blood collected via heart puncture. Those free of ectoparasites or infected only with ticks were identified with fur clips and released outside the study areas without collecting blood. Length of body, tail, ear, and hind-foot were measured as for rodents. Sex of each shrew was also recorded as above. Reproductive activity was determined only for females by the existence of swollen nipples. All procedures were approved by the University of California, Davis Animal Use and Care Administrative Advisory Committee and met guidelines recommended by the American Society of Mammalogists (Gannon et al. Reference Gannon and Sikes2007).
Chigger identification and measurement
We randomly selected at least one-fifth of chiggers from each host individual for species identification. Chiggers were soaked in deionized water for 2–3× 30-min baths, then slide-mounted in Berlese fluids (Asco Laboratories, Manchester, UK). Slides were dried at 40°C for at least 1 night, and chiggers were examined under a light microscope and identified with published keys (Wang and Yu, Reference Wang and Yu1992; Li et al. Reference Li, Wang and Chen1997).
Leptotrombidium imphalum was the most common chigger identified (84·47%), and was recovered from most small mammal host individuals (>82·65%, except Mus musculus) (Kuo Reference Kuo2010); consequently, we measured RDE only for this chigger species. Feeding leads to enlargement of the idiosoma (the body part bearing the legs) of chiggers. The idiosoma is elliptical in shape, and we measured both the length and width with an ocular micrometer. The area of the idiosoma was calculated as π(length/2)(width/2). Because the thickness of the idiosoma is difficult to assess, we only calculated area. We defined RDE as the increase in idiosoma area relative to the smallest idiosoma area measured, which was assumed to be similar to that of an unfed chigger.
Because recovering fully engorged chiggers requires caging hosts in the laboratory, which would greatly limit the number of animals that can be sampled, we detached feeding chiggers directly from the hosts. Additionally, the very small size of even highly engorged chiggers (length ca. 0·5 mm) hindered their recovery from the collecting water pans placed under the holding cage. Consequently, chiggers recovered in this study were in different feeding stages, with some close to repletion while others were just preparing to feed. To avoid confounding our conclusions by including non- or little engorged chiggers, we analysed only chiggers with RDE in the upper 25th percentile within each host individual (analyses with all chiggers included got qualitatively similar results, see Appendix 1 for details). Additionally, we excluded hosts with <2 measured chiggers in the upper 25th percentile (or <8 measured chiggers).
Statistical analysis
Individual hosts were treated as replicates, except where otherwise mentioned. Differences in RDE among host species were first compared across the entire study region. RDE for each individual host was defined as the mean RDE of chiggers recovered from that host. We confirmed normality and homogeneity of variance with Shapiro-Wilk and Levene tests, respectively, and transformed the data if necessary. We applied analysis of variance (ANOVA) or Kruskal-Wallis tests, depending on whether assumptions were or were not fulfilled. When necessary, these were followed by Tukey tests (ANOVA) or by pair-wise Mann-Whitney U tests after Bonferroni adjustment (α=0·05/n comparisons).
For each of the major hosts (A. agrarius, R. losea, B. indica), we examined whether RDE of each individual host was related to its gender, reproductive condition, and body condition. We also examined the association of RDE to chigger loads to evaluate any negative density dependence. Body condition was defined as the residual of ordinary least-squares regressions between log(body weight) and log(length of body) (Schulte-Hostedde et al. Reference Schulte-Hostedde, Zinner, Millar and Hickling2005), after controlling for the effect of gender, reproductive condition, and season. Season was categorized for each month of trapping effort as above vs below the long-term (1971–2000; Central Weather Bureau; URL: http://www.cwb.gov.tw/) mean temperature of Hua-lien (23·3°C). Pregnant females were excluded from analysis. RDE was expressed as a linear combination of explanatory variables: RDE=α + β 1*(gender) + β 2*(reproductive condition) + β 3*(body condition) + β 4*(chigger load) + ε i. Because heteroscedasticity could not be corrected for even after transformation, we estimated the 95% confidence interval for each explanatory variable by bootstrapping our data (1999 iterations, with replacement).
We tested whether RDE of chiggers was similar both among and within host species within trapping plot, considering each individual host as a habitat patch and plots as replicates, and assuming chiggers could encounter all rodents trapped within the same plot. Six plots were sampled in both sampling periods, and therefore may not be independent; we excluded these 6 plots from analysis of the first time-period but retained them for analysis of the second time-period because trapping effort was greater and more hosts were trapped in the second period. For tests among host species but within plot, we first calculated the mean RDE for each individual host, and then calculated the mean RDE for each host species. Because B. indica was very uncommon in most plots (7 plots with >1 B. indica), we only compared A. agrarius vs R. losea. Because normality and homoscedasticity could not be fulfilled even after transformation, we used a non-parametric Wilcoxon signed rank test to assess if RDE differed between the two host species (A. agrarius and R. losea) within plots.
To evaluate any intra-host species variation within plots in RDE, each host species was separately analysed with a nested ANOVA; because normality and homogeneity of variance could not be fulfilled even after transformation, we used ranked input rather than raw data, making this a non-parametric analysis. RDE of chiggers across all plots was ranked, and rank was then treated as the dependent variable, with plots as the fixed factor, and host individual as a random factor nested within plots as in parametric nested ANOVA (Conover and Iman, Reference Conover and Iman1981). Mean values are followed by ±1 s.e. All procedures were implemented in SPSS 16.0 (SPSS Inc. Chicago, Illinois, USA).
RESULTS
We captured 1393 small mammals belonging to 8 species, from which we recovered 110 760 chiggers. We identified 26 274 chiggers (23·72% of the sample), and measured (length and width of idiosoma) 19 464 L. imphalum. This sample included 10 475 L. imphalum from 483 A. agrarius, 7101 from 82 R. losea, 1744 from 35 B. indica, 74 from 4 C. attenuata, 65 from 5 C. suaveolens, and 5 from 1 S. murinus. The smallest L. imphalum (idiosoma 8·5×103 μm2) was recovered from A. agrarius.
We excluded S. murinus, C. attenuata, and C. suaveolens from further analysis because we captured few individuals infested with L. imphalum. After considering only chiggers with RDE in the upper 25th percentile within each host individual, and excluding hosts with <2 measured chiggers in the upper 25th percentile, a total of 4600 L. imphalum was analysed. This included 2519 L. imphalum from 391 A. agrarius, 1667 from 79 R. losea, and 414 from 31 B. indica.
Variation in RDE across host species and across study region
RDE was significantly different among A. agrarius, R. losea, and B. indica (H=132·1, D.F.=2, P<0·001). Pairwise Mann-Whitney U tests followed by Bonferroni adjustment (3 tests, α=0·017) revealed that engorgement differed significantly among these groups (Fig. 1). Mean RDE from chiggers on R. losea and B. indica was about 1·4x and 1·3x that from A. agrarius (Fig. 1).
Fig. 1. A comparison of mean (+1 s.e., 104 μm2) relative degree of engorgement (x104 μm2) of larval Leptotrombidium imphalum recovered from different host species. Letters above bars denote significant differences (P<0·05) and are based on a Kruskal-Wallis test followed by pair-wise Mann-Whitney U tests after Bonferroni adjustment (α=0·05/n comparisons) (only included chiggers with relative degree of engorgement in the upper 25th percentile within each host individual).
A plot of engorgement against the proportion of chiggers with greater engorgement, considering each L. imphalum (instead of host) as an independent sample, also revealed differences among the 3 host species (Fig. 2). For instance, 50·0% of chiggers from R. losea engorged more than 1·2×105 μm2, compared with only 10·0% of chiggers from A. agrarius (Fig. 2).
Fig. 2. A comparison of the proportion of larval Leptotrombidium imphalum (y-axis) greater than certain relative degree of engorgement (x104 μm2, x-axis) among 3 host species (only included chiggers with relative degree of engorgement in the upper 25th percentile within each host individual).
Relation of RDE with chigger loads and intrinsic factors of hosts
Chigger loads varied greatly within A. agrarius (0 to 918, mean=105·3±5·3), R. losea (17 to 2082 544·1± 49. 8), and B. indica (0 to 1607 288·6±58·1). Within A. agrarius, bootstrap analysis revealed a negative association between RDE and chigger loads, and no relationship with gender or with reproductive or body condition (Table 1, Fig. 3A). Similarly, for R. losea, RDE was negatively associated with chigger loads, but not to other factors (Table 1, Fig. 3B). RDE recovered from B. indica was also negatively associated with chigger loads (Table 1, Fig. 3C). Male B. indica also hosted more engorged chiggers than did females, while engorgement of chiggers was not associated with reproductive or body condition of B. indica (Table 1).
Fig. 3. Distribution of mean relative degree of engorgement (x104 μm2) of larval Leptotrombidium imphalum (y-axis) against chigger loads (x-axis) in individuals of (A) Apodemus agrarius, (B) Rattus losea, (C) Bandicota indica (only chiggers with relative degree of engorgement in the upper 25th percentile within each host individual were plotted).
Table 1. Mean and 95% confidence interval of intercept (α) and explanatory variables (β i) in relation to the relative degree of engorgement (μm2, square rooted transformed) of larval Leptotrombidium imphalum recovered from 3 major host species using 1999 bootstrap replicates with replacement (only included chiggers with relative degree of engorgement in the upper 25th percentile within each host individual)
a Females designated as 0; males as 1.
b Inactive reproduction designated as 0; active reproduction as 1.
* Means significantly different from 0.
A test of similarity of RDE among and within host species
(a) Variation in RDE across host species and within plots
For this analysis we only analysed plots with >1 individual of each host species. As such, we compared only A. agrarius and R. losea (19 plots) due to the rarity of B. indica in most plots (7 plots with >1 B. indica). A. agrarius and R. losea averaged 8·6±1·4 and 3·4±0·8 ind. per plot, respectively. Mean numbers of measured chiggers were 120·7±16·6 for each plot. Chiggers were always more engorged when retrieved from R. losea than from A. agrarius (Wilcoxon signed rank test, Z=3·82, P<0·001).
(b) Variation in RDE within host species and within plots
All 3 host species exhibited significant variation in chigger engorgement within plots. Forty-three plots had >1 captures of A. agrarius, with an average of 9·1±0·9 individuals, and 58·5±8·5 measured chiggers per plot. RDE differed among A. agrarius within plots (F=8·6, D.F.=347, P< 0·001, nested ANOVA using ranked data). Similarly, R. losea had an average of 3·4±0·8 individuals across 19 plots, and these supported 69·9±8·3 chiggers each plot, with significant variation in engorgement (F=10·00, D.F.=45, P<0·001). Finally, B. indica averaged 2·4±0·2 individuals, and 28·9±6·6 chiggers per plot, and RDE also differed significantly among B. indica within plots (F=2·90, D.F.=10, P=0·002).
DISCUSSION
RDE of larval trombiculid mites differed among host species. Larval L. imphalum recovered from R. losea and B. indica were 1·4x and 1·3x as engorged, respectively, as those from A. agrarius. In the current study, RDE was calculated based on the area of the idiosoma rather than its volume. The difference in RDE of the two hosts, however, should be even greater if we considered volume, since the latter increases as the cube of linear dimensions whereas area increases only as their square. It is therefore possible that even though R. losea hosted fewer chiggers than A. agrarius (40·77% vs 47·82% of total chiggers), the much more engorged chiggers recovered from the former species may contribute more to the subsistence of trombiculid mites. Besides, less engorged chiggers from A. agrarius may contribute disproportionally less to the chigger population such that an increase of this host species could reduce chigger numbers, especially when there is competition between A. agrarius and R. losea. A similar ‘dilution effect’ (LoGiudice et al. Reference LoGiudice, Ostfeld, Schmidt and Keesing2003) may occur in vectors with less mobility and thus more limited host selection (e.g., ticks), as opposed to vectors with more mobility (e.g., mosquitoes, fleas). More studies can corroborate this hypothesis.
Unlike differences in RDE within host species, those among host species did not appear to reflect negative density dependence. More engorged chiggers were recovered from host species with higher chigger loads (mean chigger loads: R. losea > B. indica > A. agrarius, with significant difference, Kuo et al. unpublished data). In host-parasite interactions, which generally are analogous to host-vector interactions, it is well documented that hosts vary in their defensive responses against parasites (Combes, Reference Combes2001). Our study is consistent with predictions from one such defence, grooming behaviour, which is very effective against disease vectors (Hart et al. Reference Hart, Hart, Mooring and Olubayo1992; Kelly et al. Reference Kelly, Mustafa and Dye1996; Hawlena et al. Reference Hawlena, Bashary, Abramsky and Krasnov2007c; Slowik and Lane, Reference Slowik and Lane2009). Due to the larger amount of energy reserve, larger host species are more tolerant of vector infestation, while small species are more inclined to interrupt any infestation (Hart et al. Reference Hart, Hart, Mooring and Olubayo1992; Olubayo et al. Reference Olubayo, Jono, Orinda, Groothenhuis and Hart1993). Larger species should therefore host more chiggers; additionally, chiggers are allowed to engorge longer, thus becoming more fully engorged. This agrees with our findings that larger species (B. indica: ca. 300 g; R. losea: ca. 120 g) generally hosted more chiggers, and chiggers were more engorged, while small hosts (A. agrarius: ca. 30 g) harboured fewer, and less engorged chiggers. The 2 smallest rodent species in our study site were not highly parasitized with chiggers (M. musculus, ca. 10 g, mean chigger load=0·06; M. caroli, ca. 15 g, mean chigger load=0·01; Kuo, Reference Kuo2010). Similarly, other than vector competition for resource (Hawlena et al. Reference Hawlena, Abramsky, Krasnov and Saltz2007b), negative density-engorgement relations within three major host species may be due to increased grooming activity in response to an increased vector burden (Kelly, Reference Kelly2001). Grooming behaviour may therefore be one of the mechanisms leading to the difference in RDE.
We found that chiggers did not have similar RDE, both within and among host species. Because a positive relation of the amount of food intake to egg production has been widely reported among disease vectors (Balashov, Reference Balashov1972; Sonenshine, Reference Sonenshine1991; Kelly et al. Reference Kelly, Mustafa and Dye1996; Takken et al. Reference Takken, Klowden and Chambers1998), RDE likely is a useful proxy for fitness, although at least one ectoparasite – the flea Xenopsylla conformis – engorged more but produced fewer eggs on one host (Gerbillus dasyurus) than another (Meriones crassus) (Krasnov et al. Reference Krasnov, Khokhlova, Burdelova, Mirzoyan and Degen2004). Nevertheless, the significant difference in RDE suggested that the prediction of the IFD that the per capita fitness of vectors should be similar among hosts was likely to be violated in our study.
The very limited mobility of chiggers (Philip et al. Reference Philip, Traub and Smadel1949), combined with their susceptibility to desiccation (Traub and Wisseman, Reference Traub and Wisseman1974), may render host switching costly, considering the risks of exposure when questing for hosts and the uncertainty of acquiring suitable hosts. Chiggers may therefore be less selective of hosts, remaining on the hosts even when energetic returns are relatively low, leading to violation of predictions of the IFD. This may also cause the negative density-engorgement relations observed in the 3 host species studied here. That ticks naturally recovered from some hosts had much lower moulting success (LoGiudice et al. Reference LoGiudice, Ostfeld, Schmidt and Keesing2003) could be due to similar reasons, although some species have remarkable mobility (Balashov, Reference Balashov1972), and host choice can be assisted by ascending to certain questing height (Randolph, Reference Randolph2004) . In fact, the distribution of some more mobile vectors, such as sandflies, also did not abide by IFD, but was determined more by mating opportunities (Kelly et al. Reference Kelly, Mustafa and Dye1996). Our study provided another example of possible lack of compliance with IFD. Whether the IFD holds for other disease vectors under natural conditions warrants additional research.
ACKNOWLEDGEMENTS
We thank H. Y. Wu and her students for logistical support during the field work. We are grateful to D. A. Kelt, J. E. Foley, D. H. Van Vuren, and 3 anonymous reviewers for the constructive comments and N. Willits for the valuable statistical advice on the manuscript. All procedures were approved by the University of California, Davis Animal Use and Care Administrative Advisory Committee, and complied with the law of Taiwan.
FINANCIAL SUPPORT
This study was financially supported by Taiwan Centers for Disease Control (DOH96-DC-2019).
APPENDIX 1
(All chiggers were included in the analyses instead of only upper 25th percentile engorged)
Relative degree of engorgement (RDE) across host species
A total of 19 320 L. imphalum were analysed, including 10 475 from 483 A. agrarius, 7101 from 82 R. losea, and 1744 from 35 B. indica. RDE was significantly different among A. agrarius, R. losea, and B. indica (H=123·8, D.F.=2, P<0·001). Pairwise Mann-Whitney U tests followed by Bonferroni adjustment (3 tests, α=0·017) revealed that engorgement differed significantly among these groups except between R. losea and B. indica (Fig. S1, online version only). Mean RDE from chiggers on R. losea and B. indica was about 1·4x and 1·3x that from A. agrarius (Fig. S1).
Fig. S1. A comparison of mean (+1 S.E., 104 μm2) relative degree of engorgement (x104 μm2) of larval Leptotrombidium imphalum recovered from different host species. Letters above bars denote significant differences (P<0·05) and are based on a Kruskal-Wallis test followed by pair-wise Mann-Whitney U tests after Bonferroni adjustment (α=0·05/n comparisons).
A plot of engorgement against the proportion of chiggers with greater engorgement, considering each L. imphalum (instead of host) as an independent sample, also revealed differences among the 3 host species (Fig. S2, online version only). For instance, 50·0% chiggers from R. losea engorged more than 8×104 μm2, compared with only 16·8% of chiggers from A. agrarius (Fig. S2).
Fig. S2. A comparison of the proportion of larval Leptotrombidium imphalum (y-axis) greater than certain relative degree of engorgement (x104 μm2, x-axis) among 3 host species.
Relation of RDE to chigger loads and intrinsic factors of hosts
Chigger loads varied greatly within A. agrarius (0 to 918), R. losea (17 to 2082), and B. indica (0 to 1607). Within A. agrarius, bootstrap analysis revealed a negative association between RDE and chigger loads, but no relationship to gender or to reproductive or body condition (Table S1, Fig. S3A, online version only). Similarly, for R. losea, RDE was negatively associated with chigger loads, but not to other factors (Table S1, Fig. S3B). RDE recovered from B. indica was also associated only with chigger loads, but not with the other factor (Table S1, Fig. S3C).
Fig. S3. Distribution of mean relative degree of engorgement (x104 μm2) of larval Leptotrombidium imphalum (y-axis) against chigger loads (x-axis) in individuals of (A) Apodemus agrarius, (B) Rattus losea, (C) Bandicota indica.
Table S1. Mean and 95% confidence interval of intercept (α) and explanatory variables (β i) in relation to the relative degree of engorgement of larval Leptotrombidium imphalum recovered from three major host species using 1,999 bootstrap replicates with replacement
* Significantly different from 0.
A test of similarity of RDE among and within host species
(a) Variation in RDE within plots and across host species
Only plots with >1 individual of each host species were analysed. As such, we compared only A. agrarius and R. losea (19 plots) due to the infrequent occurrence of B. indica in most plots (7 plots with >1 B. indica). A. agrarius and R. losea averaged 10·3±2·4 and 3·4±0·8 ind. per plot, respectively. Mean numbers of measured chiggers were 491·6±112·8 for each plot. Wilcoxon signed rank test revealed significant within-plot difference in RDE between A. agrarius and R. losea (Z=3·82, P<0·001). Chiggers were always more engorged when retrieved from R. losea than from A. agrarius.
(b) Variation in RDE within plot and within host species
Forty-three plots had >1 captures of A. agrarius, with an average of 10·8±1·1 individuals, and 242·4±32·8 measured chiggers per plot. RDE differed among A. agrarius within plots (F=7·02, D.F.=424, P<0·001, nested ANOVA using ranked data). Similarly, RDE differed among R. losea (F=8·30, D.F.=45, P<0·001) within 19 plots. Abundance of R. losea, and numbers of measured chiggers averaged 3·4±0·8, and 283·9±33·7, respectively. Only 7 plots yielded >1 B. indica, with a mean of 2·4±0·2 individuals, and 116·1±26·6 measured chiggers per plot. RDE also differed significantly among B. indica within plots (F=11·67, D.F.=10, P<0·001).
