Using DNA metabarcoding for simultaneous inference of common vampire bat diet and population structure

Abstract Metabarcoding diet analysis has become a valuable tool in animal ecology; however, co‐amplified predator sequences are not generally used for anything other than to validate predator identity. Exemplified by the common vampire bat, we demonstrate the use of metabarcoding to infer predator population structure alongside diet assessments. Growing populations of common vampire bats impact human, livestock and wildlife health in Latin America through transmission of pathogens, such as lethal rabies viruses. Techniques to determine large‐scale variation in vampire bat diet and bat population structure would empower locality‐ and species‐specific projections of disease transmission risks. However, previously used methods are not cost‐effective and efficient for large‐scale applications. Using bloodmeal and faecal samples from common vampire bats from coastal, Andean and Amazonian regions of Peru, we showcase metabarcoding as a scalable tool to assess vampire bat population structure and feeding preferences. Dietary metabarcoding was highly effective, detecting vertebrate prey in 93.2% of the samples. Bats predominantly preyed on domestic animals, but fed on tapirs at one Amazonian site. In addition, we identified arthropods in 9.3% of samples, likely reflecting consumption of ectoparasites. Using the same data, we document mitochondrial geographic population structure in the common vampire bat in Peru. Such simultaneous inference of vampire bat diet and population structure can enable new insights into the interplay between vampire bat ecology and disease transmission risks. Importantly, the methodology can be incorporated into metabarcoding diet studies of other animals to couple information on diet and population structure.


Table of contents
Supporting Information S1: Metabarcoding of common vampire bat blood meal and faecal samples Page 2--7 Table S1a: Overview of analysed common vampire bat blood meal and faecal samples Page 2 Table S1b: Samples with common vampire bat and vertebrate prey taxa assignments from metabarcoding analyses using the 16s and COI primer sets.
Page 3 Table S1c: Number of prey taxa identified within the common vampire bat samples that had prey taxa assignments in metabarcoding analyses using 16s and COI primer sets Page 3 Table S1d. Vertebrate prey availability and prey eaten at the six areas included in the study Page 4 Table S1e. Mammalian prey availability and prey eaten at the six areas included in the study Page 5 Figure S1a. Species accumulation curve for vertebrate prey detected in common vampire bat blood meal samples through metabarcoding of blood meal and faecal samples with 16s and/or COI primer sets Page 6 Figure S1b: Alignments of 16s and COI common vampire bat haplotypes Page 7 Supporting Information S2: Metabarcoding of blood meals from hairy--legged vampire bats Page 8-- 10  Table S2. Overview of the taxa identified in three blood meal samples from hairy-legged vampire bat (Diphylla ecaudata) Supporting Information S1: Metabarcoding of common vampire bat blood meal and faecal samples  Table S1b. Samples with common vampire bat and vertebrate prey taxa assignments from metabarcoding analyses using the 16s (Taylor 1996) and COI (Geller et al. 2013;Leray et al. 2013) primer sets.  Table S1c. Number of prey taxa identified within the common vampire bat samples that had vertebrate prey taxa assignments in metabarcoding analyses using 16s and COI primer sets.

Blood meal
Individual faecal samples

Pooled faecal samples
Min. -max. (average) no. vertebrate prey taxa per sample where prey is detected  Table S1e. Mammalian prey availability (expected counts, calculated using livestock densities) and prey eaten (observed counts, detected as prey through metabarcoding) in the six areas included in this study, and for all sites combined. All p-values show significant differences between observed and expected prey counts (p<0.05). However, for some areas, there are expected counts less than 5, and for these the results must be interpreted with care.  Figure S1b. Pairwise alignment of 16s and COI common vampire bat haplotype sequences. Image: Geneious version 6.1 created by Biomatters, http://www.geneious.com.

Supporting Information S2: Metabarcoding of blood meals from hairy-legged vampire bats
Three hairy-legged vampire bat (Diphylla ecaudata) blood meal samples collected in the MDD134 site in the Amazon were metabarcoded alongside the common vampire bat (Desmodus rotundus) samples with the mammal 16s (Taylor 1996) and metazoan COI (Geller et al. 2013;Leray et al. 2013) primer sets. Blood meal samples were collected from captured and morphologically identified hairy-legged vampire bats. One hairy-legged vampire bat OTU was identified by each of the markers. The 16s OTU only had a 96% match to Diphylla ecaudata in ncbi genbank, while the COI hairy-legged vampire bat OTU had matches to Diphylla ecaudata spanning from 92.88-99.35% indicating mitochondrial intraspecific diversity in the hairy-legged vampire bat. As only bird prey was detected, only the metazoan COI primer set could be used to identify prey. With the COI primer set, four vertebrate prey OTUs were identified of which three were found in one sample and two samples contained the same OTU (Table S2). The BOLD database (boldsystems.org) was used to identify the OTUs. Criteria for taxonomic assignments were as follows: i) Prey species assignment: 100% match to only one species. ii) Prey genus assignment: 100% matches to more than one species within the same genus. As the remaining OTUs had poor reference database coverage, comparisons were made against all barcode records in BOLD. iii) Prey family assignment: identity from 96.12 to 88.03 within the same family. In the sample where Tinamous sp. (Tinamoformes) were identified (Table S2), an OTU with amount of sequences and low identity matches to the order Tinamiformes were also found. Therefore, this OTU was assumed to be an artefact. Nematode taxa assignment followed that of the common vampire bat. Hairy-legged vampire bat has been reported to rely solely on birds as prey (Greenhall et al. 1984) (although potential predation on humans have been reported (Ito et al. 2016)). In agreement with this, we only detected birds in blood meal samples from three individual hairylegged vampire bats. Two of the bats had preyed on chicken (Gallus sp.), while the third bat had preyed on two wild birds, spix's guan (Penelope jacquacu) and tinamous (Tinamus sp.) (Table  S2). To our knowledge, no other study has identified other birds than chicken in any vampire bat species' blood meal or faeces. The spix's guan is a large, fairly common arboreal bird weighing up to ca 1.4 kg and known to occur in rainforest in the area where the samples were collected (IUCN 2016;Del Hoyo et al. 2017). Species in the tinamous genus are larger bird species weighing up to ca. 1.9 kg, they roost in trees, and the genus is known to occur in the area where the hairy-legged were caught (Khanna 2005;Del Hoyo et al. 2017). Although based on just one sample from one individual, this gives the first evidence that the hairy-legged vampire bats can prey on spix's guan and tinamous and that these wild birds might offer them a reliable and accessible food source in the MDD134 site. Table S2. Overview of the taxa identified through metabarcoding with the COI primer set (Geller et al. 2013;Leray et al. 2013) in three blood meal samples from hairy-legged vampire bat (Diphylla ecaudata) collected in the MDD134 site in the Amazon ecoregion. Vertebrate prey are listed as English name, order, family, genus and species.