INTRODUCTION
Cross-species transmission of parasites among humans, chimpanzees (Pan troglodytes) and baboons (Papio spp.) is a significant public health and conservation concern in Africa (Wolfe et al. Reference Wolfe, Escalante, Karesh, Kilbourn, Spielman and Lal1998; Wallis and Lee, Reference Wallis and Lee1999; Cooper et al. Reference Cooper, Griffin, Franz, Omotayo and Nunn2012; Gómez et al. Reference Gómez, Nunn and Verdú2013; Muehlenbein and Wallis, Reference Muehlenbein, Wallis, Russon and Wallis2014). Pedersen and Davies (Reference Pedersen and Davies2010) refer to central Africa as a ‘hotspot’ of potential cross-species infection among humans and other primates. Most of the population of Africa is at risk of infection by at least one of three common soil transmitted nematodes, Ascaris lumbricoides, Trichuris trichiura and hookworm (Pullan and Brooker, Reference Pullan and Brooker2012). All the three have been reported in chimpanzees (e.g. Zommers et al. Reference Zommers, Macdonald, Johnson and Gillespie2013; Drakulovski et al. Reference Drakulovski, Bertout, Locatelli, Butel, Pion, Mpoudi-Ngole, Delaporte, Peeters and Mallié2014) and baboons (e.g. Ravasi et al. Reference Ravasi, O'Riain, Adams and Appleton2012; Mafuyai et al. Reference Mafuyai, Barshep, Audu, Kumbak and Ojobe2013). The risk of parasite transmission between humans and primates is also of concern in conservation of endangered primates, including chimpanzees (McGrew et al. Reference McGrew, Tutin, Collins and File1989; Muriuki et al. Reference Muriuki, Murugu, Munene, Karere and Chai1998; Wallis and Lee, Reference Wallis and Lee1999; Zommers et al. Reference Zommers, Macdonald, Johnson and Gillespie2013; Muehlenbein and Wallis, Reference Muehlenbein, Wallis, Russon and Wallis2014).
Because of their associations with humans (through crop raiding and use of human refuse as food) and large populations, transmission of parasites from baboons to humans has long been identified as a particular zoonotic threat (e.g. Miller, Reference Miller1960; Goldsmid, Reference Goldsmid1974; Crockett and Dipeolu, Reference Crockett and Dipeolu1984; Muriuki et al. Reference Muriuki, Murugu, Munene, Karere and Chai1998; Weyher et al. Reference Weyher, Ross and Semple2006; Ravasi et al. Reference Ravasi, O'Riain, Adams and Appleton2012; Mafuyai et al. Reference Mafuyai, Barshep, Audu, Kumbak and Ojobe2013). Many of the same parasites have been reported in sympatric chimpanzees and baboons (Table 4 and references therein), making baboons a potential risk to chimpanzees as well as to humans (see also Cooper et al. Reference Cooper, Griffin, Franz, Omotayo and Nunn2012; Gómez et al. Reference Gómez, Nunn and Verdú2013). While the possibility that these two hosts share parasites is well established (all the parasites discussed in this paper have been found in both chimpanzees and baboons), what has not been investigated is whether, in the absence of humans, the parasite communities in these two hosts are similar with respect to prevalence, richness and community composition.
Five parasitological studies of sympatric P. troglodytes and Papio spp. groups have been conducted. Three occurred in Tanzania, East Africa: two at Gombe National Park, on Papio anubis (McGrew et al. Reference McGrew, Tutin, Collins and File1989; Murray et al. Reference Murray, Stem, Boudreau and Goodall2000), and one in Mahale Mountains National Park, on Papio cynocephalus (Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012). In Tanzania, interactions among chimpanzees, humans and baboons occur daily. The presence of humans adds a third possible host to the dynamics of infection, and human activities have many potential effects on prevalence (Brearley et al. Reference Brearley, Rhodes, Bradley, Baxter, Seabrook, Lunney, Liu and McAlpine2013). For example, forest fragmentation (Sa et al. Reference Sa, Petrasova, PomajbÍkova, Profousova, Petrzelkova, Sousa, Cable, Bruford and Modry2013) and crop-raiding (Weyher et al. Reference Weyher, Ross and Semple2006) alter parasite epidemiology in primates, and just the presence of trails used for primate field study can have an impact on prevalence (Zommers et al. Reference Zommers, Macdonald, Johnson and Gillespie2013). Comparing the dynamics of infection between P. troglodytes and Papio in the absence of the confounding effects of humans and their activities is important to understand the interaction between the three hosts when those complications are present. Two studies of P. troglodytes and Papio papio occurred in West Africa, in Niokolo-Koba National Park, Senegal (at Mt. Assirik, McGrew et al. Reference McGrew, Tutin, Collins and File1989 and at Fongoli, Howells et al. Reference Howells, Pruetz and Gillespie2011) where interactions with humans are far less likely. This report is on the populations at Mt. Assirik which, at the time of the study, was uninhabited by humans or their domestic plants and animals.
The five previous studies have other limitations in addressing the issue of whether P. troglodytes and Papio spp. are comparable hosts. Only one of these studies (Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012) presented statistical analysis comparing prevalence in the two hosts; our study makes these statistical comparisons. None of the previous studies conducted statistical comparisons of within-host richness or reported on patterns of parasite co-occurrence in mixed infections; we report on both. All of the previous studies were confined to single surveys of animals in one location, leaving open the possibility that prevalences in P. troglodytes may be more closely correlated with variance in time or space than with variance in the prevalences in Papio spp. We surveyed animals in two locations over two time periods.
In 2000, we returned to Mt. Assirik to survey the parasite communities of sympatric P. troglodytes and P. papio. We reported on data from baboons earlier (Ebbert et al. Reference Ebbert, McGrew and Marchant2013). Here we present data on the chimpanzees and use our data and those from other studies to ask if the two hosts harbour similar communities, or if, instead, host is a significant factor in explaining variation in prevalence, within-host richness, and co-occurrence. We tested three hypotheses:
-
(1) prevalence, within-host richness and co-occurrence of parasites is the same in P. troglodytes and P. papio regardless of host location (either of two valleys at Mt. Assirik);
-
(2) prevalence is the same in the two hosts regardless of census time (two periods about 20 years apart); and
-
(3) prevalence and within-group richness are the same in the two hosts regardless of study (six reports differing in the year conducted, location and methods).
METHODS
We compare the current results with those from P. papio collected at the same sites and over the same time period (Ebbert et al. Reference Ebbert, McGrew and Marchant2013). Field, laboratory, identification and statistical methods were identical to those reported previously and so are briefly summarized here.
Field
WCM and LFM collected fecal specimens of P. troglodytes versus at Mt. Assirik, Parc National du Niokolo-Koba, Republique du Senegal, between 9 March and 6 April 2000. Mt. Assirik (12° 53′N, 12° 46′W) is a low flat hill (elevation 311 m) from which flows only three water courses with year-round running water (McGrew et al. Reference McGrew, Baldwin and Tutin1981). Specimens were collected in two valleys, Lion Valley (LV) and Stella's Valley (SV), which are about 7 km apart (straight-line distance) on opposite sides of Mt. Assirik. At the time of collection no humans lived in this area, nor were there domestic plants or animals. Neither cultigens nor refuse were available to the chimpanzees, who ate only natural foods. We collected fresh specimens (n = 49) into 10% neutral-buffered formalin in the morning near sleeping sites, at 1–6 h post-deposition. Feces were deposited upon arising, usually intact on boulders or outcrops, and were undisturbed before collection. Samples were collected anonymously, but we sampled separate sleeping aggregations and so assumed the samples represent distinct individuals.
Field methods and conditions in LV were similar to those of McGrew et al. (Reference McGrew, Tutin, Collins and File1989), when 70 samples were collected from P. troglodytes. Those samples were collected between March 1976 and June 1978, in both the wet and dry seasons; our current collection was conducted during the dry season.
Laboratory
In 2001, the P. troglodytes and P. papio samples were provided to MAE as numbered specimens and so were analysed blind. We prepared 7·5 ml of homogenized feces per sample using standard protocols for formalin-ether sedimentation (e.g. Price, Reference Price1994). Formalin-ether sedimentation and zinc sulphate flotation was used in our earlier (McGrew et al. Reference McGrew, Tutin, Collins and File1989) study. Four slides were prepared from each sample. We scrutinized each slide under light microscopy at 150X, using a standardized grid to examine all the material under the cover slip (No.1, 22 × 22 mm). We identified helminth eggs and ciliate cysts using published descriptions of primate parasites (e.g. Hasegawa et al. Reference Hasegawa, Kano and Mulavwa1983; Price, Reference Price1994; Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010; Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012). To assist in identification of some parasites, we randomly chose eggs and cysts for measurement using a micrometre at either 300X or 600X.
Statistical analysis
We used JMP statistical software (JMP, Ver. 8·0·2, SAS Institute, Inc., 2008). Throughout, we judged results as significant if these analyses indicated a <0·05 (two-tailed) probability of obtaining results at random. We present means as x ± s.d. and percentages as p ± binomial error.
We compared the length and width of eggs between valleys and hosts using MANOVA and Wilk's lambda approximation of the F-distribution. We use logistic models and the significance of likelihood ratios (G) to test effects of host and valley (Mt. Assirik data), host and year (Mt. Assirik data from LV) and host and study (data across six studies) on prevalence (number of specimens infected with a particular parasitic taxon, divided by the number of specimens examined). Within-host richness (number of parasitic taxa in a specimen) and within-group richness (number of parasitic taxa across specimens within a group) had skewed distributions; we therefore used a non-parametric assessment (Wilcoxon rank test) to test for differences in richness across valleys and hosts, and across studies and hosts. We used Fisher's exact tests and data from the most common genera (n ⩾ 10) to test whether genera co-occurred within a host more often than expected from the group prevalence.
RESULTS
Parasite identification
Among our P. troglodytes samples we identified seven parasites to genus and four other taxa (Table 1). As detailed below, parasite morphology was similar to that seen in P. papio (Ebbert et al. Reference Ebbert, McGrew and Marchant2013), which supports our assumption that the two hosts were exposed to similar species within the genera.
Percent prevalence (binomial error) of parasite taxa in samples from P. troglodytes (Pan Pan) and P. papio (Papio) collected in two valleys (SV and LV) in 2000
Results are for logistic models (G, P) testing whether host, valley and their interaction effect prevalence. P values ⩾ 0·05 are indicated by NS, those <0·0001 by asterisks; other values are specified. Data for P. papio previously reported (Ebbert et al. Reference Ebbert, McGrew and Marchant2013).
Watsonius watsoni (Paramphistomata: Echinostomida) was a new record for P. troglodytes. There was no effect of host, location or their interaction on egg size (MANOVA of length and width, d.f. = 6, F = 1·5, P = 0·19 for full model). Pooling across valleys (n = 47), the P. troglodytes eggs averaged 116 ± 8·0 by 69 ± 2·6 μm. In the original description of Watsonius watsoni (Stiles and Goldberger, Reference Stiles and Goldberger1910) eggs ranged from 122 to 130 μm in length and from 75 to 80 μm in width. Little is known about this parasite (Toft and Eberhard, Reference Toft, Eberhard, Bennett, Abee and Henrickson1998) and we know of no other reports of Watsonius in wild African primates other than a report of infection in a mandrill (Pick, Reference Pick1951) which did not include a description of the eggs.
Both Trichuris and the Trichuris-like morph (referred to here as cf. Trichuris) we reported in P. papio were present in P. troglodytes. There was no effect of host, location or their interaction in the egg size of cf. Trichuris (MANOVA, d.f. = 6, F = 1·0., P = 0·47). Pooling across valleys (n = 6) cf. Trichuris averaged 37 ± 3·8 by 16 ± 1·9 μm in P. troglodytes. Because we randomly selected samples for parasite measurement, Trichuris was measured only in Papio. It averaged 55±2·8 by 25±1·6l μm, (n = 10), a size similar to that seen previously in P. troglodytes (e.g. Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010; Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012).
There were two Enterobius morphs present in P. troglodytes. The more numerous had the same length and width as those seen in P. papio: there was no effect of host, valley or their interaction on egg size (MANOVA, d.f. = 6, F = 2·0, P = 0·06 for model). This morph is referred to here as Enterobius. Pooling over locations (n = 21) it measured 37 ± 3·3 by 33 ± 2·1 μm in P. troglodytes. The second morph (cf. Enterobius), which was confined to LV and to P. troglodytes, measured an average of 56 ± 2·4 by 27 ± 1·7 μm (n = 10). It was consistent with Enterobius anthropopitheci from P. troglodytes samples in Tanzania (Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010; average 55 by 27 μm) and oxyurid eggs from Pan paniscus samples in Democratic Republic of Congo (DRC) (Hasegawa et al. Reference Hasegawa, Kano and Mulavwa1983; 49–53 by 23–25 μm).
Protospirura egg size did not differ between hosts or locations (MANOVA, d.f. = 4, F = 1·2, P = 0·34 for model; data did not allow a host by location comparison). Eggs (n = 9) measured 54 ± 4·2 by 44 ± 2·8 μm in P. troglodytes. These eggs have a thick shell covered with a hyaline coat; the coat often appeared rough. They were consistent with photos of Protospirura muricola in P. troglodytes (52 by 39 μm, Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010) and ‘cf Protospirura muricola’ (51 × 38·5 μm) described by Petrášová et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010 from the same population of P. troglodytes. We chose to identify the spirurid eggs in our samples as Protospirura because they were so similar to those described in Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010, an identification that was confirmed with adult worms. However, we note that the eggs in our samples also closely resemble those found in P. troglodytes from Gombe National Park in Tanzania (pictured, but no average measurements presented, in Gillespie et al. Reference Gillespie, Lonsdorf, Canfield, Meyer, Nadler, Raphael, Pusey, Pond, Pauley, Mlengeya and Travis2010, Fig. 2) and identified as Abbreviata caucasica. An apparently identical micrograph in Howells et al. (Reference Howells, Pruetz and Gillespie2011, Fig. 2) was labeled as Physaloptera sp. in a P. troglodytes sample collected in Fongoli, Senegal. Goldsmid and Rogers (Reference Goldsmid and Rogers1978) pictured an egg with the same thick shell and hyaline coat from Papio ursinus in South Africa (average 57 × 25·6 μm) and identified it as ‘Abbreviata (=Physaloptera) caucasica’.
We noted a second spirurid in our samples, which we did not identify further. The one egg in P. troglodytes was similar to two eggs seen in P. papio, with an undifferentiated embryo and without the hyaline coat. In P. troglodytes it measured 43 by 29 μm.
As in P. papio, two sizes of strongylid eggs were noted. The first was larger, averaging 68 ± 3·5 by 38 μm (n = 2); one egg had about 32 cells and the other about 64. These are similar to unidentified strongylids noted in Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010 (74 by 42 μm) and to Oesophagostomum eggs reported by File et al. Reference File, McGrew and Tutin1976 (85 by 50 μm), Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012 (57–93 by 37–60 μm) and Hasegawa et al. Reference Hasegawa, Kano and Mulavwa1983 (75–80 by 43–48 μm). The second (n = 1) had only about 16 cells, and measured 52 × 30 μm. It was similar to unidentified strongylids noted in Petrželková et al. Reference Petrželková, Hasegawa, Appleton, Huffman, Archer, Moscovice, Mapua, Singh and Kaur2010 (62 by 33 μm), unidentified hookworm eggs (Hasegawa et al. Reference Hasegawa, Kano and Mulavwa1983; 75–80 by 43–48 μm) and to Necator eggs reported by File et al. Reference File, McGrew and Tutin1976 (64 by 40 μm).
Although other humans were rarely seen at our field sites, the identification of a few Ascaris eggs, a parasite of humans and pigs (Roberts and Janovy, Reference Roberts and Janovy2005), raises the possibility of P. papio and P. troglodytes contact with human feces. These infections could also stem from chance encounters with red river hog (Potamochoerus porcus) or warthog (Phacochoerus aethiopicus) feces (McGrew et al. Reference McGrew, Baldwin, Marchant, Pruetz and Tutin2014).
Within-host richness
We found 15 morphologically distinct taxa in P. papio and 11 in P. troglodytes. Two taxa (cf. Enterobius and Troglodytella) were present in P. troglodytes but not in P. papio. Neither G tests (Table 1) nor Fisher exact tests (not shown) could distinguish between low prevalences in P. papio and absence in P. troglodytes of seven taxa (Streptophargus, Strongyloides; data not shown for flukes ‘A’ and ‘B’ or for nematodes ‘A’, ‘B’ and ‘C’, each identified in 1·0 ± 1·0% of P. papio samples, Ebbert et al. Reference Ebbert, McGrew and Marchant2013).
The P. troglodytes samples were less likely to be infected than the P. papio samples. All of the P. papio samples were infected with at least one parasite taxon and averaged 3·6 taxa each (Ebbert et al. Reference Ebbert, McGrew and Marchant2013); we identified three or four taxa in most hosts. In contrast, only 65·3 ± 6·8% (32 of 49) of the P. troglodytes samples were infected; samples with no patent infections were more common than any other class. Among infected samples, those with a single parasite type were most common (12 of 32, 37·5+8·6%). The mean number of taxa per P. troglodytes sample was 1·6 ± 1·7. In P. papio the average was more than twice that, 3·6 ± 1·3 taxa per sample, a significant difference (Wilcoxon rank test, P < 0·0001). Richness did not vary between locations in P. troglodytes (Wilcoxon rank test, P = 0·3) or P. papio (Ebbert et al. Reference Ebbert, McGrew and Marchant2013).
Prevalence and community composition
Prevalences differed between host species in four cases (Table 1). Balantidium and Trichuris prevalences were higher in P. papio, Watsonius prevalence was higher in P. troglodytes and Troglodytella was absent in P. papio. Streptophargus and Strongyloides were absent in P. troglodytes and at very low prevalence in P. papio; in these cases, G tests could not distinguish between the two hosts.
Location had an effect on prevalence in only a few cases. Prevalence in Enterobius and cf. Trichuris showed host by location interactions without corresponding host effects. Prevalence of Enterobius in P. troglodytes samples collected from LV was about one-fourth of that seen in SV and in P. papio from both valleys. Prevalence of cf. Trichuris was 4–10X higher in P. troglodytes samples from SV than that seen in LV and P. papio samples from both valleys.
In P. papio, a trio of species (Trichuris, Balantidium and Protospirura) was present in most of the samples, each with a prevalence of 70% or more (Ebbert et al. Reference Ebbert, McGrew and Marchant2013). In P. troglodytes, Protospirura was the most common infection (found in 55 ± 7·1% of samples), but the second most common infection (Trichuris) was only about half of this value (26 ± 6·3%). None of the other taxa was found in more than 16% of P. troglodytes samples (Table 1).
Similar to P. papio (Ebbert et al. Reference Ebbert, McGrew and Marchant2013), community composition in P. troglodytes shifted over the 22 years between sampling dates at LV (Table 2). Three genera seen in the present study were absent from the previous collection, despite the earlier study's larger (n = 70) sample size. Watsonius was absent from both hosts in the first study; Trichuris and Balantidium were previously present in P. papio, but not P. troglodytes. For reasons outlined above, we cannot be sure that the Physaloptera identified in the first study was different than the Protospirura identified in the present study, and so draw no conclusions about the influence of time on their prevalences.
Percent prevalence (binomial error) of parasite genera in samples of P. troglodytes and P. papio in LV from 1976 to 1979 (McGrew et al. Reference McGrew, Tutin, Collins and File1989, n = 70 for P. troglodytes, n = 39 for Papio)
Results are from logistic models (G, P) testing whether host, time (compared with 2000 collections, Table 1) or their interaction effect prevalence. P values ⩾ 0·05 are indicated by NS, those <0·0001 by asterisks; other values are specified.
Correlations among taxa
Contingency table analysis of taxa with sample sizes of 10 or more pointed to associations among a triad of genera, Trichuris, Watsonius and Enterobius in both P. troglodytes and P. papio (Table 3). In each case, there were significantly more samples in which either both were present or both were absent when compared with the counts expected from the group prevalence. In P. troglodytes, Trichuris and Balantidium were positively associated, with more cases of co-occurrence (6) than expected (2), and a deficit of observations with only one parasite present.
Observed vs expected occurrences of parasites in P. troglodytes (A) and P. papio (B) samples collected at Mt. Assirik
P values are from comparisons of observed vs expected using Fisher's exact test.
Comparison across studies
We compared the prevalence of 10 parasite types in sympatric P. troglodytes and Papio spp. using data from this and five other studies (Table 4).
In every case, host was a significant factor in explaining prevalence and in seven cases the direction of the effect was clear. These seven included three cases where a parasite found in one host was not found in the other: Probstmayria was absent from Papio spp. and Streptophargus and Schistosoma were absent from P. troglodytes. In two cases, infection was common in one host and found only once in the other: Balantidium was recorded only once in P. troglodytes, and Troglodytella only once in Papio spp. Oesophagostomum was always at a higher prevalence in P. troglodytes, while Trichuris was always at a higher prevalence in Papio spp. In the remaining three cases, host as a main factor was either not significant (Necator, Strongyloides, the interaction term was significant for these analyses) or weakly significant (0·04, Physaloptera, the interaction term was not significant in this case).
Within-group richness was estimated by counting all the parasitic animal taxa identified in the study and pooling across strongylids, as not all studies distinguished among the strongylid genera. Within-group richness averaged 6·3 ( ± 2·7) in P. troglodytes and 8·2 (±3·5) in Papio, a non-significant difference (Wilcoxon rank test, P = 0·26).
Studies varied in their isolation method which could affect within group-richness, however, within-group richness did not differ among studies (Wilcoxon rank test, P = 0·35). Murray et al. (Reference Murray, Stem, Boudreau and Goodall2000) was the only study to use a direct smear technique, and the only study not to use formalin-ether concentration. Three studies used flotation methods (McGrew et al. Reference McGrew, Tutin, Collins and File1989; Murray et al. Reference Murray, Stem, Boudreau and Goodall2000; Howells et al. Reference Howells, Pruetz and Gillespie2011). Two studies used only one method (formalin-ether concentration; Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012; Ebbert et al. Reference Ebbert, McGrew and Marchant2013), however, richness in these two studies did not differ from that found in the studies using two methods (Wilcoxon rank test, P = 0·12).
DISCUSSION
Our analysis contradicted each of our hypotheses, showing clear differences between P. troglodytes and P. papio as parasite hosts. At Mt. Assirik, within-host richness in P. troglodytes was lower than in P. papio, and more P. troglodytes samples were free of detectable parasites. Regardless of whether we compared across valleys (Table 1), years (Table 2) or studies (Table 4), host was a main effect in explaining prevalence in Balantidium and Trichuris: the prevalence of both was much lower in P. troglodytes than in Papio spp. Of the five common (>10% prevalence in at least one host and location) genera at our site, three (Balantidium, Trichuris and Watsonius) were more prevalent in P. papio when compared across hosts and valleys. We did not observe Oesophagostomum at Mt. Assirik; in the comparison among studies, its prevalence was consistently and significantly higher in P. troglodytes than in Papio spp. Streptophargus and Schistosoma were not found in P. troglodytes in any study, and Probstmayria was not found in Papio spp. The positive association of Trichuris and Balantidium we detected in P. troglodytes did not occur in P. papio.
Percent prevalence (binomial error) in sympatric P. troglodytes and Papio spp. populations
Parasitic taxa are those found in at least one host in at least three studies. Effect tests are results (χ 2, P) of logistic models. P values ⩾ 0·05 are indicated by NS, those <0·0001 by asterisks; other values are specified. Within-group richness (R) includes all parasitic animals identified in the study and pooling across strongylids.
a McGrew et al. Reference McGrew, Tutin, Collins and File1989, Papio papio.
b This study, Ebbert et al. Reference Ebbert, McGrew and Marchant2013, Papio papio.
c Howells et al. Reference Howells, Pruetz and Gillespie2011, Papio papio.
d McGrew et al. Reference McGrew, Tutin, Collins and File1989, Papio anubis.
e Murray et al. Reference Murray, Stem, Boudreau and Goodall2000, Papio anubis.
f Kooriyama et al. Reference Kooriyama, Hasegawa, Shimozuru, Tsubota, Nishida and Iwaki2012, Papio cynocephalus.
g d.f. = 1.
h d.f. = number of studies in which genus is present, minus 1.
i Data for ‘hookworm’.
j Strongylids not identified to genus.
k Data for Troglodytella abrassarti.
We found some similarities between the hosts. At Mt. Assirik, egg morphology was consistent between hosts: this supports our assumption that we were detecting the same parasites in both hosts. Genera that were positively associated in P. troglodytes (Trichuris, Watsonius and Enterobius) also co-occurred more than expected in P. papio. Our data and the comparison among studies suggested that host was not an important factor in explaining the prevalence of Necator, Physaloptera, Protospirura and Strongyloides. Within-group richness did not differ between the hosts when compared across studies.
The positive relationships we found among Trichuris, Watsonius and Enterobius in both hosts, and between Trichuris and Balantidium in P. troglodytes raise the intriguing possibility these co-infections can improve host habitat, perhaps through immunosuppression, excluding a common competitor, or altering the microbial community (reviews in Graham et al. Reference Graham, Cattadori, Lloyd-Smith, Ferrari and Bjørnstad2007; Pedersen and Fenton, Reference Pedersen and Fenton2007; Eswarappa et al. Reference Eswarappa, Estrela and Brown2012; Leggett et al. Reference Leggett, Brown and Reece2014).
Although we cannot be sure that the Protopirura we identified in this study differ from the Physaloptera identified earlier, we argue that the other three new genera detected in the present study represent a shift in parasite community composition. The alternative explanations, that the changes are due to the difference in the methods between the two studies, or to issues of identification, seem unlikely. The possibility that Watsonius, Trichuris and Balantidium were present in P. troglodytes but not correctly identified in the earlier studies seems remote: the eggs of Watsonius and Trichuris are large and distinctive, as are Balantidium cysts, and the latter two genera were identified in P. papio. There were two differences in methods between our studies, and both would be likely to increase the number of genera in the first study over the second. The first sampling effort was conducted over a much longer time span, 28 months, during wet and dry seasons; the present study occurred during the dry season over a period of 1 month. In the earlier study (McGrew et al. Reference McGrew, Tutin, Collins and File1989) we used two methods of preparing samples for examination. Both a longer sampling period and a second method of preparing each sample should yield more, not less, variation in the parasite community. Our first study included collections during the wet season; how this difference would affect the results is not clear. For example, we observed a decline in Strongyloides prevalence between our studies. Previous studies of P. troglodytes in Tanzania have shown both a decrease in Strongyloides during the dry season (Gillespie et al. Reference Gillespie, Lonsdorf, Canfield, Meyer, Nadler, Raphael, Pusey, Pond, Pauley, Mlengeya and Travis2010) and no effect of season on Strongyloides prevalence (Huffman et al. Reference Huffman, Gotoh, Turner, Hamai and Yoshida1997; Bakuza and Nkwengulila Reference Bakuza and Nkwengulila2009). Huffman et al. (Reference Huffman, Gotoh, Turner, Hamai and Yoshida1997) also found no effect of season on Trichuris prevalence.
We identified parasites by morphology, as is standard in the current primate literature. The lack of molecular tools for confirmation of these identifications is a serious problem, as highlighted by the issues with distinguishing among the genera Protospirura, Physaloptera and Abbreviata already mentioned. If P. troglodytes, humans and Papio spp. harbour morphologically indistinct species specialized to each primate host, then the concerns about cross-species transmission become much less pressing, although not moot. Recent studies on genetic variation in Balantidium coli (Pomajbíková et al. Reference Pomajbíková, Oborník, Horák, Petrželková, Grim, Levecke, Todd, Mulama, Kiyang and Modrý2013), Oesophagostomum (Ghai et al. Reference Ghai, Chapman, Omeja, Davies and Goldberg2014a ) and Trichuris (Ravasi et al. Reference Ravasi, O'Riain, Adams and Appleton2012; Ghai et al. Reference Ghai, Simons, Chapman, Omeja, Davies, Ting and Goldberg2014b ) have shown both specialist and generalist clades. For example, isolates of Trichuris from South Africa showed two clades shared by humans and P. ursinus (Ravasi et al. Reference Ravasi, O'Riain, Adams and Appleton2012). Isolates of Trichuris from Uganda revealed a different pattern, clustering in three clades, one specific to humans, one specific to red colobus and black-and-white colobus, and one found in every primate tested, which included humans, P. anubis and P. troglodytes (Ghai et al. Reference Ghai, Simons, Chapman, Omeja, Davies, Ting and Goldberg2014b ).
Understanding the mechanisms for the patterns we observed, and their relevance to either conservation of P. troglodytes or human health will require further study. Surveys tracking prevalence, community composition and within-host richness over extended time periods would be of particular value in determining whether our observations can be replicated. We speculate that, in those parasites that are spread via fecal contamination of food or water, host physiology is more likely to explain differences between the hosts than differences in contact rates. There are no obvious gross behavioural or environmental differences between these two primates that might make one more likely to contact a fecal-borne parasite than the other. In the open, dry and hot habitat of Mt. Assirik, both species spend most of their days on the ground, and all of their nights in the trees (Sharman Reference Sharman1981; Baldwin et al. Reference Baldwin, McGrew and Tutin1982), and their ranges overlap (LFM, WCM personal observation). In contrast to the fecal-borne parasites, contact with infective stages of parasites with intermediate hosts depends on more specific behaviours that could differ between hosts. These behaviours might include, among the parasites we identified at Mt. Assirik, differential rates of consuming particular arthropod intermediates of Streptophargus or encountering infective metacercaria of Watsonius encysted on plants.
We have shown that two closely related primates, P. troglodytes and P. papio, are significantly different as hosts of gastrointestinal parasites when compared across locations, times and studies. Of the five common genera at our site, three (Balantidium, Trichuris and Watsonius) were more prevalent in P. papio when compared across hosts and locations. This result was echoed in our comparison across studies, and suggests that baboons may be of particular public health concern in the epidemiology of Trichuris, a major human parasite found across Africa (Pullan and Brooker, Reference Pullan and Brooker2012). Our results provide a baseline for comparison in studies of the interactions among humans, baboons and chimpanzees as hosts of gut parasites. For example, we showed that individual P. troglodytes were much more likely to be free of parasites than baboons, and those that were infected carried less than half as many types of parasites as P. papio. If, contrary to our results, chimpanzees showed a higher within-host richness than baboons in an area impacted by humans, this could suggest that factors associated with human activity (e.g. fragmentation, crowding and altered diet) disproportionally impact chimpanzees.
ACKNOWLEDGEMENTS
The authors thank: Directeur, Service des Parcs Nationaux, and Conservateur, Parc National du Niokolo-Koba, for permissions and help; J. Arno, J.D. Pruetz and P. Stirling for field assistance; S.K. File for processing of 1976–79 specimens; two anonymous Parasitology reviewers whose helpful and constructive critique greatly improved this paper.
FINANCIAL SUPPORT
This work was supported by LSB Leakey Foundation (LFM and WCM, unnumbered grant) and the Philip and Elaina Hampton Fund of Miami University (LFM and WCM, unnumbered grant).
