Rift Valley Fever: Does Wildlife Play a Role?

Abstract

Introduction

Rift Valley fever virus (RVFV) is a vector-borne pathogen that is known to affect multiple species, including wildlife, domestic ruminants, and people. Once confined to Africa, it has spread to the Arabian peninsula (Balkhy et al. 2003) and threatens other regions in the world where competent vectors exist (Turell et al. 2008). Despite a significant amount of research focused on periodic outbreaks, the role wildlife plays in the ecology of the virus continues to be undescribed, particularly whether or not there is a vertebrate reservoir host. The clinical epidemiology of the virus in wildlife is poorly represented in peer-reviewed literature (Davies and Karstad 1981), though anecdotal reports and international disease reporting (OIE 2008; ProMED mail 1999) have demonstrated that several wild ruminant species can have clinical signs, including abortion. This review seeks to use a systematic approach to compile literature relevant to the occurrence, clinical course, and ecology of RVFV in wildlife species. It synthesizes this information and recommends future directions for continued research of the virus in an ecological system that includes both wildlife and livestock.

Rvfv Vector Ecology

Mosquito vectors that transmit diseases are often heavily dependent on seasonal rainfall or localized water pools. Field and laboratory research focused on mosquitoes that can spread RVFV have identified two types of vectors. The first is a subset of floodwater Aedes spp mosquitoes that are hypothesized to transmit the virus transovarially (Linthicum et al. 1984; Swanepoel and Coetzer 2004). The eggs are desiccation resistant and believed to be able to survive long periods on the top of the soil until flooding occurs (Linthicum et al. 1985). During periods of abnormally high rainfall it is believed that transovarially infected Aedes mosquitoes hatch and initiate infection primarily in wildlife or domestic ruminants (Linthicum et al. 1984). Once a ruminant host is infected several mosquito species (particularly Culex spp) are capable of transmitting the virus horizontally and play an important role in the amplification and spread of the virus (Meegan et al. 1980). Some RVF outbreak prediction systems were developed based on environmental factors that may affect the hatching of Aedes eggs (Davies et al. 1985; Linthicum et al. 1999). Normalized difference vegetation index, precipitation, and temperature detected via remote sensing are used to predict the environmental conditions that the Aedes require for hatching (Anyamba et al. 2002). At least one outbreak of Rift Valley fever (RVF) has been accurately predicted in East Africa (Anyamba et al. 2006). However, predictions in other regions of Africa are not as accurate (Anyamba et al. 2010). The difficulty in predicting outbreaks using climate and vegetation variables indicates that RVFV is reliant on factors beyond those that predict vector abundances. These external factors may involve a single host reservoir and/or be related to a low-level transmission cycle including wild and domestic ruminant hosts. The ability of Aedes spp to be the sole reservoir for RVFV through transovarial transmission has yet to be demonstrated in a laboratory setting (no isolations from >3138 mosquitoes hatched from eggs laid by infected females [Turell et al. 2008]).

Reservoir Hosts and Systems

No mammalian maintenance hosts have been described for RVFV, although evidence of previous infection (Evans et al. 2008), clinical signs of infection (Davies and Karstad 1981), and, rarely, virus has been detected in wildlife (Capobianco Dondona et al. 2016). Olive et al. (Olive et al. 2012) define the qualities of a “good” reservoir host as a taxon that has an extended viremic period of sufficient viral titer to infect vectors with little or no clinical disease, a population with enough turnover to maintain a stable proportion of susceptible individuals, and that has regular contact with the vector population. In addition to considering a single host species or taxonomic group as the reservoir, it may be important to consider a multispecies framework (Haydon et al. 2002) that incorporates possible (non)maintenance host(s). A multispecies framework involves multiple populations of maintenance and nonmaintenance hosts, which are required for the persistence of the virus; if only one member of the community were to be present the virus would not be able to persist in the ecosystem (Haydon et al. 2002). Using this understanding of reservoir host(s) and systems and the current understanding of the vector ecology, the role of wildlife in the epidemiology of RVF is presented using a systematic approach to assess avenues for future research in wildlife.

Methods

A systematic literature search was conducted as follows: Web of Science and PubMed were queried using the search terms “Rift Valley Fever” and “Wildlife.” Duplicates were excluded, and titles and abstracts were then reviewed. Only publications that contained a reference to RVF and wildlife (or a specific species or taxa that are not domestic animals or mosquitoes) were considered topical, and only publications with an abstract that indicated it investigated the role or status of wildlife with respect to RVFV were included for full review. Finally, publications that did not provide new data (diagnostic, experimental, clinical cases, etc.) regarding the epidemiology of RVF in wildlife species were excluded (e.g., papers that anecdotally made a mention of wildlife being affected or playing a role in the ecology of the virus were excluded).

Given the difficulty in using more specific search terms (e.g., specifying species or genera for all wildlife species that may have been investigated), the references of each publication included for full review were examined and used to identify further references. This allowed the identification of additional publications surrounding a specific species or taxon, when the generic term wildlife was not used. For all papers, serological and testing results were summarized by species, country, and test type irrespective of location within a country or time period.

Results and Discussion

The systematic literature review yielded 84 publications from Web of Science and 136 records from PubMed. Of the 220 records, 38 were duplicates and 90 were excluded as nontopical, and a further 21 records were excluded from full review as nonrelevant. A total 61 records were included in the full review. Following full review, 34 additional publications were excluded for not providing new data regarding epidemiology of RVF in wildlife, were non-English, or were not available to the authors. However, all 61 records were utilized to identify additional references. An additional 85 articles were identified based on references and 30 were included in the full review. In total 57 records were included; there were two additional unpublished sources that were included as they had pertinent results to this review.

Rvfv in Free-Ranging Wildlife

The methods used to investigate RVFV in wildlife have varied greatly between publications, making direct comparison and/or summary statistics difficult to assess. Considering this, we will briefly review various diagnostic assays that can be used and will be discussed below. The World Organisation for Animal Health (OIE) determines international trade standards for terrestrial animals as well as reporting requirements for internationally important diseases (including RVF). The OIE permits the following assays for RVF (OIE 2009):

• Virus culture

• Agar gel immunodiffusion

• Real-time polymerase chain reaction (RT-PCR) assay

• Histopathology with immunostaining

• Virus neutralization assays including: microneutralization, plaque, reduction neutralization and neutralization in mice. Virus neutralization tests (VNTs) are the prescribed test for international trade.

The following tests are described in the OIE Terrestrial Manual, which may also be conducted, but are not considered prescribed tests for international trade (OIE 2009):

• ELISAs: An indirect IgG ELISA based on a whole or recombinant nucleocapsid protein as antigen and an IgM-capture ELISA for diagnosis of recent infection (OIE 2016).

There are several other ELISAs that have been published, including: inhibition ELISA based on a whole antigen detecting either or both IgG and IgM (total antibodies) (Ellis et al. 2014; Paweska et al. 2005) and IgG sandwich ELISA (Paweska et al. 2003a), among many others. Generally hemagglutination inhibition, complement fixation, agar immunodiffusion, and radioimmunoassays are no longer used (OIE 2016). The ELISAs mentioned above depend on detecting host antibodies that bind to recombinant or whole viral antigen. Species-specific differences in antibody creation and other hematological analytes can impact the accuracy of the serological test (OIE 2013). This type of species-dependent interference could cause a significant change in either false-negative or false-positive results by a given test (OIE 2013). Without validation of ELISAs by species, the accuracy of the results remains difficult to interpret. Only limited validation data have been published for the use of ELISA-based assays in wildlife (Paweska et al. 2005, 2003b, 2008, 2010). Neutralization tests are considered the gold standard (OIE 2009). These assays can be very costly and require BSL-3 facilities to work with the live virus, as well as a source of live virus for the assay. The ELISA and hemagglutination tests have been also commonly used in the studies discussed below; however, the lack of validation for wildlife species make their results difficult to interpret.

This review will focus on the results of the neutralization assay-based serosurveys (Table 1), with the other serological assays being included in the Supplementary Information (SI) Table for thoroughness. Performing neutralization assays requires experienced laboratories, since the test methodology is complex and test results dependent on well-standardized in vitro and/or in vivo systems and the reference virus used for the assay. With this understanding, and not accounting for geographic location, time relative to the most recent outbreak, or age of the host (all of which will affect the apparent prevalence), it can still be useful to compare serological results across studies to direct future work toward species that may be affected by RVF and/or may play a role in maintenance of RVFV. Table 1 contains results from species that have had a sample size (cross-study) of at least 60 individuals and used a neutralization assay or rodent inoculation assay for the presence of RVFV antibodies. Cross-study sample sizes range from 60 to 2394 and apparent prevalences range from 0 to 0.65 (Table 1). Of the 29 species that had a cumulative sample size >60, seven (27%) had no antibodies detected and five (17%) had an apparent seroprevalence >10% (Table 1). Additionally, nine other species had a seroprevalence >10% and sample sizes ≧10 individuals (see SI Table). Interpretation of serological data can be nuanced, as it simply identifies exposure and no inferences can be made as to the likelihood of a species being clinically affected by RVFV. Seropositive results may indicate the species acts as a reservoir host; or has high contact rates with vectors. However, regardless of the reason for exposure to RVFV, these data indicate that the species mentioned above and/or habitats these species prefer should be further investigated to understand RVFV ecology and epidemiology.

As with serology, direct viral detection in free-ranging wildlife cannot be used to directly infer the role a species plays in the ecology and epidemiology of the virus. While a significant number of free-ranging wildlife have been tested directly for RVFV (n = 3202), only 31 positive animals of 14 species (8 animals were positive both on RT-PCR and virus isolation [ProMED mail 1999]) have been detected and reported in peer-reviewed literature (Table 2). This includes: springbok (Antidorcas marsupialis), buffalo (Syncerus caffer), waterbuck (Kobus ellipsiprymnus), and giraffe (Giraffa camelopardalis) reported to have aborted or died due to RVFV infection, which was confirmed via virus isolation and RT-PCR (ProMED mail 1999), one field RVFV isolation in buffalo (Bird et al. 2008), and three isolations from bats (Hipposideros abae, Micropteropus pusillus, Hipposideros caffer) (Boiro et al. 1987), ([Oelofsen and van der Ryst 1999] as cited in [Olive et al. 2012]; see Table 2). Of the species identified above as warranting further investigation based on RVFV neutralization results, there have been reports of attempted viral detection in 8 Namaqua rock rats (Swanepoel et al. 1978), 32 buffalo (Bird et al. 2008; ProMED mail 1999), 8 impala (Capobianco Dondona et al. 2016), 8 giraffe (Bird et al. 2008; ProMED mail 1999), and 6 waterbuck (R. Bengis, unpublished data) (ProMED mail 1999). No studies that attempted to detect virus in kudu (lesser or greater), nyala, gemsbok, black wildebeest, black rhinoceros, orange weaver, or Hubert’s Mastomys were identified during this review. The results of the buffalo, waterbuck, and giraffe were given above. No positives were detected in the Namaqua rock rats (n = 8) or gerenuk (n = 5). All other attempts to detect RVFV from free-ranging wildlife has proven unfruitful (as reported by the reviewed publications) (Davies 1975; Henderson et al. 1972; Hoogstraal et al. 1979). Deaths have been reported in springbok, blesbok (Damaliscus pygargus phillipsi), buffalo, sable (Hippotragus niger), nyala (Tragelaphus angasii), eland (Tragelaphus oryx), waterbuck, bontebok (Damaliscus pygargus pygarus), fallow deer (Dama dama), and llamas (Lama glama) during previous RVF outbreaks in South Africa (R. Bengis, unpublished data). Wild springbok and blesbok (Damaliscus pygargus phillipsi) were also reported to have aborted during an RVF outbreak though no diagnostics were reported (Bengis and Erasmus 1988).

Experimental Infections Conducted in Wildlife

The search for a wildlife reservoir species for RVFV has been ongoing since it was first discovered by Daubney et al. (Daubney et al. 1931) in 1930. For a species to be identified as a reservoir, the following host characteristics are generally assessed: minimal clinical signs and/or mortality associated with a high and long-lasting viremia, range overlap with enzootic regions, detection of the virus in free-ranging members of the species, and relatively high or constant seroprevalence levels across populations (Olive et al. 2012). Experimental infection of wildlife seems to have been restricted primarily to rodents, primates, and other laboratory animals (rabbits, guinea pigs, rats, mice, and ferrets), with an occasional study on buffalo, bats, and birds. Death following inoculation (subcutaneous, intraperitoneal, intradermal, and/or intracranial) with RVFV occurred in wood mice (10/10, Apodemus sylvaticus [Findlay 1932c]), voles (6/6, Cricetidae [Findlay 1932c]), dormice (3/3, Gliridae [Findlay 1932c]), golden hamster (2/2, Mesocricetus auratus [Findlay 1932c]), rats (35/57, Rattus sp [Findlay 1932c]), squirrels (1/2, Sciurus sp [Findlay 1932c]), African pouched rats (1/2, the other was euthanized, Saccostomus sp [McIntosh 1961]), Abyssinian grass rats (19/57, Arvicanthis abyssinicus nairobae [Daubney and Hudson 1932]) and 27/91, (Arvicanthis abyssinicus nubilans [Weinbren and Mason 1957]), and the southern multimammate mice (6/35, Mastomys coucha [Daubney and Hudson 1932]). Nearly all primate species had clinical signs postinoculation (Davies et al. 1972; Findlay 1932a); however, mortalities were reported only in macaques (9/49 died, 19/49 had clinical signs, Macaca mulatta [Findlay 1932b, 1932c; Morrill et al. 1990; Peters et al. 1988]). Of the buffalo, most (4/6) had subclinical disease similar to subclinical RVF in cattle, one (of two pregnant buffalo) aborted and one buffalo had clinical signs of RVFV (Daubney and Hudson 1932; Davies and Karstad 1981). None of the other 23 species/genera that were inoculated had clinical signs or signs were not reported post-RVFV inoculation (Davies and Addy 1979; Davies and Linthicum 1986; Findlay 1932b; Francis and Magill 1935; Gora et al. 2000; Hoogstraal et al. 1979; Oelofsen and van der Ryst 1999; Pretorius et al. 1997; Swanepoel et al. 1978). Of 43 species/genera that were inoculated, 16 species were tested for and produced neutralizing antibodies, 23 species had a detectable viremia, and three species seroconverted as measured on nonneutralizing assays (see Table 3 for details). The longest period of time an animal remained viremic was 13 days (or the longest period in which viremia was assessed) in rhesus macaques (Findlay 1932c). It was also noted that the macaques that had clinical signs and/or more severe illness remained viremic longer (Findlay 1932c).