Cryptosporidium infections in terrestrial ungulates with focus on livestock: a systematic review and meta-analysis

Background Cryptosporidium spp. are causative agents of gastrointestinal diseases in a wide variety of vertebrate hosts. Mortality resulting from the disease is low in livestock, although severe cryptosporidiosis has been associated with fatality in young animals. Methods The goal of this systematic review and meta-analysis was to review the prevalence and molecular data on Cryptosporidium infections in selected terrestrial domestic and wild ungulates of the families Bovidae (bison, buffalo, cattle, goat, impala, mouflon sheep, sheep, yak), Cervidae (red deer, roe deer, white-tailed deer), Camelidae (alpaca, camel), Suidae (boar, pig), Giraffidae (giraffes) and Equidae (horses). Data collection was carried out using PubMed, Scopus, Science Direct and Cochran databases, with 429 papers being included in this systematic analysis. Results The results show that overall 18.9% of ungulates from the investigated species were infected with Cryptosporidium spp. Considering livestock species (cattle, sheep, goats, pigs, horses and buffaloes), analysis revealed higher Cryptosporidium infection prevalence in ungulates of the Cetartiodactyla than in those of the Perissodactyla, with cattle (29%) being the most commonly infected farm animal. Conclusions Overall, the investigated domestic ungulates are considered potential sources of Cryptosporidium contamination in the environment. Control measures should be developed to reduce the occurrence of Cryptosporidium infection in these animals. Furthermore, literature on wild populations of the named ungulate species revealed a widespread presence and potential reservoir function of wildlife.


Background
Cryptosporidium, the causative agent of cryptosporidiosis, is an ubiquitous protozoan parasite. It causes gastrointestinal disease in a wide variety of vertebrate hosts, including ungulates of the orders Artiodactyla and Perissodactyla, as well as humans. Several Cryptosporidium species are known to be zoonotic with animals as major reservoirs [1]. In resource-limited settings, cryptosporidiosis is a leading cause of diarrhoeal death in children younger than five years across the globe, only second to rotaviral enteritis [2]. Cryptosporidiosis is also a significant contributor to health care cost in developed countries. It is estimated that in the USA 748,000 cases of human cryptosporidiosis occur annually [3]. Residents of and travelers to developing countries may be at greater risk of infection due to poor water treatment and food sanitation [4,5]. Cryptosporidiosis typically induces self-limiting diarrhea in immunocompetent individuals, but the infection can be severe and life-threatening in immunocompromised subjects [6]. It is one of the most important diseases in young ruminants, especially neonatal calves [7,8]. The clinical presentation of cryptosporidiosis varies from asymptomatic to deadly, leading to important economic losses due to growth retardation, reduced productivity and mortality [9,10]. Considering that an infected bovine calf can shed up to 1.1 × 10 8 oocysts per gram of feces at the peak of the infection, cattle (and very likely wild ruminants) are significant contributors of environmental Cryptosporidium oocysts [11,12], causing water-borne [13][14][15] and foodborne [16,17] diarrhea outbreaks in humans worldwide. The worldwide annual excretion of Cryptosporidium spp. oocysts by livestock has been calculated to be 3.2 × 10 23 [18], with cattle being the host species causing most environmental contamination. Cattle are able to carry different species including C. hominis which implies an associated significant public health risk [19]. In addition, Cryptosporidium oocysts are infective at the time they are passed in feces and are highly resilient to a wide range of environmental factors including disinfection and water treatment processes. Moreover, low infection doses are sufficient to cause disease in suitable hosts, e.g. 10-100 oocysts are described to provoke diarrhea in humans [20,21].
Over the past few decades, a major subject of debate and controversy in the epidemiology of Cryptosporidium is whether, and to what extent, domestic and wildlife species may act as natural reservoirs of human cryptosporidiosis [22,23]. This is principally due to the fact that the genus Cryptosporidium encompasses nearly 40 valid species with marked differences in host range, among which over 10 (mainly C. hominis, C. parvum and C. meleagridis) have been reported in humans [24] with a variety of genotypes being zoonotic [1,22,25]. The public health significance of animal cryptosporidiosis varies greatly depending on factors such as geographical variation in prevalence and genotype distribution, seasonality, load of environmental contamination with oocysts and access to surface waters intended for human consumption or recreation [9,26]. In particular, genotyping data from epidemiological surveys conducted globally indicate that infected calves are the major reservoir for zoonotic C. parvum in many areas [26,27], with lambs, kids and foals being potential additional sources of C. parvum infection for humans in some areas of the world [28][29][30][31]. Pigs are only sporadically infected with zoonotic Cryptosporidium species and are therefore considered minor contributors to the zoonotic transmission of cryptosporidiosis in humans [32]. Adult livestock typically harbor low level and asymptomatic infections but are epidemiologically important as cryptic carriers of the parasite, enabling re-infections at the herd level. Little is known of the molecular epidemiology and transmission cycles of cryptosporidiosis in wild ungulates. However, recent surveys have revealed the presence of C. parvum in wild hoofed species including the American mustang (Equus ferus caballus) [33], Scottish roe deer (Capreolus capreolus) and red deer (Cervus elaphus) [34], and Spanish wild boars (Sus scrofa scrofa) [35], which may represent a threat to water quality and public health [34].
In the present study, we conducted a systematic review of publications on the prevalence of Cryptosporidium infections and Cryptosporidium species distribution in domestic and wild ungulates in order to ascertain the extent to which hoofed animals should be considered as relevant reservoirs of human infection.

Search strategy
To evaluate the prevalence of Cryptosporidium infection in hoofed animals, we performed a comprehensive review of literatures (full text or abstracts) published online. English databases including PubMed, Scopus, Science Direct and Cochran were searched for publications related to Cryptosporidium infection of animals worldwide, from 1984 to 2016. We used the following MeSH terms alone or in combination: "Cryptosporidium" or "cryptosporidiosis" and "prevalence" and "livestock" or "cattle" or "buffaloes" or "sheep" or "pigs" or "camels" or "alpacas" or "horses" or "ruminants" or "wildlife". To identify additional published articles, we used the PubMed option of "related articles" and checked the reference lists of the original and review articles. The more agricultural and veterinary focused database CAB abstracts was searched using the following search terms: "Cryptosporidium" or "cryptosporidiosis" and "prevalence" and "cattle" or "cows" or "calves" or "buffaloes" or "sheep" or "lambs" or "goats" or "kids" or "camels" or "alpacas" or "crias" or "llamas" or "pigs" or "piglets" or "horses" or "foals" or "deer" or "fawns" or "farm animals" or "ruminants" or "livestock" or "wildlife". A protocol for the literature review was devised ( Fig. 1) in accordance with the PRISMA guidelines [36] (Additional file 1: Table S1).

Inclusion and exclusion criteria
As part of the eligibility for inclusion, titles that suggested the topic Cryptosporidium in domestic and wild hoofed animals were selected. The abstracts from the selected reference titles were reviewed by two independent reviewers to determine if the studies met the inclusion criteria and, if so, the entire articles were reviewed in full. If more than one report was published from the same study, only one was included. Exclusion criteria included studies only on human cryptosporidiosis or case reports. Studies on epidemiology of Cryptosporidium spp. in groups unrelated to hoofed animals, or studies presenting overall prevalence estimates, where samples were collected from the ground, and data from each animal were not independently retrievable, were also excluded. The language of data collection was limited to English. In order to provide contemporaneous and representative estimates, studies were excluded if they presented data collected prior to 1984. On several occasions, we contacted the authors for the collection of raw data.

Data extraction and tabulation
A data extraction form was used to collect the following data from each study: first author, year of publication, location of study, period of study, host species, age range, clinical signs (diarrhoeic versus non-diarrhoeic), population nature (e.g. domestic, captive or wild), total number of fecal samples, utilized detection method (conventional microscopy, CM; immunofluorescence antibody test, IFA; enzyme-linked immunosorbent assay, ELISA; immunochromatographic test, ICT; quantitative latex agglutination, QLAT; and polymerase chain reaction, PCR), number of Cryptosporidium-positive samples and identity of Cryptosporidium species and genotypes.

Retrieving sequences and phylogenetic analyses
To examine the genetic relationships among Cryptosporidium spp. (C. hominis, C. felis, C. parvum, C. erinacei, C. xiaoi, C. ryanae, C. scrofarum, C. muris, C. andersoni, C. ubiquitum, C. bovis and C. suis) in ungulates, a phylogenetic tree was constructed using the program Splits Tree v.4.0 based on the Neighbor-Net method and Median-Joining analysis of sequences of the 18S rRNA gene [37]. For this purpose, the sequences of the 18S rRNA gene of these Cryptosporidium spp. were retrieved from the GenBank database in the FASTA format. These sequences were initially obtained from various herbivores, including cattle, buffaloes, yaks, camels, goats, sheep and deer, as well as pigs.

Meta-analysis
A meta-analysis was performed for studies describing Cryptosporidium infection prevalence in domestic animals that are common in many parts of the world, i.e. cattle, sheep, goats, buffaloes, horses and pigs. This analysis was performed to enhance knowledge on the potential role of livestock in zoonotic Cryptosporidium Table 1 Summarized Cryptosporidium prevalence data for major domestic farmed animals. Data for wild populations of the given species not included (see for full datasets and other host species in Additional file 2:  transmission since these animals feature a close contact to humans. The pooled prevalence of Cryptosporidium infection as well as its 95% confidence interval (CI) was calculated for each study. A forest plot was generated to display the summarized results and heterogeneity among the included studies. To ensure comparable sensitivity of tests used in analyzed studies, only results from studies based on PCR as a diagnostic method were included. Studies using PCR methods only for molecular Cryptosporidium species/genotype identification but utilizing alternative diagnostic methods to determine prevalence were not included. The heterogeneity was expected in advance and statistical analyses including I 2 and Cochrane's Q test (with a significance level of P < 0.1) were used to quantify these variations. The meta-analysis considering the random effects model [38] was performed using the Stats Direct statistical software (http:// www.stats direc t.com).

Results
The initial database search retrieved 14,970 publications. The screening of these records enabled us to exclude 14,456 studies due to not meeting the inclusion criteria. Altogether, 514 studies were retained for further investigation. During the secondary assessment of these papers, another 85 were excluded because of one of the following reasons: other host species including wild hoofed animals; report of the same results as another paper published by the same author; and language of publication (e.g. Chinese, Spanish, etc.). Papers evaluating cryptosporidiosis in camels, yaks, donkeys, alpacas and llamas were excluded in the secondary analysis of data, as the meta-analysis focused on Cryptosporidium infection in cattle, sheep, goats, pigs, buffaloes and horses. Eventually, 429 studies which evaluated Cryptosporidium infection during three decades met our eligibility criteria and were retained for analysis ( Fig. 1).
In total, 196,638 stool samples from Artiodactyla and Perissodactyla ungulates were evaluated, of which 37,206 (18.9%) subjects were positive for Cryptosporidium infection. Among the 196,638 stool samples, 90,744 were associated with the domestic hoofed animals (including camels, yaks, donkeys, alpacas and llamas), displaying a Cryptosporidium infection prevalence of 13.6% (n = 12,377) ( Table 1 and Additional file 2: Table S2).
All subsequent analyses included only the studies that focused on Cryptosporidium infection in cattle, sheep,   Table 2). Analysis of available data by regions (continents and New Zealand) showed a moderate geographical variation of observed prevalence (Table 1). Although diagnostic tests varied among regions, the observed prevalence mostly fell within the 5-30% range ( Table 2). Regarding cattle, a considerably lower maximum prevalence was seen in New Zealand compared to other regions. Cryptosporidium prevalence in goat tended to be lower in Asia; however, only one study was available for Australia. For sheep it was the highest in the regions with most intensive sheep production, i.e. Australia, Europe and North America (Table 1). Cryptosporidium prevalence in pigs was the highest in Asia, Africa and Europe. In horses, The forest plot diagrams of prevalence of Cryptosporidium infection in domestic hoofed animals derived from studies using a PCR method are shown in Figs. 2, 3, 4, 5, 6, 7. As forest plots show, there is a considerable variation of study numbers and observed prevalence in a given host species within each defined geographical region, even if only studies based on PCR methodology are included. Considering a wider range of studies, i.e. studies that use either CM or PCR (Table 2), cattle are most commonly infected globally while horses feature the lowest Cryptosporidium prevalence.
The highest and lowest prevalence rate of Cryptosporidium infection in domestic hoofed animals was observed in America (26%) and Africa (14%) continents, respectively (Table 3, Fig. 8). Among 53 countries with data, Canada (60%) showed the highest infection rate whereas China, Thailand and Germany (8%) had the lowest infection rate (Table 3, Fig. 8).
The distribution of Cryptosporidium species/genotypes by host and geographical region is summarized in Table 4. Cryptosporidium parvum (monoinfections 4172/10,583; 39.4%) and C. andersoni (monoinfections 1992/10,583; 18.8%) were the most commonly detected Cryptosporidium species (Table 4). A phylogenetic network was constructed based on sequences of Cryptosporidium spp.
( Fig. 9) using the Neighbor-Net method. On the basis of this phylogenetic analysis, 10 clades (I, II, III, IV, V, VI, VII, VIII, IX and X) containing 12 Cryptosporidium spp. were identified (Fig. 9). Interestingly, C. andersoni and C. muris were placed together in Clade I, and C. xiaoi and C. bovis were both placed in Clade III. It further demonstrated a pairwise sister relationship between clades III and IV (clustering C. xiaoi, C. bovis, and C. ryanae), VI and VII (containing C. ubiquitum and C. suis) and VIII and IX (containing C. hominis and C. erinacei), respectively. Interestingly, the result of the phylogenetic analysis indicated that clades II (C. scrofarum), III (C. bovis and C. xiaoi) and IV (C. ryanae) could have originated from a common ancestor. The distribution of Cryptosporidium spp. in a wide range of domestic and wild ungulates is presented in Table 4. The C. parvum is the most common genotype in cattle (54.1%), goats (42.1%) and horses (40.2%), followed by C. ryanae in buffaloes (66.6%), C. suis in pigs (54.1%), and C. xiaoi in sheep (48.9%). In terms of transmission dynamics and clinical importance of zoonotic Cryptosporidium spp., C. hominis, C. parvum, C. andersoni, C. bovis and C. ubiquitum were identified in sheep/goats, cattle/goats/horses/pigs/ sheep, cattle/camels/sheep/yaks, buffaloes/cattle/sheep/ pigs/red deer and alpacas/buffaloes/cattle/goats/impalas/ sheep/red deers, respectively (Table 4).

Discussion
In this systematic review and meta-analysis, we found that 18.9% of the overall populations of the investigated ungulate species were infected with Cryptosporidium spp. Our study showed that although the prevalence of Cryptosporidium infection was higher in ungulates of the Cetartiodactyla than in Perissodactyla, the prevalence in the latter was not negligible and needs to be considered in terms of pathogen transmission and cycling. From the data collected and summarized on wild animals (as included in Table 4, and Additional file 2: Table S2), it is obvious that sylvatic cycles play a major role in Cryptosporidium transmission. Wild terrestrial ungulates are likely serving as important reservoir for the parasite, and certainly the infection of livestock and humans may occur by contact to wildlife feces. For meta-analysis, worldwide Cryptosporidium prevalence and species/genotype identity common livestock species have been scrutinized. Overall, Cryptosporidium prevalence in farmed animals is the highest in the Americas and Europe (Table 3) which could be attributed to the intensive farm animal production in these regions. More specifically, considering domestic farm animals, the pooled prevalence of equine Cryptosporidium infection was 4.7%, compared to the pooled prevalence of 29.1%, 26.0%, 24.4%, 22.6% and 8.2% in cattle, buffaloes, sheep, pigs and goats, respectively. Regarding the number of studies published for the different geographical regions, our analysis does not support under investigation of certain regions (e.g. Asia) as cause of a detection bias. This reinforces the suggestion that animal production intensity affects the prevalence of Cryptosporidium spp. Concentrated animal feeding operations (CAFOs) are most common in cattle and pigs. For example, in the USA, in 2002 more than 71% of all produced beef were derived from operations holding more than 5000 heads of cattle each. It is known that CAFOs pose a major problem due to the high amounts of manure that are released to the environment, facilitating potential pathogen transmission to humans, wildlife and other agricultural operations [39]. Furthermore, pathogen transmission within a CAFO seems much more likely than in more extensive farming systems. Accordingly, a high prevalence of Cryptosporidium was observed in animals from countries with many CAFO operations, especially in studies in Asia and Europe, with both regions harboring the majority of the commercial pig raising industry [40]. High prevalences in pigs in Africa may be attributed to the opposite effect of extensive farming with high exposure to environmental contamination. Other host animals displaying a high prevalence, such as buffaloes and sheep, are also generally kept in larger groups on commercial operations. The comparatively low prevalence rates in equines and goats may potentially result from smaller animal groups and free-range nature of the animal management. Between wild and domestic animals, it appears that Cryptosporidium prevalence is lower in wild populations than in farmed populations in the same host species. For example, Zahedi et al. [41] reported Cryptosporidium infection rates of 30% in farmed buffalo but 12% in wild buffalo. This suggests that animal density and confinement to the same (contaminated) environment facilitate Cryptosporidium transmission in domestic animals, and there is no clear host species disposition in terms of general susceptibility to infection with the genus Cryptosporidium despite the observed variation in Cryptosporidium infection rates among host species (Table 4).
Cryptosporidiosis in ungulates, especially ruminants, has several economic and health implications. Cryptosporidiosis in neonatal calves can lead to profuse watery diarrhea, loss of appetite, lethargy, dehydration and even death, thus may require costly treatments [42]. Moreover, as shown in sheep and goats, cryptosporidiosis can exhibit long-term effects on the growth of animals [43,44]. Additionally, infected calves can shed over 1 × 10 10 oocysts each day, which can survive in the environments for months. The ingestion of very few oocysts can cause infection in susceptible hosts, including humans [23,45]. a One study was performed in these countries It has been shown that the median infection dose of C. parvum for humans range from below 10 to over 1000 oocysts [22]. Zoonotic transmission of Cryptosporidium spp. can easily occur seasonally from young animals such as bovine calves to humans, frequently as an occupational hazard [45,46]. Nearly 40 Cryptosporidium species have been recognized based on molecular, morphological and biological characteristics of the parasites. Previous studies have shown that four major species are responsible for bovine cryptosporidiosis, namely C. parvum, C. andersoni, C. bovis and C. ryanae [1]. We showed that the most prevalent Cryptosporidium species in ungulates are C. parvum and C. andersoni, comprising 39.4% and 18.8% of detected parasites, respectively.
The data also suggest that some Cryptosporidium species are shared among ungulate hosts (Table 4). This indicates the occurrence of some inter-species transmission of Cryptosporidium spp. among ungulate species, making wildlife an important reservoir for infections in domestic animals. Currently, most data on the distribution of Cryptosporidium species and genotypes are available on domestic animal populations. Amazingly, there are clear differences in the distribution of Cryptosporidium species within the same host species among geographical regions. For example, studies from Ethiopia and Nigeria indicate that C. andersoni and C. bovis are the most prevalent species in cattle. In contrast, in countries with concentrated animal feeding operations (CAFO) such as Australia, Iran, Japan and New Zealand, as well as many European and North American countries, C. parvum is prevalent in cattle (Table 4). Similarly, alpacas in their region of origin are mostly infected with C. parvum and C. ubiquitum, while alpacas in the UK only tested positive for C. parvum (Table 4). Calves, lambs and goat kids in areas with more human activities can even have C. hominis infections [19,41,47,48]. Thus, it might be speculated that husbandry systems and contact to other livestock and humans strongly influence the distribution of Cryptosporidium species in an ungulate population.
There are geographical differences in the estimated pooled prevalence of Cryptosporidium infection. The prevalence was highest in the continent of America, followed by Europe, Australia, Asia and Africa. Canada had the highest prevalence among countries. Study design, time of sampling, age of animals, and conditions of keeping animals are other factors that can contribute to the observed heterogeneity in cryptosporidiosis prevalence, in addition to the nature of animal management.
The outcome of our study is probably affected by the publication bias. Publication bias occurs when the results of studies affect the likelihood of their inclusion in the systematic review and meta-analysis [49]. Our systematic review was limited to studies published after 1984 in English. Moreover, many studies did not provide enough information to be included in the meta-analysis.

Conclusions
Results of the meta-analysis suggest that Cryptosporidium infection is highly prevalent in ungulates, especially ruminants. Geographical differences in Cryptosporidium prevalence and distribution of Cryptosporidium species are seen for most domestic ungulate hosts. These within-host-species differences could be partially attributed to differences in animal management among geographical regions. The highest prevalence in farmed ungulates occurs in America and Europe where CAFO is widely practiced. The major farm animal hosts of Cryptosporidium spp. appear to be cattle, buffalo, sheep and pigs. These farm animals are potent reservoirs for a variety of Cryptosporidium species. Cryptosporidium prevalence is also clearly higher in farmed animals than in wild ungulate populations. Inter-species transmission of Cryptosporidium spp. appears to be affected by contact with other host species (humans or other animals) and infection pressure (intensive farming), rendering the investigated ungulate hosts capable of propagating both zoonotic and nonzoonotic Cryptosporidium species. Fig. 9 The phylogeny of Cryptosporidium spp