Feeding sites promoting wildlife-related tourism might highly expose the endangered Yunnan snub-nosed monkey (Rhinopithecus bieti) to parasite transmission

An increasing number of studies have found that the implementation of feeding sites for wildlife-related tourism can affect animal health, behaviour and reproduction. Feeding sites can favour high densities, home range overlap, greater sedentary behaviour and increased interspecific contacts, all of which might promote parasite transmission. In the Yunnan snub-nosed monkey (Rhinopithecus bieti), human interventions via provisioning monkeys at specific feeding sites have led to the sub-structuring of a group into genetically differentiated sub-groups. The fed subgroup is located near human hamlets and interacts with domesticated animals. Using high-throughput sequencing, we investigated Entamoeba species diversity in a local host assemblage strongly influenced by provisioning for wildlife-related tourism. We identified 13 Entamoeba species or lineages in faeces of Yunnan snub-nosed monkeys, humans and domesticated animals (including pigs, cattle, and domestic chicken). In Yunnan snub-nosed monkeys, Entamoeba prevalence and OTU richness were higher in the fed than in the wild subgroup. Entamoeba polecki was found in monkeys, pigs and humans, suggesting that this parasite might circulates between the wild and domestic components of this local social–ecological system. The highest proportion of faeces positive for Entamoeba in monkeys geographically coincided with the presence of livestock and humans. These elements suggest that feeding sites might indirectly play a role on parasite transmission in the Yunnan snub-nosed monkey. The implementation of such sites should carefully consider the risk of creating hotspots of disease transmission, which should be prevented by maintaining a buffer zone between monkeys and livestock/humans. Regular screenings for pathogens in fed subgroup are necessary to monitor transmission risk in order to balance the economic development of human communities dependent on wildlife-related tourism, and the conservation of the endangered Yunnan snub-nosed monkey.

1. We determined if feeding sites overexpose fed monkeys to parasite, by measuring Entamoeba prevalence in fed and wild subgroups. We hypothesized that host aggregation at feeding stations might lead to higher prevalence in fed compared to wild monkeys. Little data are available to compare these two subgroups (e.g. age structures or demographic rates), nevertheless we have previously shown that fed individuals exhibit a deficit in heterozygotes and a mean relatedness two times higher in fed than in wild individuals 26  www.nature.com/scientificreports/ highly debated, some examples suggest that a relationship exists between heterozygosity and some aspects of parasitism 38 . Inbred individuals might have higher chances to exhibit homozygosity for genes involved in disease resistance, and individual heterozygosity is thus expected to be a predictor of host susceptibility 39 . We thus hypothesized that monkeys with a low heterozygosity might have the highest probability to be positive for Entamoeba. 2. One of the most obvious strategies to mitigate the negative influence of feeding sites on parasite transmission is to space feeding stations more broadly, especially by avoiding domesticated animals and human settlements 11 . We thus determined if the distance from monkeys to other hosts might be used as a proxy for parasite exposure. 3. Extending our investigations to the hosts likely to frequent the feeding stations (mainly pigs and cattle), as well as those present in the nearest village (mainly domestic chickens and humans), we sought to determine whether different Entamoeba assemblage profiles co-existed in the different host. All or some of the hosts sharing the same Entamoeba parasites might have consequences both for the conservation of monkeys but also for human health and the health of domesticated animals. The fact that Entamoeba species are not highly host-specific makes this possible.

Materials and methods
Study area and sampling. The study area covers about 82.9 km 2 in the subtropical-temperate mountain Samage Forest (part of the Baimaxueshan National Nature Reserve) in the vicinity of Xiangguqing (响古箐) and Gehuaqing (格花箐) hamlets 23,40,41 , north-west of Tacheng (Fig. 1). Here, Yunnan snub-nosed monkeys form a large group that may comprise more than 900 individuals 23,42 . Reserve officers of the Baimaxueshan National Nature Reserve provision feeding sites located near Xiangguqing and Gehuaging hamlets with food (Bryoria sp. and Usnea sp. lichens, the natural staple food of this species, collected in the neighbourhood). They move the feeding sites a few hundred meters every two to three days to simulate the natural displacement of monkeys and to minimize the behavioural impacts of feeding. During feeding sessions, tourists can easily observe monkeys, but the reserve officers strongly limit potential physical contacts by maintaining a reasonable distance between the fed subgroup and tourists. Visitors are not allowed to give food to monkeys. The wild subgroup is quite elusive and distributed over the surrounding mountains.
Faeces of Yunnan snub-nosed monkeys were collected with the assistance of the reserve officers, from December 2016 to January 2017 and then from March to May 2017. Faeces from the fed subgroup were collected on the feeding sites after the feeding sessions to avoid disturbing monkeys. Faeces from the wild subgroup were collected opportunistically in the mountains while hiking on trails with the reserve officers 43 . We carefully collected only fresh faeces, relying on their general appearance. Faecal samples were georeferenced and stored in ethanol (70%) until laboratory analyses. A total of 91 faecal samples of Yunnan snub-nosed monkeys were collected in the field between December 2016 and May 2017. Using molecular genotyping, we determined that these samples corresponded to 44 distinct fed individuals and 30 distinct wild individuals (see details in Afonso et al. 26 ).
Because a high Entamoeba spp. faecal prevalence was observed in Yunnan snub-nosed monkeys (see "Results" section), we went back to the field in May 2018 to collect faeces from domesticated animals and humans in and around the Xiangguqing hamlet, which is the nearest to the feeding sites. We visited every house, checking for the presence of pets and livestock (primarily pigs Sus domesticus, cattle Bos taurus, and domestic chickens Gallus gallus). One faecal sample was collected for each livestock species found in each house, coming from one individual or from latrines. Three human latrines were also sampled for faecal material. Livestock faeces were also collected directly on the ground in the feeding sites, with the assistance of the reserve officers. This systematic sampling lead to the collection of faecal samples in 16 pig groups, 11 individual cattle, 10 individual domestic chickens, one individual dog, and three human latrines. Due to the low sample size, the dog sampled in this study (PCR-negative for Entamoeba) was excluded from statistical analyses.

Molecular analyses.
To minimize potential contamination from the external surface of the faeces, we washed each sample (using ultra-pure water) before processing for DNA extraction. To collect faecal material for analysis, faeces were opened and 180-220 mg of stool was sampled from inside the faecal mass, avoiding the external surface. Total genomic DNA was extracted using the QIAamp Fast DNA Stool Mini Kit (Qiagen, Courtaboeuf, France) according to the manufacturer's recommendations, after a step in 600 µl of lysis buffer (ASL, Qiagen) at 56 °C during 8-12 h in order to ensure a good homogenization of faecal material. Each sample was processed independently in an automated manner using the QIAcube platform (Qiagen). DNA concentration was then measured using a NanoVue Plus spectrophotometer (Biochrom). DNA extracts were stored at − 20 °C until DNA amplification.
Microsatellite genotyping of Yunnan snub-nosed monkeys is fully detailed in Afonso et al. 26 and in the Supplementary Material S1. Individual genotypes were used to assign each faeces collected in the field to distinct individuals, then individual faecal prevalence of Entamoeba species were assessed. When genotyping revealed that several faeces originated from the same individual monkey, one faeces sample was randomly selected per individual for the subsequent analyses, after ensuring consistent results (positive/negative for Entamoeba) between replicates.
Entamoeba DNA was detected using the protocol developed by Vlčková et al. 44 . First, conventional PCR was applied to all samples. An approximatively 270 bp long region of 18S rDNA was amplified using Entamoebaspecific primers 673f. (5′-ATY AGA TAC CGT CGT AGT CC-3′) and 942r (5′-GTW CGG TCT TGG TAA GTT TTC-3′) dual-indexed following a methodology adapted from Fadrosh et al. 11    www.nature.com/scientificreports/ Data processing. Bioinformatics. Read demultiplexing and primers trimming was performed using the Cutadapt software 45 . We then performed all subsequent analyses using R 3.5.1 software (R Core Team, 2018) and the packages Ape 46 , Biostrings 47 , Dada2 48 , Decipher 49 , Phangorn 50 , Phyloseq 51 , and ShortRead 52 . We followed recommendations of Vlčková et al. 44 and Galan et al. 53 to filter sequences and perform denoising. Sequences with an expected number of sequencing errors of at least one were removed. We discarded one individual (a monkey) yielding fewer than an arbitrary threshold of 500 sequences 53 , supposing that a low number of sequences per sample might limit the completeness of Entamoeba assemblage detection 53,54 . After creating an operational taxonomic unit (OTU) table, we removed from each sample all OTUs that account for < 0.5% of overall sequences, supposing that these OTUs might be remaining chimera or incorrectly assigned sequences 53 . Finally, we removed sequences that were present in only a single sample 44 .
Taxonomic assignment and phylogenetic analyses. OTUs generated in this study were aligned with sequences from Stensvold et al. 55 and Jacob et al. 28 , using the Decipher algorithm as implemented in the Decipher R package 49 . The phylogenetic analyses included 29 partial sequences of the 18S rDNA gene in Entamoeba species. We tested which nucleotide substitution model was better suited to our sequence data by using the phangorn package in the R software 50 .
We used the terminology proposed by Stensvold et al. 55 and Jacob et al. 28 to describe Entamoeba taxonomic diversity: OTUs were assigned to an Entamoeba species when this species was previously described in terms of morphology and molecular data. Subtype (ST) is a genetic cluster within the range of diversity of a defined species, with sequence divergence within a ST being not greater than 3%. Ribosomal lineages (RL) corresponds to organisms for which ≥ 80% of the SSU rDNA gene has been sequenced and there is a divergence of ≥ 5% with known sequences, while conditional lineages fills the same criteria, except that < 80% of the SSU rDNA gene has been sequenced.
Taxonomic assignment of the OTUs generated in this study was then performed by matching BLASTn searches and phylogenetic analyses; sequences published in Stensvold et al. 55 and Jacob et al. 28 were used as a reference.
Statistical analyses. Because monkey faeces were collected during a period of six months, we tested for temporal autocorrelation in Entamoeba PCR-positivity using a Durbin-Watson test implemented in the R package lmtest. A faecal sample was considered positive for Entamoeba when at least one Entamoeba OTU was detected. Fed monkeys were expected to be more frequently positive for Entamoeba compared to wild monkeys, and we analysed separately the data for the two subgroups to avoid statistical confusion.
We determined if the faecal prevalence of Entamoeba OTUs or Entamoeba species/lineages differed among host types using Pearson's Chi-squared Test for Count Data, or Fisher's Exact Test for Count Data when data did not meet Cochran rules. Faeces samples from monkeys and other hosts were not collected during the same sampling session and were thus analysed separately to avoid incorrect interpretations due to inter-annual variations.
Entamoeba polecki was widespread in Yunnan snub-nosed monkeys (see results). Therefore, we used a logistic regression to link the logit of the probability of an individual to have a faecal sample PCR-positive for E. polecki (i.e. at least one of the four E. polecki OTU) to predictor variables: distance to livestock and humans, individual heterozygosity, and subgroup. Livestock distribution did not always coincided with human locations, and overlapped feeding sites, especially pigs and cattle. Distance to livestock and humans was then assessed for each monkey as the distance of one given faeces to the centroid of livestock and human sample locations. Individual heterozygosity in monkeys was approximated by the proportion of heterozygote loci over the 10 microsatellites used to determine individual genotypes (i.e. multilocus heterozygosity). Both distance to livestock and humans and individual heterozygosity in monkeys are confounded with the subgroups. The fed individuals are close to Xiangguqing, and they exhibit lower heterozygosities than wild individuals 26 . We thus permutated the order of the subgroup covariate among explanatory variables to proceed to model selection, in order to detect possible collinearity (see models tested Table 1). The models were compared using the Akaike's Information Criterion, corrected for small sample size (AICc 56 ). AICc differences between the best model and all other considered models (Δi = difference between AICc and the lowest AICc value) were calculated to determine the relative ranking of each possible model. The model with the lowest AICc represented the best compromise between the residual deviance and number of parameters 56 . When Δi < 2, the most parsimonious model (i.e., that with the fewest parameters) was selected.
We then searched to determine if different Entamoeba assemblage profiles co-existed in the different host types. OTU richness (i.e. the number of OTUs per faecal sample) was compared among fed and wild monkeys using a Fisher exact test for count data. OTU diversity was compared among fed and wild monkeys using a Permutational Multivariate Analysis of Variance Using Distance Matrices (PERMANOVA), with 999 permutations and Bray-Curtis dissimilarities. Finally, a Principal Coordinate Analysis (PCoA) for Bray-Curtis dissimilarities was used to visualise among-sample differentiation in relative abundances of Entmoeba OTUs.
All statistical analyses were performed using the R software, using the packages Phyloseq and Vegan 57 .
Ethics statement. Field Table 2 and Fig. 2 www.nature.com/scientificreports/ identity with E. bovis (FN666252) and was close to sequences that Jacob et al. 28 referred as "E. bovis and related lineages" common in cattle and some non-human primate species; this OTU was designated as CL10. OTUs that did not share 100% identity with available sequences were deposited on Genbank, accession numbers are indicated ( Table 2). The phylogenetic relationships of the OTUs amplified in this study suggest three clusters in our data, which coincided with different host profiles (Fig. 3): (1) an E. polecki cluster, with four OTUs amplified in faeces of monkeys, humans, and pigs, (2) a cluster comprising Entamoeba RL4 and Entamoeba CL9, amplified in faeces of cattle, monkeys, pigs and domestic chickens, and (3) a cluster of E. bovis and the related conditional lineage CL10 amplified only in faeces of cattle and pigs.
The model with the lowest AIC showed that the probability that a given individual monkey had a faecal sample PCR-positive for at least one E. polecki OTU was related to both the distance from the centroid of faecal sample locations of livestock and humans (ΔAICc = 19.01, β = − 0.0006, SE β = 0.00003, LRT P < 0.001), and individual heterozygosity in monkeys (ΔAICc = 5.53, β = − 5.62, SE β = 2.21, LRT P < 0.001; see detailed results on model www.nature.com/scientificreports/ selection Table 1). The subgroup within which monkeys were distributed (fed or wild) was not related to faecal sample positivity after taking into account these two variables (ΔAICc = 0.49). Probability rapidly decreased with distance from livestock and humans to reach only negative samples after 4000 m, and probability decreased when heterozygosity increased (Fig. 6). The highest risk of faecal positivity in monkeys was thus reached when individuals were close to livestock and humans, and had a low individual heterozygosity. These two criteria are likely to be found mainly in fed monkeys, for which the lowest values of distance and heterozygosity were recorded (Fig. 7). Conversely, individual monkeys far from livestock and humans and with high heterozygosity were mainly recorded in the wild subgroup (Fig. 7) and were predicted to have a very low probability of E. polecki PCR-positivity in their faeces.
Entamoeba polecki was the only species detected in humans in one faecal sample (Fig. 8A), while Entamoeba CL9 was the only species detected in domestic chickens (1/10 faecal sample; Fig. 8C). Pigs and cattle were more frequently PCR-positive for Entamoeba (Fig. 4B). All species/lineages were found in pig faeces (Fig. 8), with E. polecki being predominant in this host. Conversely, E. polecki was not found in cattle faeces, in which Entamoeba RL4 and E. bovis were frequently found (Fig. 8).
Entamoeba OTU co-occurrences and assemblages in hosts. Several Entamoeba OTUs frequently co-occurred in faecal samples. The co-occurence of more than one Entamoeba OTU in PCR-positive samples was higher in in fed monkeys (88.64%) than in wild monkeys (33%), and was observed in 54.5% of cattle, and 50% of pigs. OTU richness was higher in fed than wild monkeys ( Fig. 9A; Fisher's exact test for count data, P < 0.001) and was 2.77 in fed monkeys and 0.53 in wild monkeys on average. Consequently, co-occurrence of Entamoeba species/lineages was observed more frequently in fed than wild individuals (Fig. 9B). Entamoeba OTU diversity in positive samples (i.e. with at least one Entamoeba OTU) did not significantly differed between fed and wild monkeys (PERMANOVA, F = 2.15, df = 1 and 47, P = 0.121). Co-occurrences were frequently observed in cattle and pigs, regarding both OTU richness and Entamoeba species/lineages (Fig. 9C,D).
PCoA analysis confirmed that Entamoeba OTU assemblages overlapped in fed and wild monkeys (Fig. 10A): wild monkeys showed assemblages similar to some of the OTU profiles observed in fed monkeys. Cattle (and one PCR-positive faeces of domestic chicken) formed a group relatively differentiated from pigs (Fig. 10B).

Discussion
In this study, we took advantage of the recent development of high-throughput sequencing for the PCR-diagnosis of Entamoeba to investigate parasitism in a local host assemblage strongly influenced by wildlife-related tourism via food provisioning. The 13 Entamoeba OTUs found in this study were distributed in three phylogenetic clusters, no one of these clusters being specific to a given host. We assigned these OTUs to E. polecki, Entamoeba RL4, E. bovis-related lineages, and we named two conditional lineages (CL9 and CL10) that have not been yet reported.
In Yunnan snub-nosed monkeys, the individual faecal prevalence of Entamoeba OTUs was higher in fed (89%) than in wild monkeys (33%). We also found co-occurences of Entamoeba OTUs more frequently in the fed subgroup, all these OTUs being assigned to E. polecki or Entamoeba RL4. These elements suggest that fed individuals face a higher exposure to parasite transmission than wild individuals. While E. polecki is commonly reported in non-human primates, Entamoeba RL4 has only been found in cattle previously 28,55 . Our study is to www.nature.com/scientificreports/ our knowledge the first report of this lineage in faeces of non-human primates and pigs. One possible explanation is that Yunnan snub-nosed monkey and pigs might carry the parasite after ingestion from an environment contaminated by cattle droppings. The fact that cysts of several Entamoeba species were detected in environmental samples (soil and water) in previous studies 58 supports this hypothesis. It is also possible that monkeys and pigs ingest other host faeces (particularly from cattle grazing in feeding sites), either accidentally or on purpose. However, DNA amplified in host faeces might actually originate from soil and have contaminated faeces before sampling collection. If we assume that Yunnan snub-nosed monkeys or pigs carried the parasite in their digestive tract, we also do not know if they are a natural host for Entamoeba RL4 or if they are transient hosts.
In Yunnan snub-nosed monkeys, the probability for a faecal sample to be PCR-positive for at least one E. polecki OTU was related to the distance from livestock and humans, and to monkey individual multilocus heterozygosity. This probability was the highest when faecal samples were collected close to livestock and humans, and when individual heterozygosity was low. Multilocus heterozygosity has already been related to parasite infection likelihood in a variety of hosts and parasites 38 . Although highly debated, multilocus heterozygosity is believed to be linked to individual fitness 59,60 . In our case, fed monkeys show low genetic diversity and high relatedness 26 , which would theoretically lead to decreased chances of surviving disease. However, the distribution of individual heterozygosity overlaps with a gradient of domesticated animal and human density. Our sampling design does not allow to sort out this covariation, and further investigations should be conducted to determine if there is a higher risk of diseases in the fed subgroup due to fitness variability.
Our main hypothesis to explain the high prevalence for Entamoeba observed in the fed subgroup is that host aggregation at feeding sites can promote parasite transmission, and increase inter-specific contacts. Rather than study the distance to feeding sites, we focused on the distance to livestock and human settlements and found here www.nature.com/scientificreports/ a proxy which might be used to determine a reasonable distance between anthropized areas and places where feeding sites can be implemented (Fig. 11). We believe that if this measure is repeated in other systems, it can help managers to mitigate the bidirectional risk of disease transmission due to a wild-domestic interface. Here, we found that the probability of finding a faeces positive for E. polecki in monkeys rapidly decreased with the distance from livestock and humans. Faeces of wild monkeys are often difficult to collect, due to evasiveness in large and scarped mountain, and we probably do not have enough data to support an empirical cut-off value. Nevertheless, the few faeces from wild monkeys collected over 4 km from livestock and humans were all Entamoeba free. In this study, E. polecki was found in faeces of Yunnan snub-nosed monkeys, pigs, and humans. These results are consistent with previous studies which all concluded that E. polecki ST1 to ST4 are not host-specific and are generally found in these hosts 28,35 . Entamoeba polecki (also refered as E. chattoni in non-human primates) is commonly found in non-human primates, especially in Asia, where it is not rare to observe high prevalence 27,61,62 . Entamoeba polecki is also a common parasite of domesticated pigs 28 . Here, we show evidences that humans, domesticated and wild animals can all be exposed to the same parasite. However, the fact that some habituated individuals of the endangered Yunnan snub-nosed monkey were highly exposed to Entamoeba, including an Entamoeba species also found in humans and domesticated animals raises some questions for the conservation of this group of monkeys. Even if the Entamoeba species found in this study have not been related to infectious diseases in hosts, the pathogenicity of Entamoeba species is largely unknown and has never been explored in the case of the Yunnan snub-nosed monkey. Moreover, if we assume that feeding sites overlapping livestock distribution ranges are local hotspots of interspecific parasite transmission, parasite exposure for these monkeys may concern a wide range of parasites (including bacteria, viruses, helminths, and other protozoa). To some extent, our study is a model of how contacts between wildlife and domestic animals and humans reinforce interspecific exchange of parasitic organisms. As distance to other hosts seems to be a determinant of positivity for parasites, www.nature.com/scientificreports/ we recommend avoiding overlapping grazing areas and feeding sites. More generally, a systemic Ecohealth approach should be considered to ensure both conservation of Yunnan snub-nosed monkey and the health of human and domesticated animals, which are inseparable in this social-ecological system.