A participatory approach to marsh deer (Blastocerus dichotomus) morbidity and mortality passive surveillance in Argentina: rst results.

Background: The comprehensive approach to diseases in broad ecological contexts appears as a new challenge for global health. Winter mortality of marsh deer ( Blastocerus dichotomus ) was historically described in Argentina even though eld diculties prevented their in-depth study. Results: Between May 2014 and April 2017, we investigated marsh deer morbidity and mortality through passive surveillance in the two largest populations in Argentina using a participatory approach. The latter spanned a network of researchers, eld partners (veterinarians, park rangers, and local community), and decision makers. We sampled dead deer during as well as outside mortality events. A total of 44 marsh deer with different body condition scores were evaluated. We obtained reference haematology and biochemistry values from animals with good body condition score. Marsh deer with poor body condition had a high burden of the ticks Amblyomma triste and Rhipicephalus microplus . Vector-borne agents (VBAs), such as Theileria cervi , Trypanosoma theileri, Trypanosoma evansi, Ehrlichia chaffeensis, Anaplasma platys, Anaplasma odocoilei, Anaplasma marginale , and Candidatus Anaplasma boolense were also found. Haemonchus spp., Ostertagia spp., and Trichostrongylus spp. were the most frequent gastrointestinal parasites in deer with poor body condition. A Multiple Correspondence Analysis reinforced a possible association of winter period with lower body score condition, high tick loads, infection with E. chaffeensis , and presence of harmful gastrointestinal parasites. Conclusions: Our approach allowed the establishment of a participatory surveillance network of marsh deer morbidity and mortality in Argentina. We report and analyse the rst data obtained opportunistically within the framework of this network, providing information on the infectious and parasitic agents in marsh deer populations. The occurrence of Fasciola hepatica and Leptospira interrogans serovar pyrogenes is reported for the rst time in wild marsh deer from Argentina. The histopathological ndings in succumbed animals allowed us to distinguish lesions associated with their body condition from those that were unrelated. Our data will be useful to improve the interpretation of future surveillance


Results
First steps towards building a participatory surveillance network of marsh deer morbidity and mortality The participatory surveillance network was integrated by researchers, eld partners (veterinarians, park rangers, livestock and timber producers, and local community), and decision makers in the two largest populations of marsh deer in Argentina located in Ibera Wetlands (IW) and Lower Delta (LD). In IW, the network included personnel from national and provincial protected areas, NGOs, and private veterinarians. In LD, the activities were integrated into the work programme and management structure of the "Marsh Deer Scienti c Technical Committee" (CCP) and included samplings during live captures carried out by "Pantano Project".
The work was gradually articulated in both study areas. Training classes and workshops led to an increased knowledge and improved sampling and biosecurity skills of local partners. During 2014 and 2015, local partners received practical training in the eld. In 2016, together with the National Parks Administration, we organized a workshop for the local community in IW. Participants exchanged information on marsh deer and discussed issues, concerns, and approaches in wildlife health, with the purpose of expanding and strengthening the network. The theoretical-practical workshop was attended by 60 persons, including representatives from multiple public, private, and academic institutions and the local community. A similar workshop, held during 2016 in LD, was attended by 15 local partners (Fig. 1A).

First reports of mortality events
With the support of Provincial Wildlife Agencies and the National Parks Administration, the rst reports of mortality events of marsh deer within the framework of the network took place in May 2014 and June 2015 (Fig. 1A). A total of 27 and 100 carcasses of marsh deer were reported in the Mburucuyá National Park (western IW) and in Cambyretá Area -Iberá National Park (north-eastern IW), respectively. Although most of the dead animals were registered by park rangers during eld monitoring in remote areas, a few marsh deer with apparent signs of weakness were observed in the days prior to their death.
Necropsies of three marsh deer were performed by our group at each site; the remaining individuals could not be evaluated due to advanced autolysis. During the winter of 2016, trained eld partners registered only seven dead animals in IW. A complete necropsy was performed by local partners (veterinarians and park rangers) on two of them. In LD, an extraordinary ood between April and August 2016 coincided with wildlife mortality due to disease and starvation, but some individuals died because they were hit by vehicles, attacked by dogs, or hunted due to increased visibility of animals grouped in the few remaining high and dry areas [28]. Several cases of dead, injured, and hunted marsh deer (n=233) were reported to the CCP [28]. Complete necropsies were performed on nine animals, two of which were found dead with visible signs of disease, four had been sighted sick in the days prior to their death, while the remaining three had been killed by poachers.
In both areas, local partners also reported dead deer that had been hunted, hit by vehicles o killed by dogs outside mortality events; they were sampled opportunistically.

Marsh deer survey
A total of 44 marsh deer, mostly adult males, were analysed between May 2014 and April 2017. Thirty-ve dead animals were sampled in both areas, 19 of which were clinically sick individuals sampled during mortality events. Upon visual examination, clinical signs frequently included notable weakness, pale mucous membranes, loss of one or both antlers (Fig. 2), drooping of head and ears, rough dull coat, decreased muscle tone, severe emaciation, dehydration, loss of coordination, and inability to stand before death. Live marsh deer (n=9) and the other 16 dead animals, including 10 hunted deer, one road-kill, three dog-kills and two visibly sick individuals, were sampled outside of mortality events. Some deer sampled during and outside mortality events (n=17) had been seen alive before their death (Table 1, Fig. 1A).
Half of the studied individuals (n=22) had a good body condition (score 3). The other half showed a regular (score 2) or a poor body condition (score 1), and most of them were sampled during mortality events (Table 1). All marsh deer with body score 1 showed cachexia and skin laceration. Submandibular oedema and diarrhoea were frequent in individuals with regular or poor body condition score. During winter, ve females were pregnant at time of death, in correspondence with the reproductive season. Five adult or juvenile males had lost their antlers ( Table 1). The temporal distribution of sampling, body scores, and cause of death (if applicable) of marsh deer are detailed in Fig. 1B.

Haematology and biochemistry values
Haematological and serum biochemistry parameters of marsh deer were determined in animals sampled alive and with good body condition. In addition to the nine live animals, we included three individuals that had been rescued with gunshot wounds and were sampled while alive, and then died (Table 2). Almost all of the mean haematology values were within the range previously reported for free-living marsh deer in Brazil [29]. Signi cant differences with the values reported by Szabó et al. [29] were observed in the packed cell volume for both sexes; red blood cell count for both sexes; mean cell haemoglobin concentration in females; and total protein in both sexes [29] (Supplementary material, Additional les 1, 2).

Serological analyses
The results of the serological analyses are shown in Table 3. Two marsh deer from LD showed evidence of exposure to Leptospira interrogans serovar pyrogenes. One of them, sampled alive during a CCP rescue, had a good body condition and a L. pyrogenes pyrogenes titre of 1/200; the other one, sampled during a mortality event, had a poor body condition and a titre of 1/100. One hunted marsh deer from IW with regular body condition showed evidence of exposure to brucellosis in all four serological techniques (Table 3).

Identi cation of ticks and diagnosis of tick-borne agents
All marsh deer from IW were parasitised with the ticks Rhipicephalus microplus and/or Amblyomma triste, and all ticks collected from individuals in LD were identi ed as A. triste. Tick loads were estimated in 37 marsh deer, including live and dead animals sampled up to eight hours post-mortem. High tick loads were found in nine individuals (24.3%), eight of which were from IW and showed regular or poor body condition. In LD, most marsh deer (74%) had a low tick load ( Table 1).
The molecular detection of vector-borne agents (VBA) was performed in 40 marsh deer (Table 4). Regarding piroplasmid parasites, Babesia sp. was not detected in any deer, whereas T. cervi was found in 21 individuals, 11 of them with regular or poor body condition score. Ehrlichia chaffeensis was found in ve marsh deer, four of which had a poor body condition score. In both areas, different Anaplasma species occurred in marsh deer with good, regular, and poor body condition scores. The identi ed species of the family Anaplasmataceae included E. chaffeensis, A. platys, A. odocoilei, A. marginale, and Candidatus A.
boolense. Trypanosoma theileri and T. evansi occurred in marsh deer with good, regular, and poor body condition scores in both areas. Rickettsia sp. was not found in any of the examined individuals.
A total of 20 individuals in both areas had co-infections with different VBAs during and outside mortality events. Co-infections involved two VBAs in 13, and three VBAs in six marsh deer. Only one animal with poor body condition sampled during a mortality event in LD was co-infected with four VBAs.    Thirty-one out of 43 animals (72.1%) had less than 10 oocysts per gram (OPG) of faeces. Eleven deer had OPG values between 11 and 100. The highest value (OPG=790) was found in a dead marsh deer during a mortality event. This animal also had a poor body condition, high tick load, and EPG=156, including Trichostrongylus spp., Haemonchus spp., Ostertagia, and Paramphistomum spp. (Fig. 3B).
Results of the qualitative faecal analyses are shown in Table 5. Trichostrongylina [30], Strongyloides spp., Capillaria spp., and Paramphistomum spp. eggs were found. A high frequency of infection (79%) was detected for Trichostrongylina eggs. Larval culture of 18 faecal samples evidenced the occurrence of Trichostrongylus spp. and Strongyloides spp., and third-stage larvae morphologically compatible with the genera Haemonchus, Ostertagia, Oesophagostomum, and Cooperia. Nine marsh deer were positive for Haemonchus sp.; seven of them had a regular or poor body condition and died during mortality events, whereas the three marsh deer positive for Ostertagia spp., also sampled during mortality events, had a poor body condition.
Adult parasites of Paramphistomum cervi were identi ed macroscopically in the rumen of marsh deer during necropsies, and their eggs detected in faeces ( Table 5). Six of the seven animals positive for this parasite were from IW; ve of them had died during mortality events and showed a poor or regular body condition.

Gross and histopathological ndings
Complete necropsies were performed on 23 marsh deer (14 males and 9 females); 12 of them were adults, 10 juveniles, and one was a fawn. Eight (35%) deer had a good, ve (21.5%) a regular, and 10 (43.5%) a poor body condition score. Table 6 summarizes the main microscopic ndings in dead animals with score 1 and 2 sampled during mortality events vs. animals with score 3 that died of traumatic causes. One marsh deer with score 2 and sampled outside a mortality event was excluded from Table 6. This animal was reported dead in a eld in LD. The deer had in ammation and myiasis in the metacarpophalangeal joint of the left forelimb and a fracture in the second phalanx in the same forelimb. Eight of the necropsied marsh deer died of traumatic causes. Three animals had received gunshot wounds to the thorax and and were diagnosed with oedema, congestion, and pulmonary haemorrhages ( Table 6). Three individuals had been wounded in other parts of the body, and two animals had been attacked by dogs, showing haemorrhages and non-speci c lesions in different tissues.
In addition to the necropsied individuals that died of traumatic causes, histopathological analysis allowed us to determine the cause of death of seven animals (two from IW and ve from LD) that had poor body condition and died during mortality events. The severe leucocytosis in most organs and multiple histopathological lesions (multifocal hepatic necrosis, lymphoid necrosis of the spleen, and interstitial pneumonia) ( Table 6) detected in one marsh deer from IW were compatible with septicaemia. This animal also showed cachexia, submandibular oedema, skin laceration, high tick load, OPG=790, and was coinfected with T. cervi., E. chaffeensis, Trichostrongylus spp., Haemonchus spp, Ostertagia spp., and Paramphistomum sp. The other marsh deer from IW had a severe acute cortical nephrosis; it also showed high tick load and was co-infected with T. cervi. and T. theileri.
One deer from LD had an acute multifocal myocardial necrosis (Table 6). In addition, it showed oedema in multiple organs, a mild nephritis, abomasitis, and EPG=1880. Cachexia, submandibular oedema, skin laceration, and high tick load were also found, and the deer was co-infected with T. cervi, T. theileri, Trichostrongylus spp., and Ostertagia spp.
Fibrinous bronchopneumonia could have been the cause of death of one deer sampled in 2016 during the mortality event in LD. This animal was coinfected with T. cervi, T. theileri, and Trichostrongylus spp. Another marsh deer sampled in LD died of nephrotoxic nephrosis (Table 6). Clinical conditions and tissue lesions related to malnutrition were detected in this animal and in two other marsh deer sampled in the same area during the mortality event. All three animals were males, showed cachexia, and two of them had lost their antlers days before their death (the third one was a fawn). One deer was co-infected with T. cervi, A. odocoilei, A. marginale, and T. theileri, and had a hepatocellular atrophy. The fawn had a multifocal pustular rumenitis, atrophy of the thymus, absence of fat reserves, and hepatic lipidosis, and was infected with E. chaffeensis.
In the other eight animals, the lesions were non-speci c and did not allow us to establish the cause of death. One deer died with a regular body condition score during the period outside the mortality events, while the other eight individuals showed poor or regular body condition and died during mortality events. Six marsh deer showed cachexia, four of which also had submandibular oedema. Five had high, and four intermediate tick loads.
Cysts of Sarcocystis sp. were detected in the cardiac muscle of six animals; in two of them, cysts were also found in skeletal muscle. Embryonated eggs and larvae of Metastrongylus nematodes were found in seven marsh deer (three juveniles and four adults), whereas adult nematodes were observed only in one adult deer. Adult forms of F. hepatica were found in the liver of ve animals from IW (22%). Unidenti ed scattered pyriform microorganisms were observed in erythrocytes in brain vessels of one of these animals, a male with poor body condition.
Skin lesions from two marsh deer were examined. A juvenile male presented a chronic, locally extensive, ulcerative dermatitis with associated panniculitis on the neck. The other lesion was a malignant melanoma in the upper right eyelid of an adult male, characterized by fusiform, anaplastic melanocytes, many of them with intracytoplasmic melanin granules.

Data analysis
The MCA showed that lower body condition (score=1), higher tick loads (tick load =1), infection with E. chaffeensis, and presence of harmful gastrointestinal parasites (Ostertagia sp., Paramphistomum sp., and Haemonchus sp.) were correlated (Fig. 4). Low body scores and high parasite loads were correlated with the sampling period from May to August (winter season). In the opposite way, the MCA showed that a higher body score was correlated with better weather conditions (sampling period September to December) and the absence of harmful parasites. These associations were not statistically con rmed.

Discussion
In order to establish and strengthen a participatory approach for the passive surveillance of marsh deer morbidity and mortality in Argentina, we focused on the development of local networks, improving report and sampling of visibly sick or dead marsh deer, and taking advantage of access to samples of healthy animals during captures and / or rescue procedures. Accurate data about health parameters in wildlife species are relevant to detect changes in the occurrence of pathogens over time.
Thanks to the local networks we established, we were able to compile information from three mortality events that occurred in 36 months. Although case reports increased, access to biological samples in remote areas remains a bottleneck for the comprehensive study of mortality events. A large number of dead animals were found, but in most cases their late detection made it impossible to collect samples and/or determine their cause of death.
In this study we describe the health conditions of 44 marsh deer from the two largest populations in Argentina. We provide blood parameters of adults with a good body condition, report the most frequent infectious and parasitic agents in the assessed populations, and describe the lesions found in tissues of dead animals. Necropsies evidenced that mortality events did not have a single cause. On the contrary, they were due to a combination of factors, such as unfavourable environmental conditions, increased parasite load, and loss of body score. The MCA reinforced a possible association between adverse weather conditions during winter periods with lower body score, higher tick loads, infection with E. chaffeensis, and presence of well-known harmful gastrointestinal parasites. Further studies including a larger sample size are needed to understand the cascade of events that trigger mortality events.
The haematological data obtained in this work are the rst values reported for free-ranging marsh deer in Argentina; in general, they are in accordance with those from Brazilian populations [29]. The serum chemistry values are, to our knowledge, the rst reported for the species.
Exposure to infectious agents in the analysed marsh deer was low. None of the studied animals showed evidence of exposure to BTV, IBRV, BVDV, foot-andmouth disease, Johne's disease, bovine leukosis, Q fever, chlamydial abortion, or VSV. Published information about these infectious agents in marsh deer is scarce. Some reports describe a high prevalence of antibodies against herpesvirus-1, although they do not distinguish this virus from cervid herpesvirus-2 due to their antigenic similarity [31]. A high seroprevalence (74%) and typical lesions of epizootic haemorrhagic disease virus (EHDV) (an orbivirus related to bluetongue) were detected in Brazil [32], whereas bluetongue virus was reported only in captive Mazama sp. In Argentina, serum samples from 14 free-ranging Ozotocerus bezoarticus celer were negative to Johnes' disease, BVDV, EHDV, and BTV [33].
Surprisingly, only few individuals tested positive for Leptospira interrogans, although sampling was done during and after a ood event. In Argentina, human exposure during ood events was the most important single risk for leptospirosis [34]. In the central region where LD is located, a high number of cases of leptospirosis in humans was recorded between 2015 and 2016 [35], coinciding with our sampling period. Only two marsh deer from this area were seropositive for the pathogenic L. interrogans serovar pyrogenes; one of them was found dead without lesions and the other was found alive, with no alterations in haematology and biochemistry values. A single positive titre of 1:100, as found in one of the individuals, can be interpreted as a residual background titre, whereas those between 100 and 200, as found in the second one, are considered relevant in wild herbivores [36]. Although the information on leptospirosis in marsh deer is extremely scarce, studying this pathogen is especially relevant in severe ood scenarios, as its incidence is strongly associated with rainfall and high temperature and humidity levels [37]. Antibodies against L. interrogans were found in related species, such as Mazama gouazoubira [38] and O. bezoarticus (serovars hardjo, mini, wol , and pomona) [33,39]. This study documents for the rst time antibodies against L. interrogans serovar pyrogenes in the southernmost marsh deer population in Argentina.
One hunted marsh deer from IW showed evidence of exposure to brucellosis, with relatively high titres. No previous serological analyses in marsh deer from Argentina or Brazil reported evidence of Brucella infection [36,40]. In our study area, bovine brucellosis has been detected in cattle herds [41], and marsh deer could become infected in environments shared with livestock.
Although we were unable to estimate the tick load on all marsh deer at the moment of death, the occurrence of high burdens of A. triste and R. microplus was an important nding in animals with poor body condition, in which skin lesions contributed to an impoverished general condition. This was especially evident during mortality events in IW, where R. microplus was the most frequent tick. Rhipicephalus microplus mainly infests cattle and is endemic to north-western Argentina, although its geographical distribution does not include LD. The VBSs found in the analysed marsh deer were the same as those observed in Brazil [10][11][12][13].
Ehrlichia chaffeensis, which causes a zoonotic disease, has recently been described in marsh deer in Argentina [22]. In the present study, its occurrence in animals from IW was positively associated with poor body condition score. Positive deer, except for a fawn, had medium or high tick loads. We suspect that the in ammatory lesions found in the kidney of a positive deer (i.e., glomerulonephritis secondary to immune complex deposition) were caused by E. chaffeensis; future immunohistochemical studies might con rm the possible association.
Infection with T. cervi was associated with poor or regular body condition scores. Theileria cervi has been historically considered of low pathogenicity, probably because of a long evolutionary relationship between parasite and host [42]. Theileria cervi infections are therefore often asymptomatic in naturally infected cervids, except for animals with high parasite load, concurrent disease, malnourishment, immunosuppression, in areas with high deer population densities, or in stressful situations [42][43][44]. Histologically, records of hemosiderosis in tissues of positive marsh deer suggest a possible relationship between the lesions and the agent [45,46]. In this study, of a total of 11 T. cervi-positive deer with poor or regular body condition, 91% were assessed during mortality events, and all of them had co-infections with other infectious or parasitic agents. In addition, scattered pyriform microorganisms were found in red blood cells in the brain of one positive deer. Detailed studies of tissues of T. cervi-positive deer will be essential in order to correlate the infection with speci c lesions, especially in populations under stress conditions.
Trypanosoma theileri, T. evansi, A. platys, A. odocoilei, A. marginale, and Candidatus A. boolense were found in marsh deer from both areas, regardless of their body condition. Candidatus A. boolense was rst identi ed in mosquitoes in China [47], and this is the rst report in marsh deer in Argentina.
Most of the assessed animals had low EPG and OPG values, and Trichostrongylina eggs, including Haemonchus spp., Ostertagia spp., and Trichostrongylus spp., showed the highest infection frequency. In Argentina and Brazil, helminthic diseases are an important cause of morbidity in this species [16][17][18], and H. contortus was found to be one of the most pathogenic agents involved in mortality events [19,20,48]. Ostertagia sp. was also described in marsh deer in Brazil [17,18,49]. This parasite causes abomasal epithelial hyperplasia and an imbalance in the protein digestion process. During hypobiosis, it can cause petechiae and ecchymotic haemorrhages in the abomasal mucosa [50]. The submandibular oedema found in marsh deer with poor score that died during mortality events could be associated with multiple coinfections with VBAs and harmful gastrointestinal parasites, such as Haemonchus spp and Ostertagia spp., as suggested in the exploratory investigation of a winter mortality event in 2007 [20]. The three deer positive for Ostertagia spp. were sampled during mortality events, showed co-infection with Trichostrongylus spp., and two of them also with Haemonchus spp. This record is signi cant due to the high pathogenicity of these agents in domestic livestock and wildlife [51].
Paramphistomum cervi and Fasciola hepatica were detected in more than 70% of the necropsied marsh deer from IW. This is the rst record of F. hepatica in this species in Argentina. The occurrence of P. cervi was previously described in animals in Corrientes province [21], and high loads of trematodes, including P. cervi, were found in individuals from the Parana River basin, Brazil [48]. Both agents had a high prevalence in IW, favoured by suitable ambient temperatures and the simultaneous presence of Lymnaea sp. and known de nitive hosts (e.g., domestic sheep). In IW, the infection rate in Lymnaea sp. increases until the end of summer and autumn, and the clinical signs of disease appear 2-4 months later [52]. Most of the analysed marsh deer had died during mortality events associated with extraordinary oods in IW. The oods increase the area of suitable habitat for Lymnaea sp., favouring the production of metacercariae and increasing the risk of infectation of the de nitive host of this trematode. Marsh deer have a higher infection risk due to their habit of feeding in swampy environments. Given the ecological characteristics of B. dichotomus, the detection of liver lesions in some infected individuals, and the high prevalence of F. hepatica in the area, additional studies are needed on its role in the maintenance of the infection in the area.
The histopathological ndings in succumbed animals allowed us to recognize different lesions, some of which were associated with their body condition.
Marsh deer wounded during illegal hunting often ee to swampy areas, where detection is di cult for poachers. Days later, they may appear prowling near human dwellings, where they end up dying. In road-killed or hunted marsh deer, most of the lesions were agonal, such as congestion, oedema, and pulmonary haemorrhages. Other ndings were incidental, such as in ammatory reactions in liver, kidneys, lungs, and abomasum. In agreement with the ndings described by Navas-Suárez and collaborators [53], they were frequently associated with parasitic agents, such as F. hepatica, Sarcocystis sp, Metastrongylus, or Trichostrongyloidea nematodes.
Lesions of the respiratory system were frequent in marsh deer with and without signs of disease. Pneumonia has been described as one of the most frequent in ammatory processes in cervids [53]. In this study, pneumonia were usually mild, except in a juvenile with a locally extended brinous bronchopneumonia.
Similar to the observations of Navas-Suárez and collaborators [53], lung congestion was the most common hemodynamic disorder, followed by oedema and haemorrhage.
Hepatic in ammations were frequent, but usually not signi cant or incidental, in the examined deer. Digestive tract lesions had a high frequency in marsh deer with poor body condition score. Abomasitis was the most common lesion in the abomasum and was generally associated with the presence of Haemonchus sp. and Ostertagia sp, with some animals showing depressions in the abomasal mucosa. A high percentage of intestinal samples showed some degree of autolysis, which hindered bacterial culture and isolation, or a proper histopathological study. Some animals had clinical signs of diarrhoea. Campylobacter spp. was isolated from faeces of marsh deer in Brazil [54]; in most species, the main clinical sign caused by this bacterium is diarrhoea, which is usually selflimited. Other enteric infections by Salmonella spp. and Yersinia spp. may affect cervids [36], but they have never been reported in marsh deer.
Only 15 necropsied animals showed lesions that allowed us to determine their cause of death. Eight animals with good body condition died of trauma.
Hepatic lipidosis was found in three animals with regular or poor body condition, which could be associated with a negative energy balance during mortality events in the two areas. The nephrosis found in two marsh deer suggested that they died of renal failure, although the cause of the kidney disease could not be established.
In LD, the oedema found in multiple organs of a dead deer with myocardial necrosis may have been associated with heart failure. One individual that died in LD had brinous bronchopneumonia, but we were not able to isolate the causal agent due to advanced autolysis at the time of sampling and logistical limitations. The clinical conditions may have been caused by bacteria such as Mannheimia hemolytica, Histophilus sommnys or Pasteurella multocida, which are frequently detected in stress situations or during co-infections with viral agents [53].
Three dead individuals were sampled in LD in an extraordinary ood during the mortality event of 2016, when food was very scarce. Clinical conditions and tissue lesions of the three dead individuals were related to malnutrition, and one of them had renal failure (see above). The absence of fat reserves and hepatic lipidosis indicated a negative energy balance. The male reproductive cycle and the antler-weakening process are mostly unknown in marsh deer [32].
According to our observations, in both study areas antlers may be shed in almost every month, and older males retain hard antlers for long periods. We were unable to establish whether the loss of antlers in the two adults (the third animal was a fawn) was associated with the antler-weakening process or in uenced by malnutrition during mortality events.
With the development and strengthening of the participatory network, we were able to explore marsh deer populations in remote areas, training local collaborators and making them part of the investigation of marsh deer mortalities, a situation that they observe year after year. Through sustained and collaborative work in both areas, we hope to further shorten detection and sampling times, thus improving the number and quality of the biological samples. In this study we focused on the agents most commonly mentioned as causing disease in marsh deer and related species, and on those infectious and parasitic agents prevalent in the two study areas. Many of the reported results lay the foundation for future research on the association of infectious agents with pathological lesions in marsh deer, spillover of agents to and from populations of domestic ruminants, and the role of these cervids in the transmission cycles of certain pathogens. Our work adds important data to the scarce available information on marsh deer health in Argentina [36] and may serve as the basis for future investigations of mortality events.

Methods
Study area. Fieldwork was conducted in Ibera Wetlands (IW) and Lower Delta (LD) of Parana River in eastern Argentina. IW is included in a macro-wetland system of 1,300,000 hectares (27º40'S, 56º38'W). The region comprises complex ecosystems dominated by paludal environments in Corrientes province. While part of the area is protected (Iberá Provincial Reserve and Iberá National Park), a large portion still consists of private lands. Main productive activities include extensive cattle ranching, agriculture, and tourism [55]. The climate is subtropical with average temperatures of around 15-16 °C, and an absolute minimum temperature of -2 °C in winter. Mean annual rainfall is around 1,500-1,800 mm [55]. LD is part of the "Paraná Delta" Biosphere Reserve, located in the Parana River oodplain in Buenos Aires and Entre Ríos provinces (34°15'S, 58°58'W). The area (88,724 ha) comprises the typical deltaic morphology with a permanent additional growth of alluvial lands on the outer front of the Parana River. For more than 150 years, LD was subjected to intensive forestry associated with the construction of dams and roads. At present, subsistence and sport hunting are frequent, although illegal. The climate is temperate with average temperatures of around 16-18 °C and annual rainfall of 1,073 mm [56].
Establishment of a participatory network. A surveillance network was developed in each area. Key partners were identi ed to assist the development of a functional system. Local park rangers and veterinarians were invited to join the participatory approach in order to act as focal points to raise awareness and for biological sample collection, respectively. Livestock and timber producers and other members of the local community were trained in marsh deer morbidity and mortality recognition and immediate reporting. Results of the study were shown to decision makers from Provincial Wildlife Agencies and the National Parks Administration, in order to contribute to management strategies.
Methods for training activities ranged from hands-on training for eld staff to workshops and informal community gatherings to convey study advances.
Workshop presentations were made by epidemiologists, pathologists, and eld veterinarians. Appropriate teaching materials and methodologies were developed and used depending on the target audience. The training consisted in data recording, necropsy and sampling techniques, and biosafety and biosecurity standards. The participants were also provided with disposable materials and biosafety kits for sample collection.
At the laboratory, a group of specialists were involved in the diagnostic procedures (histopathological, serological, and molecular diagnosis of infectious and parasitic agents), contributing to the national surveillance system for noti able diseases. Some samples were sent to the laboratory of the National Service of Agri-Food Health and Quality (SENASA). The information obtained was reported online in the World Animal Health Information System (WAHIS) of the World Organisation for Animal Health (OIE), which summarizes the animal health situation worldwide.
Marsh deer sampling. Field work consisted of the investigation of animal carcasses or tissues from dead animals during and between mortality events, and was complemented by sampling live animals in one population. Necropsies and sample collection of dead marsh deer were performed by trained local "key partners" or personnel from our group. Free-ranging marsh deer were immobilized by parenteral anaesthesia using tiletamine hydrochloride and zolazepam chlorhydrate (Zelazol®, Fort Dodge) 1.2-1.5 mg/kg combined with xylacine (Xilacina 100®, Richmond Vet Pharma) 0.5-0.8 mg/kg and butorphanol tartrate (Butormin®, Holliday -Scott S.A.) 0.3-0.4 mg/kg [32,57]. The animals were maintained on thermal padded surfaces and oxygenated through mask (5-8 l/min). Vital signs were continuously measured using a multi-parameter medical monitor (Veterinary Monitor, GT9003C, Fridimex S.A.). The anaesthesia was reversed using naloxone hydrochloride (Naloxona®, Denver Farma) 0.05 mg/kg and yohimbine (Yohimbine Vet ®, Richmond Vet Pharma) 0.2 mg/kg.
Live and dead deer were assigned to an age class (fawn, yearling, or mature adult) according the classical tooth-wear aging technique [58], and body condition (scores 1, 2, 3) was determined using the Body Condition Score Chart [59]. Data about reproductive status, characteristics of the antlers, evidence of submandibular oedema, cachexia, and bone fractures were recorded. All animals were sexed and checked for skin lacerations and presence of ticks by visual inspection and palpation. Ticks were collected manually using acarological tweezers and stored in tubes containing 70º alcohol. External tick load was estimated on each animal using a prede ned part of body surface (head, neck or ears) and categorized into three levels according to the abundance: category 1 (high load) corresponded to more than 50 ticks, category 2 (medium load) to 30-50 ticks, and category 3 (low / null load) to 0 or less than 30 ticks on one of the prede ned parts of body surface. Blood samples were collected by jugular vein puncture (10-15 ml, live individuals) or cardiac puncture (15-20 ml, dead individuals). Aliquots of 1 ml each were stored at 4 ºC with EDTA, and at -20°C and -80°C without additives. The remaining blood was centrifuged and serum aliquots were stored at -80°C. A faecal sample was collected directly from the rectum of the marsh deer and stored without air in nylon bags at 4 ºC.
During necropsies, tissues were macroscopically evaluated and two samples (1 x 1 cm) of selected organs (heart, lungs, abomasum, liver, kidneys, intestine, lymph nodes, spleen, and brain) were collected, including part of normal tissue and part of injured tissue, if present. Tissue samples were xed in 10% buffered formalin solution (BFS) and frozen at -80°C.
Laboratory diagnosis. The aliquot of blood collected in EDTA tubes was used to determine haematological parameters using manual methods and an automated analyser (Re otron Plus, Roche, Mannheim, Germany). Blood chemical analyses were performed in an automated analyser (Metrolab 2100, Wiener lab, Rosario, Argentina), and concentration of total protein was determined using a portable refractometer (REF 302, Arcano, China). The blood parameters were compared with previously published data of wild marsh deer from Brazil [29].
Serological diagnoses of bluetongue virus (BTV), infectious bovine rhinotracheitis virus (IBRV), bovine viral diarrhoea virus (BVDV), brucellosis, foot-andmouth disease virus, Johne's disease, Leptospira interrogans, bovine leucosis, Q Fever, chlamydial abortion, and vesicular stomatitis virus (VSV) were performed by the National Service of Agri-Food Health and Quality (SENASA) according to the procedures described by the World Organization for Animal Health [60].
DNA was extracted from blood by phenol/chloroform method followed by a standard ethanol precipitation [62]. Blood samples were screened using PCR protocols targeting a fragment of the 16SrRNA gene for the Anaplasmataceae family [63,64], a fragment of the internal transcribed spacer 23S-5S of Rickettsia sp. [65], and a fragment of the 18SrRNA gene for Trypanosoma sp. [66] and Babesia/Theileria [67]. For positive samples, both strands of the ampli ed fragment were sequenced with a Big Dye Terminator v3.1 kit from Applied Biosystems and analysed in an ABI 3130XL genetic analyser from the same supplier (Genomic Unit, Consorcio Argentino de Tecnología Genómica (CATG), Instituto de Biotecnología, CICVyA, INTA). Raw les from each gene target were processed using the Vector NTI Advanced 10 program (Invitrogen). Both chromatograms were used for assembling a consensus sequence. The nal le in FASTA format was used for further sequence analysis.
Parasitic elements in faeces were identi ed, counted, and expressed as EPG (Number of eggs per gram of faeces) or OPG (Number of oocysts per gram of faeces) using a modi ed Wisconsin technique [68]. Infective larvae were cultured. Taxonomic identi cation of parasitic elements was performed according to the literature [69][70][71] using a magnifying glass (10-40X, Carl-Zeiss SV11, Germany).
Tissue samples xed in 10% BFS were processed using conventional histopathological protocols [72]. Samples were then embedded in para n wax and 5 μm sections obtained, which were stained with haematoxylin and eosin. Microscopic lesions found in each tissue were identi ed and categorized according to severity using a ve-degree scale (mild, mild-moderate, moderate, moderate-severe, severe).

Data analysis.
Given the high number of health variables measured, Multiple Correspondence Analysis (MCA) was performed for an exploratory analysis and to visually identify possible associations between body condition and exposure to the different infectious and parasitic agents [73]. MCA is particularly useful in epidemiological studies where large numbers of categorical variables are considered, as it allows to observe clustering of variables without the need to meet data distribution assumptions or a minimum number of cases, as required in other widely used statistical techniques [74,75].
The variables included were tick load, vector-borne agents (VBA), infection with nemathelminthes species, body score condition, and the period of the year in which the sampling was performed. Given the sensitivity of the MCA for events with low case numbers and even unique cases, only agents with more than one positive individual were included. The t-test was implemented in Epitools [76] to compare the means of the blood values obtained from our work with the means of two references. A signi cance level of P≤0.05 was used (2-tailed).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. OrozcoetalARRIVEChecklist.pdf SupplementarymaterialOrozcoetal.docx