Virulent Brucella nosferati infecting Desmodus rotundus has emerging potential due to the broad foraging range of its bat host for humans and wild and domestic animals

ABSTRACT Desmodus rotundus, vampire bats, transmit dangerous infections, and brucellosis is a hazardous zoonotic disease, two adversities that coexist in the subtropical and tropical areas of the American continent. Here, we report a 47.89% Brucella infection prevalence in a colony of vampire bats inhabiting the tropical rainforest of Costa Rica. The bacterium induced placentitis and fetal death in bats. Wide-range phenotypic and genotypic characterization placed the Brucella organisms as a new pathogenic species named Brucella nosferati sp. nov., isolated from bat tissues, including the salivary glands, suggesting feeding behavior might favor transmission to their prey. Overall analyses placed B. nosferati as the etiological agent of a reported canine brucellosis case, demonstrating its potential for infecting other hosts. To assess the putative prey hosts, we analyzed the intestinal contents of 14 infected and 23 non-infected bats by proteomics. A total of 54,508 peptides sorted into 7,203 unique peptides corresponding to 1,521 proteins were identified. Twenty-three wildlife and domestic taxa, including humans, were foraged by B. nosferati-infected D. rotundus, suggesting contact of this bacterium with a broad range of hosts. Our approach is appropriate for detecting, in a single study, the prey preferences of vampire bats in a diverse area, demonstrating its suitability for control strategies where vampire bats thrive. IMPORTANCE The discovery that a high proportion of vampire bats in a tropical area is infected with pathogenic Brucella nosferati and that bats forage on humans and many wild and domestic animals is relevant from the perspective of emerging disease prevention. Indeed, bats harboring B. nosferati in their salivary glands may transmit this pathogenic bacterium to other hosts. This potential is not trivial since, besides the demonstrated pathogenicity, this bacterium possesses all the required virulent arsenal of dangerous Brucella organisms, including those that are zoonotic for humans. Our work has settled the basis for future surveillance actions in brucellosis control programs where these infected bats thrive. Moreover, our strategy to identify the foraging range of bats may be adapted for exploring the feeding habits of diverse animals, including arthropod vectors of infectious diseases, and therefore of interest to a broader audience besides experts on Brucella and bats. The discovery that a high proportion of vampire bats in a tropical area is infected with pathogenic Brucella nosferati and that bats forage on humans and many wild and domestic animals is relevant from the perspective of emerging disease prevention. Indeed, bats harboring B. nosferati in their salivary glands may transmit this pathogenic bacterium to other hosts. This potential is not trivial since, besides the demonstrated pathogenicity, this bacterium possesses all the required virulent arsenal of dangerous Brucella organisms, including those that are zoonotic for humans. Our work has settled the basis for future surveillance actions in brucellosis control programs where these infected bats thrive. Moreover, our strategy to identify the foraging range of bats may be adapted for exploring the feeding habits of diverse animals, including arthropod vectors of infectious diseases, and therefore of interest to a broader audience besides experts on Brucella and bats.

worldwide. Although most bats are beneficial and ecologically relevant, the blood-feed ing Desmodus rotundus vampire bat is considered a plague because it is a reservoir and vector of a collection of reemerging and emerging microbial pathogens (1)(2)(3)(4)(5)(6). D. rotundus is found below 1,500 m of altitude in the tropical and subtropical areas of the Americas, with an increasing range toward the Northern hemisphere (7,8) attrib uted to climate change (9). Vampire bat populations have drastically increased due to livestock farming, their primary food source (6,(10)(11)(12). Consequently, most Latin American countries control them through extermination strategies (13,14).
D. rotundus bats bite their prey directly above the blood vessels, have anticoagulants in their saliva to aid blood flow, and use their tongue to sip blood (12,15,16). One prey per bat is generally observed; however, several vampire bats can forage on a single animal (12) and even feed on the same prey on various occasions (17). Through this process, they transmit dangerous infectious diseases (1)(2)(3)(4)(5)(6). After feeding, the vampire bat flies back to its colony in man-made or natural structures, with colonies ranging from ten to hundreds (16,18). These chiropterans are highly social animals that commonly regurgitate blood meals to feed each other (19). Through many decades of observation, it has been demonstrated that vampire bats forage from diverse prey, including domestic and wildlife animal species (6,11,12,15,16). Humans are also foraged by D. rotundus; however, the frequency and the number of cases may be underreported because most incidents occur in rural landscapes (10).
Brucellosis is a zoonotic bacterial infection of veterinary and public health relevance (20). In Costa Rica, nearly 11% of all bovine herds have brucellosis, causing a high economic loss in farm animals and a zoonotic risk (21,22). Antibodies against Brucella organisms have been found in 9.4% of vampire bats in Brazil (23). In addition, Brucella suis biovar 5, closely related to those isolated from rodents, has been identified in insectivorous bats from the Caucasus (24,25). Likewise, a new species of Brucella organism has been isolated from humans working in the Amazonian rainforest in French Guiana, from which the infection source remains unknown (26).
Here, we describe the prey preferences of D. rotundus vampire bats infected with a new species of virulent Brucella organism in a region of the tropical rain premontane forest of Costa Rica and rural surroundings. This new species, named B. nosferati sp. nov., is a primary pathogen of bats that resides in different organs, including the salivary glands, with the potential to infect other mammals. The colony of D. rotundus-infected bats studied foraged on a wide range of domestic and wildlife animals and humans, which may become hosts of this newly discovered pathogenic bacterial species causing brucellosis.

Study site and bat colony characteristics
Seventy-one vampire bats (43 female and 28 male) were collected in a cave adjacent to the National Park Piedras Blancas of Costa Rica (141.2 km 2 ), located in the far eastern region of the park. All vampire bats were negative for the presence of rabies and coronavirus. Fifty-nine were adults, six were subadults, four were juveniles, and two were newborn bats. In addition, seven pregnant bat females and the corresponding fetuses were included in the study (Dataset S1). The cave of the D. rotundus colony is in a tropical premontane rainforest area (Fig. 1). According to the census and the annual survey of the Animal Health Service (SENASA) of Costa Rica (Dataset S1), all surrounding farms within a radius of ~4 km possess cattle; some have horses, two farms have sheep, and three have poultry for commercial purposes. Most farms have a few chickens and ducks (~5), dogs (~2), and cats (~1) roaming around the shelter. None of the farms in a range of 8 km possess pigs. In the park territory, there are nearly 100 species of wild mammals and over 400 bird species (27)(28)(29)(30)(31)(32).

Identification of pathogenic Brucella organisms in bats
Since vampire bats feed on cattle, and bovine brucellosis is highly prevalent in Costa Rica (21), we anticipated these chiropterans could be infected with Brucella since this is a blood-borne pathogen circulating within cells (20). Using serological assays, we found that 33.8% (24/71) of the bats had antibodies against the Brucella lipopolysaccharide (LPS) ( Table 1), a higher proportion than the 9.4% reported in Brazil (23). So far, the only Brucella species identified in Costa Rica's livestock is Brucella abortus (21,22). Since bovines are the most common prey of vampire bats, we hypothesized that bats could become infected with this bacterium.  Most of the vampire bats did not reveal noticeable internal or external lesions. However, the gross anatomical and histological examination and immunohistochemistry of 4 pregnant vampire bats out of 16 showed a severe necrosuppurative placentitis with a mummified fetus characterized by necrosis of double-layered trophoblasts, numerous degenerate leukocytes, and karyorrhectic cellular debris characteristic of brucellosis (33). Strong Brucella sp. immunolabeling was associated with necrosis and inflamma tion of the double-layered trophoblasts, demonstrating extensive intracellular bacterial parasitism (Fig. 2). Following this, we attempted isolation and characterization of these pathogenic bacteria.
Brucella organisms were isolated in 6 male and 11 female bats, of which one was a nursing newborn male and one juvenile male (Dataset S1). As in ruminants, the bacterium was isolated from the infected bats' placenta, fetus, milk, and other tissues ( Table 2). Brucella organisms were also isolated from the salivary glands of 11.8% of the infected bats, an unprecedented finding in brucellosis. This event suggests that transmission through the bats' feeding habits might be plausible to other mammals since the minimal infectious dose for pathogenic Brucella organisms has been estimated from 5 to 50 bacteria (34).
The bacteriological characterization showed that the bat isolates were smooth brucellae resembling the pathogenic Brucella sp. BCCN84.3 isolated from a dog orchiepididymitis case in Costa Rica (35,36), except that the Brucella of the bat strains were dominant M type while the BCCN84.3 strain was dominant A type ( Table 3)  The biochemical profile of the Brucella vampire bat strains was characterized by the activity of Ala-Phe-Pro-arylamidase, β-N-acetyl-glucosaminase, L-proline arylamidase, tyrosine arylamidase, α-glucosidase, and urease and by the utilization of D-glucose, Dmaltose, palitose, sucrose, and D-trehalose and the alkalinization of L-lactate and succinate (Dataset S2). The fatty acid methyl ester (FAMEs) profile revealed some clustering of the bat strains. However, several of them were intertwined with other Brucella species (Fig. 3B). All the Brucella vampire bat isolates had the characteristic longchain fatty acids and lactobacillic acid of the classical virulent Brucella (Dataset S2). The matrix assisted laser desorption/ionization timeofflight mass spectrometry (MALDI-TOF) proteomic analysis demonstrated that the Brucella bat isolates and the etiological agent of canine orchiepididymitis, the BCCN84.3 strain, were in the same cluster, separated from other brucellae (Fig. 4). Although the bacteriological profile and mass spectrometry analysis suggested a different Brucella species, the results were not definitive. These properties caught our attention since they did not match the Brucella organisms isolated in Costa Rica (22).

Phylogenetic characterization of Brucella isolates
The Brucella bat and BCCN84.3 strains displayed a Bruce-ladder profile indistinguishable from that of B. suis 1330 strains (Fig. 5A). However, the variable number of tandem  Research Article mSphere repeats-16 loci (MLVA-16) analysis resolved that the bat and BCCN84.3 isolates clustered apart from other Brucella species (Fig. 5B). Likewise, the phylogenetic analysis of the omp2a and omp2b porin genes showed that Brucella bat and BCCN84.3 isolates grouped in a distinct cluster from other Brucella species (Fig. S1), displaying a recombination event in a region close to the 5′, which is identical to the omp2b porin sequence. Among the Brucella genomes, 210,174 single-nucleotide polymorphisms (SNPs) were used for whole-genome sequence (WGSA) phylogenetic analyses (Dataset S3), showing that the bat strains clustered tight in a branch far from other brucellae. The closest relative was B. amazoniensis isolated in the rainforest of French Guiana (Fig. 6). The number of SNP differences between the Brucella bat and the BCCN84.3 strain ranged from 80 to 150, significantly less than those described in other Brucella species. For instance, the Brucella bat and Brucella sp. BCCN84.3 isolates, compared with B. amazo niensis and B. suis 1330 strain, differed in 2,378-2,387 and 7,159-7,194 polymorphic sites, respectively. From these, 6,025 are in coding region 196 with a dN/dS ratio of 0.54 (Pvalue < 0.0001) (Dataset S3). These analyses demonstrated that the bat isolates and the Brucella sp. BCCN84.3 strain belonged to the same distinctive classical species of B. nosferati, despite being isolated years apart from different hosts. Following a taxonomical approach, we named the bat isolates Brucella nosferati sp. nov., reminiscent of the Nosferatu vampire legends of Transylvania. From the genomic and bacteriological perspectives, the pathogenic Brucella sp. BCCN84.3 strain belonged to the same species and, from now on, the type strain named B. nosferati BCCN84.3, the first species of this group isolated in Costa Rica (35,36).
According to the WGSA, B. nosferati had two chromosomes with no plasmids, nine IS711 elements, recent recombination events, and a similar number of anomalous regions shared with other classical brucellae (Fig. S2), however, with a distinct lineage and genomic islands (Fig. S3). Likewise, the WGSA showed all the genes that encode virulence factors such as VirB operon (BAW_20058 to BAW_20068), VjbR (BAW_20116), two-component system BvrR/BvrS (BAW_12006 to BAW_12007), β-cyclic glucan (BAW_11641), and flagellumlike system (BAW_11648 to BAW_11649) are conserved in the Brucella bat isolates. Genes coding for the toll-interleukin-1 receptor domaincontaining protein BtpA predicted as a VirB effector (BAW_10265) and putative integra ses (BAW_10237, BAW_10274) were absent. Likewise, the manBOAg (BAW_10538), tentatively involved in the LPS-core mannose synthesis, was 48 bp, which corresponds to the size of B. ovis (BOV_0540) and B. abortus 2308W (BAW_10538), but it was shorter than the corresponding B. melitensis (BMEI1396). It is clear that B. nosferati is a classical brucellae equipped with all the virulence arsenal similar to the most pathogenic Brucella species (20,37).

The D. rotundus foraging prey taxa
Since B. nosferati was found in the salivary glands of 11.8% of the infected bats, we decided to explore the foraging prey range through proteomics of the bat's intestinal contents. Vampire bats generally travel no more than ~4 km from their lair. However, we estimated the presence of bird and mammal species as potential prey at a radius of 8 km of the bats' cave. This range is the average maximum distance that D. rotundus may fly from their lair in 1 day (18) and a significant distance for the mobility of the animal species within the examined area. The study included 2 subfamilies, 18 families, 1 parvorder, and 3 suborders representing 49 mammal and 10 bird species, which were the most likely prey due to their size, presence, and abundance in the area (Table S1). We used stringent conditions for peptide selection (see Materials and Methods section) and a sorting algorithm for protein assortment and taxon identification (Fig. 7).
Through the analysis of the available intestinal contents of 37 bats, a total of 54,508 peptides corresponding to 7,206 proteins were initially identified and finally sorted into 7,203 unique peptides corresponding to 1,019 plasma, 396 erythrocyte, and 106 mitochondrial proteins (Fig. 8A). Accordingly, 23 taxa prey groups were identified (Fig.  8B). Although presumptive for the moment, an educated guess of the genus/species prey preference can be assessed based on the abundance and accessibility of the various prey in the studied area (27)(28)(29)(30)(31)(32). As expected, the most common prey were the larger and more abundant animals such as bovines, humans, and horses; but the choices were diverse. No vampire bat preference for either native or exotic animal species and vice versa was noted. Foraging on small mammals and birds seems unexpected, though there are reports of vampire bats chasing armadillos and feeding from squirrels, rats, chickens, among others (12,38). Likewise, there have been cases of D. rotundus rabies virus variants transmitted by squirrels in Costa Rica, indicating that vampire bats also prey on these small mammals (39). In Costa Rica, nonindustrial poultry is usually kept in open chicken coops readily accessible to vampire bats attacks.
The intestinal content of one vampire bat showed up to 20 different taxa prey groups. This variety is unlikely to mirror the foraging prey spectrum of single bats. Instead, it reflects the colony's broad and diverse foraging range since vampire bats share their meals with their mates (19). Likewise, the fact that some bats are depicted with a reduced prey number, or none (Fig. 9A), does not mean they did not forage. Instead, the proteome in these animals was ambiguous, and unique peptides were not identified; therefore, they were not associated with a specific taxa group. There was no statistical difference (P < 0.13) between the number of prey foraged by individual bats with Brucella infection. Likewise, there was no statistical difference (P < 0.77) between D. rotundus sex and prey preference (Fig. 9A).
The number of proteins identified in each animal of the bat colony also indicates that cows, humans, and, to fewer extent, sheep, equines, and hog peccary (Suina) were among the preferred prey from which more blood was taken. The remaining animals were in the same range (Fig. 9B). However, the number of proteins in the bat's intestinal content may be biased by the availability of data and influenced by the amount of blood taken and the time lapse between bloody meals.

DISCUSSION
We found a high fraction (24%) of the colony of vampire bats infected with B. nosferati, being females infected in a higher proportion (65%) than males (35%). Due to the social behavior of vampire bats, it is plausible that once the bacterium reaches a colony, the infection is broadly distributed, paralleling what happens in ruminants (20). Like other pathogenic Brucella species, B. nosferati causes placentitis, induces fetal death, colonizes the mammary gland, is secreted in milk, causes orchiepididymitis, and is present in different tissues, all features of classical brucellosis (20) that can hamper the reproductive ability of bats. This pathology may have a reproductive impact on bat populations, considering that this pathogenic Brucella could potentially spread to other chiropterans since they often share roosts with other bat species (11). It may be that the high livestock prey abundance favors B. nosferati parasitism in vampire bats. It has been shown that depending on the prey's accessibility, the vampire bats' innate and adaptive immune responses toward pathogens can shift (46). In regions of high-livestock accessibility, the microbicidal capacity of bats develops toward stronger innate immunity relative to adaptive response favoring more significant chronic stress. This shift in the immune response may affect the infection rate of bats to certain diseases, such as Brucella organisms whose immune control depends on a long-lasting adaptive immune response, as shown before (47,48). B. nosferati possesses all the required virulence arsenal of other zoonotic Brucella species and, therefore, can become a primary pathogen of animals and humans, as already demonstrated in nature. Since the bacterium can invade the salivary glands of bats, it is therefore feasible that B. nosferati could be transmitted between vampire bats FIG 7 Flowchart of the strategy followed to curate the peptide data for selecting the proteins of a given taxon group. For details of the selected proteins, see Dataset S1. Except for pigeons and squirrels, which are abundant in the park and surroundings, unlikely foraging vampire prey, such as chiropterans, animals of less than 600 g, aquatic birds and mammals, raptor birds, and nocturnal birds, were excluded from the analysis. D. rotundus proteome was included as a control for excluding peptides.
Research Article mSphere and also to domestic and wildlife species, including humans. The B. nosferati BCCN84.3 strain causing dog orchiepididymitis (35,36) supports this concept.
Considering that the B. nosferati was isolated from the bats' gut contents of 47% of infected animals, we cannot dismiss the possibility that bats could have taken this bacterium from a third-party reservoir. We neither have attempted isolation in other potential hosts nor conducted a survey on antibodies against Brucella in wildlife. However, others have reported the presence of anti-Brucella antibodies in the blood of tropical wildlife mammals such as peccary, agouti, capybara, coati, raccoon, foxes, wild felids, otters, monkeys, and tyras (23,(49)(50)(51)(52), all animal species identified in this work as prey of infected vampire bats. Moreover, the B. nosferati BCCN84.3-infected dog lived in an area inhabited by D. rotundus bats (34).
Of note is the recent description of an emerging B. amazoniensis infecting humans working in the deep Amazonian rainforest, which probably acquired the bacteria through contact with tropical wildlife, including bats (26). Although this bacterium is a different species, it branched closer to B. nosferati than other species. Likewise, a high seroprevalence in wild peccaries and capybaras was found in a tropical region of Venezuela with the concomitant isolation of organisms typed as B. suis using classical bacteriology (50,53). However, conventional bacteriological typing cannot distinguish B. suis from B. nosferati (35). The closeness to areas with vampire bat attacks (40,54) and the absence of pigs (B. suis reservoirs) in the vicinities where these sylvatic bacterial species were isolated (26,50,53) suggest the presence of zoonotic Brucella organisms in these tropical South American areas.
Our approach identified the feeding preferences of a Brucella-infected D. rotundus colony. Many studies through various decades have documented that D. rotundus feeds on several domestic and wildlife animals (6,11,12,16,18). In a single study, we identified the prey preference spectrum of a colony of Brucella-infected vampire bats in a given Research Article mSphere complex area, demonstrating the potential of our approach for control strategies in Latin America, where vampire bats thrive. This practice is particularly relevant when sudden outbreaks of infectious diseases such as rabies or brucellosis emerge in a given area, and the potential range of reservoirs requires fast identification for sanitary actions and epidemiological surveillance. Moreover, the same approach may be used to identify arthropod vector prey preferences. It has been shown that many animals display defensive behavior against vampire bat attacks (41-43), escaping from their foraging range, partly explaining the bat colony's wide prey range, including wild, dangerous predators and relatively small species of mammals and birds. Nevertheless, limitations in the availability of protein data for a given animal species cannot be circumvented. We aimed to diminish this bias by increasing specificity by selecting and curating major blood proteins since most taxa have pertinent information on these proteins. In addition, we broaden the chances by selecting higher taxa levels rather than attempting species or genus identification. However, by increasing specificity, the method's sensitivity was diminished, and in some cases, we could not identify the taxa group unambiguously.
Vampire bats forage on humans in the surroundings of the Piedras Blancas National Park. Close to 35 bat bites in humans were reported in 2021 by the Costa Rican Ministry of Health (44). However, the number of bites is expected to be higher since these events are often unreported (45). Several investigators have also described a high incidence of vampire bats foraging on humans (up to 88% of the inhabitants) in tropical ecological conditions similar to those presented here (10,17,45,(55)(56)(57)(58)(59). Humans are abundant potential diurnal prey that have not developed natural defense mechanisms against vampire bat attacks when these chiropters hunt for blood at night. Bats frequently enter rural houses and shelters in tropical areas of Costa Rica (45,60). Considering the high prevalence of Brucella infections in D. rotundus, further studies are needed in prey victims to assess the zoonotic potential of B. nosferati. Unfortunately, in Latin America, including Costa Rica (22), bacterial isolation is seldom attempted in animals and humans, precluding the opportunity to identify the Brucella species (61).
It has been noticed that vampire bats reproduce better and in larger groups, precisely in those border areas where foraging is abundant, adjusting their immune response and favoring certain diseases over others (46). Following this, it is not surprising that D. rotundus is becoming a plague in certain areas where livestock and humans are abundant (62). Our recent experience suggests that bats may be reservoirs of novel and dangerous infectious diseases for humans and animals (63). Therefore, identifying a new species of pathogenic bacteria in D. rotundus vampire bats and establishing the range of potential hosts in a given environment are relevant for epidemiological studies and for preventing zoonotic infectious diseases, including brucellosis.

Vampire bat collection and pathogen detection
Seventy-one D. rotundus were captured during the day using nets in a cave in the Pacific tropical rainforest in March 2020, Piedras Blancas National Park, Puntarenas, Costa Rica (8°43'40.1''N and 83°11'15.2''W) (Dataset S1). The bats were transported inside cotton bags to the Animal Health Service of Costa Rica, euthanized through intramuscu lar injection of xylazine and ketamine, and blood and organ samples were collected through systematic necropsies. Although vampire bats were collected as part of the National Surveillance Bovine Rabies Program of Costa Rica (see Ethics Statement), their blood and tissues were also screened for additional pathogen infections. Following this, the presence of coronavirus and rabies virus was explored through polymerase chain reactions as described (13,64) and brucellosis by the rose bengal agglutination test (RBT) and cELISA for detecting antibodies against Brucella LPS (65, 66).

Bacterial isolation and pathological studies
After necropsy, several tissues were tested for Brucella organisms ( Table 2) following regular bacteriological procedures (65,67,68). Forty-two Brucella strains were isolated from the different organs of 17 D. rotundus vampire bats (Dataset S1). The placenta of a D. rotundus was subjected to histological studies following hematoxylin and eosin staining of the sections and immunolabeling of intracellular brucellae as described before (33,67). Positive control tissue consisted of B. abortus-infected bovine placenta. No primary antibody was applied to the negative control. Several Brucella species were used as a reference.

Bacterial phenotypic characterization
All Brucella-compatible colonies were analyzed as described (65). A biochemical profile was evaluated using the automated Vitek 2 System (bioMérieux) with the GN ID card to identify Gram-negative bacilli following the protocols described by the manufacturer. The vampire bat Brucella isolates and reference Brucella species, grown and harvested under the same conditions (Dataset S2), were examined by whole-cell MALDI-TOF proteomics and gas-liquid chromatography/ms (glc/ms) of the derived FAMEs as described before (35,69,70).
For MALDI-TOF, mass spectra were acquired using a VITEK MS Plus (bioMérieux), ranging from 2,000 to 20,000 Da at a laser frequency of 50 Hz. VITEK MS RUO (Research Use Only) database and SARAMIS software (bioMérieux) were used to capture spectra for each isolate. A consensus spectrum was built for reference strains and isolates. Principal component analysis was constructed using the Spearman correlation algorithm.
The glc/ms (Agilent Technologies 6850) was performed using a 25-m × 0.2-mm cross-linked phenyl-methyl silicone fused silica capillary column HP 19091B-102 (Agilent Technologies Inc., Santa Clara, CA, USA). Methyl ester determination was carried out according to the MIDI instruction manual of Technical Note #101 (MIS, MIDI Inc., Newark, DE, USA) as previously described (35,69). The same culture and harvesting conditions for all bacteria were used during the experiment. Extraction of LPS was performed using the SDS-proteinase K method as described before (71). Brucella LPS was analyzed in 4%-10% gradient SDS-PAGE stained by the periodate-alkaline silver method or revealed by Western blotting with a monoclonal antibody against the A>M LPS epitope of the O-chain (72).
Dendrograms were constructed based on the absence or presence of peaks using an agglomerative hierarchical clustering algorithm, using XLSTAT 2022 (Version 1.2, 2022, Addinsoft, Inc., Brooklyn, NY, USA). The proximities were calculated using squared Euclidean distance and the aggregation using the unweighted pair group average method. One-way analysis of variance followed by Dunnett's test or multivariate analysis of variance was used to determine statistical significance.

Determination of the prey range of vampire bats
The prey range of D. rotundus vampire bats was estimated by species-oriented proteomic analysis of the intestinal content. The intestinal content, mainly constituted of tubular clots, was removed and placed in 1 mL sterile plastic tubes and frozen at −80°C until used. About 25 mg of intestinal content was suspended in 25 µL deionized double-dis tilled sterile water, gently stirred for 30 minutes, and then vortexed twice. The tubes were centrifuged at 12,000 rpm for 5 minutes, and the supernatants separated from the precipitate. Protein concentration was estimated using a Qubit Fluorometer (Life Technologies). Then, 130 µg of the 2-mercaptoethanol reduced proteins (5 minutes at 95°C) was separated by SDS-PAGE using 4%-20% precast gradient gels (Bio-Rad). Finally, the gels were stained with Coomassie blue R-250, as described elsewhere (83). The stained SDS-PAGE protein bands were excised from gels and subjected to reduction and subsequent enzyme digestion (83). The proteomic analysis was performed through nESI-MS/MS on a Q-Exactive Plus mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) as described before (83). MS/MS spectra were processed to assign peptide matches to known protein families by similarity with sequences in the UniProt/SwissProt databases using Peaks X (Bioinformatics Solutions, Canada). Cysteine carbamidomethy lation was set as a fixed modification, while deamidation of asparagine or glutamine and methionine oxidation were set as variable modifications, allowing up to two missed cleavages by trypsin.
Animal species identification was supported on the datasets constructed from Proteins-NCBI/UniProt/wiss-Prot (https://www.uniprot.org/) of the mammal and bird species found in the perimeter of the 8 km range where the D. rotundus bats were captured (Table S1). Due to the incomplete genomic and proteomic databases for most Costa Rican wildlife animals and birds, we complemented the data with the broader available information of the closest phylogenetic species for a given group (Table S1). The estimate was narrowed to 47 mammals and 14 bird species included in 25 taxa groups located within the foraging range of bats (27)(28)(29)(30)(31)(32), making, in some cases, the identification at the species or genus level feasible (Table S1). Because of the close phylogenetic relationships of the species within target animal taxa, we expressed the identification at the subfamily, family, parvorder, or suborder level, according to the availability of the datasets. For example, in the case of the four opossum species (Didelphidae) present in Piedras Blancas National Park (27)(28)(29)(30)(31)(32), the available databases were complemented with those of Monodelphis domestica and Gracilinanus agilis and other opossums from South America. Any target peptides for proteins of these South American animals were estimated to belong to one of the Costa Rican opossums. A similar approach was used for animal species from which data were insufficient. Likewise, the BLAST search for confirming human-derived peptides converged in some cases with apes and old-world monkey proteins such as chimpanzees, macaques, or papions, and therefore assigned to humans. Once the presumptive animal species was identified, it was assigned to a generic taxadefined group (Table S1).
The data were curated as a final selection step by selecting only the most abundant plasma and red blood cell protein families (84) (Dataset S1). Due to their taxonomical value, major mitochondrial proteins were also included in the analysis. Extracellular, mucous-secreted, nuclear, and non-abundant red cell proteins were excluded from the study. Because of the close phylogenetic relationship among the various animal groups, we depicted the results at the nearest taxonomical rank identified above the genus level, such as family (e.g., Canidae and Felidae), subfamily (e.g., Bovinae and Caprinae), parvorder (e.g., Platyrrhini), or suborder (e.g., Suina and Folivora) (Table S1). For example, if one or more different presumptive proteins of Felis catus (domestic cat) and Panthera onca (jaguar) were identified, the match was considered positive and assigned to the family Felidae. This broader definition reduced the chances of taxonomic errors at the species or genus level. Whenever necessary, peptides were examined by a peptide-pro tein BLAST of the existing mammal and bird databases for unambiguous identification.
If a given protein species' peptide matched a protein from a different generic taxon in the database, the peptide and the corresponding animal taxa were disregarded. Except for pigeons and squirrels, which are abundant in the park and surroundings, unlikely foraging vampire prey, such as chiropterans, animals of less than 600 g, aquatic birds and mammals, raptor birds, and nocturnal birds, were excluded from the analysis. D. rotundus proteome was included as a control for excluding peptides. File repetitions were removed from the datasets. The protein parameters for match acceptance were set to false discovery rate <0.1%, −10lgP protein score ≥50, and unique peptides for a given taxon were considered. Proteins of the respective animals were identified with the available GenBank protein database for the corresponding species (Dataset S1).

ACKNOWLEDGMENTS
We thank Ms. Isabel Quirós for her valuable assistance during the proteomic analyses of intestinal bat content and Vicerrectoría de Investigación (Universidad de Costa Rica) for partial support to the Proteomics Lab at Instituto Clodomiro Picado. We thank Sebastian Brünink and Wendy Jo-Lei for their fieldwork support.

DATA AVAILABILITY STATEMENT
All genomic data have been deposited in the National Center for Biotechnology Information (NCBI) under the accession numbers indicated in Dataset S3. All raw data are available upon request. These genetic resources were accessed in Costa Rica according to the Biodiversity Law #7788 and the Convention on Biological Diversity, under terms of respect to equal and fair distribution of benefits among those who provided such resources under CONAGEBIO Costa Rica permit # R-CM-UNA-005-2022-OT-CONAGEBIO.

ETHICS APPROVAL
The capture of D. rotundus vampire bats is part of the National Surveillance Bovine Rabies Program by the Animal Health Service (SENASA) of Costa Rica, following the Executive Decree 24405-MAG of 8 June 1995, of the Ministry of Agriculture of Costa Rica in which the control of the D. rotundus is declared as a compulsory (13). The euthanasia of the vampire bats followed the AVMA Guidelines (85).

ADDITIONAL FILES
The following material is available online.

Supplemental Material
TABLE S1 (mSphere00061-23-s0001.docx). Animal species included in the proteomic databases to define the target taxa prey of D. rotundus in the Piedras Blancas National Park of Costa and the surrounding area. Figure S1 (mSystems00061-23-s0002.tif). Phylogenetic analysis of the omp2a and omp2b nucleotide sequences of B. nosferati and other Brucella strains. B. nosferati grouped in a distinct cluster from other Brucella species. Figure S2 (mSystems00061-23-s0003.tif). Recombination events in representative Brucella species. Each event is shown by a vertical block ordered along the genome. Figure S3 (mSystems00061-23-s0004.tif). Presence and absence of anomalous regions or genomic islands in the Brucella genomes.