Specialization of bat-fly interactions at different elevations in a montane forest of northern Peru Authors: David Minaya, Juan J. Pellón, Carla Yauris, Kristhie Pillaca, Balder Choza, Jaime Pacheco, Gustavo Graciolli, José Iannacone Published: 2026 ABSTRACT Hippoboscoidea flies exhibit highly specific ectoparasitic relationships with bats, shaped by both intrinsic factors (e.g. bat behaviour) and extrinsic factors (e.g. land use). Understanding the dynamics of these parasite–host interactions is essential for uncovering co-evolutionary patterns and informing conservation strategies. To this end, we studied bat–fly interactions across different elevations in a montane forest of Amazonas, northern Peru. The most abundant bats were Carollia brevicauda, C. perspicillata and Sturnira oporaphilum, while Paraeuctenodes similis and Trichobius joblingi were the most common flies. Most flies exhibited monoxenous host specificity. Bat–fly interaction networks revealed high modularity and specialization at both local and regional scales. Modules typically grouped bat species of the same genus or subfamily, suggesting that phylogenetic constraints and roosting behaviour may shape those interaction patterns. Nestedness within modules (compound structure) emerged in the aggregated regional network, aligning with the integrative hypothesis of specialization. Although network structures were broadly similar across sites, species turnover contributed to subtle differences in module composition and specialization. These differences were congruent with the changes in species roles of certain bats and flies. This study represents the first of its kind in Peru and addresses significant knowledge gaps in the ecology of bat–fly interactions in the Neotropics. Introduction Bat flies (Diptera: Hippoboscoidea) are obligate, blood-feeding ectoparasites that exclusively parasitize bats and are among the most abundant and frequent hematophagous parasites in this mammalian group (Hrycyna et al., 2019). These flies typically exhibit a high degree of host specificity, often associated with long-term evolutionary relationships between parasite and host lineages (Dick et al., 2010). However, host specificity can also be influenced by intrinsic factors such as host behaviour, health and body size, as well as extrinsic environmental factors (Palheta et al., 2020). Studying these tightly linked parasite–host interactions offers valuable insight into co-evolutionary processes and the mechanisms shaping host specialization (Dick and Patterson, 2006; Hiller et al., 2021). Globally, bat flies are divided into two main families with distinct biogeographic distributions: Nycteribiidae, more diverse in the Eastern Hemisphere, and Streblidae, predominantly found in the Western Hemisphere, especially in the Neotropics (Soares et al., 2013; Graciolli and Dick, 2018; Barbier et al., 2019; Graciolli et al., 2019). In South America, representatives of both families coexist, parasitizing a wide range of bat species (Biz et al., 2023; Zapata-Mesa et al., 2024). Despite their ecological relevance, detailed studies on bat–fly associations remain scarce in many parts of the Neotropics, including Peru. Peru harbours remarkable bat diversity, with 196 species currently recorded (Pacheco et al., 2021; Velazco, 2023; Diaz et al., 2025), of which at least 75 are known to host ectoparasites, including 66 species of bat flies (Minaya et al., 2021). However, most records of bat–fly diversity in Peru are concentrated in the lowland Amazonian rainforests of Loreto and Madre de Dios (Theodor, 1967; Guerrero, 1996a; Graciolli et al., 2007; Autino et al., 2011; Gettinger, 2018; Gettinger et al., 2020; Morales-Malacara and Guerrero, 2020). In contrast, montane forests – despite being among the most bat–diverse ecosystems in the Neotropics (Chaverri et al., 2016; Bogoni et al., 2021) – remain poorly studied in terms of bat ectoparasite associations. Available information from these ecosystems is sparse and typically limited to isolated records of ectoparasite presence, rather than comprehensive analyses of interactions (Biz et al., 2023; Zapata-Mesa et al., 2024). For example, in the montane forests of Amazonas, northern Peru, a key region within the Andean forest belt, only a single study on bat flies has been published (Ibáñez and Jara, 2008), highlighting a substantial gap in our understanding of host–parasite relationships in these high-elevation systems. The use of ecological network analysis has become increasingly important in advancing our understanding of host–parasite systems, particularly in bat–ectoparasite relationships (Runghen et al., 2021; Biz et al., 2023; Zapata-Mesa et al., 2024). Unlike traditional species inventories, network approaches allow researchers to explore structural properties such as modularity, nestedness and interaction specialization within ecological communities (Bezerra and Bocchiglieri, 2022). Bats and their ectoparasitic flies represent an ideal model for such studies due to their high species richness and long coevolutionary history. Network-based analyses have consistently revealed high levels of specialization and modularity in bat–fly associations, suggesting that both ecological and evolutionary factors shape these interaction patterns (Falcão et al., 2022). Moreover, understanding the structure of host–parasite networks is increasingly relevant for public health, as ectoparasites may act as vectors of zoonotic pathogens, potentially facilitating transmission between wildlife and humans (Szentiványi et al., 2019). Integrating network analysis into parasite–host studies therefore offers valuable insights for both ecological theory and applied conservation and health strategies. Given this context and the need to generate local-scale data that contribute to a broader understanding of bat–ectoparasite interactions, the aim of our study was to assess the dipteran ectoparasites associated with bats in the montane forests of Amazonas, northern Peru. Specifically, we sought to examine patterns of species distribution to explore the structure of parasite–host association using interaction networks. Materials and methods Study area The study was conducted in the hamlet of Nueva Esperanza on the Numparket Waterfall Tourist Route located in Aramango, Bagua (Amazonas, Peru). The area is characterized by low montane forest vegetation and falls within the ecoregions of Very Humid Montane Forest (Bosque Montano Muy Húmedo, BMHM) and Very Humid Premontane Forest (Bosque Premontano Muy Húmedo, BMHP) (Britto, 2017). The landscape includes both well-preserved primary forest and zones subjected to selective logging. Fieldwork was conducted between July 2023, February 2024 and August 2024 across three sites: Numparket (1800 m a.s.l.), Chontas (1560 m a.s.l.) and Higuerón (1480 m a.s.l.). Numparket is located around the Numparket Waterfall within the conservation concession Cerro El Adobe. This area also forms part of the buffer zone of both the national sanctuary Cordillera de Colán and the communal reserve Chayu Nain (Figure 1). Due to its proximity to the waterfall and its tributary rivers, Numparket maintains high humidity throughout most of the year. The site is predominantly covered by well-preserved primary forest, with the exception of some disturbed zones near the road. In contrast, Chontas and Higuerón are located within the area of influence of Cordillera de Colán. These sites are characterized by a mosaic of preserved forest, patches of secondary growth and areas affected by selective logging. The vegetation includes species such as Ficus paraensis, Cecropia spp., various species of Araceae and Rubiaceae, and abundant pteridophytes (Authors’ observation).Figure 1.Geographic location of the sampling sites in the Amazonas region, northern Peru, where bats were captured and their ectoparasitic flies collected during the 2023–2024 field campaign.Figure 1 long description.A composite image showing maps and photos of sampling sites in northern Peru, including Higuerón, Chontas and Numparket waterfall. See long description. Geographic location of the sampling sites in the Amazonas region, northern Peru, where bats were captured and their ectoparasitic flies collected during the 2023–2024 field campaign. Figure 1 long description. Bats and flies sampling At every sampling station, 10 understory mist nets (12 × 2.5 m) spaced ∼20 m apart were used for 12 nights, distributed as three blocks of four consecutive nights (July 2023, February 2024 and August 2024) (MINAM, 2015). The nets were opened from 18:00 to 00:00 h to target species with peak foraging activity during that period (Jones et al., 1996). Individuals that could not be identified in the field were collected, preserved in alcohol and deposited at the Museo Vera Alleman de la Universidad Ricardo Palma (MURP) in Lima, Peru. All individuals were checked while alive. Bat flies were removed using entomological forceps, then fixed and preserved in polypropylene cryovials containing 70% ethanol. Specimens were cleared in 10% KOH and examined under a Nikon SMZ745 stereomicroscope for taxonomic identification using the keys of Wenzel et al. (1966); Wenzel (1976) and Guerrero (1993; 1994a, 1994b, 1995a, 1995b, 1996b, 1996c). Macrophotographs of external anatomy and taxonomically important structures were taken using a TOUPCAM camera mounted on a Nikon Eclipse Si microscope with a Nikon Nii LED illumination system. Image stacking was performed with ToupView software. All ectoparasite specimens were deposited in the entomological collection of the Natural History Museum, Faculty of Natural Sciences and Mathematics, Universidad Nacional Federico Villarreal (MUFV). Specialization of bat-fly interactions Host specificity of bat flies was classified as follows: monoxenous (ectoparasitic flies utilizing only a single host species), oligoxenous (utilizing two or more congeneric species), pleioxenous (utilizing two or more host genera within the same family) and polyxenous (utilizing multiple hosts from different families) (Marshall, 1982; Seneviratne et al., 2009). The parasite population component was analysed using standard ecological parasitological indices: prevalence (P%) and mean intensity of infection (MI), following Bush et al. (1997) and Bautista-Hernández et al. (2015). Using the bat–fly interaction encounters, we constructed weighted bipartite networks for each of the three sites as well as an aggregated regional network. To assess the specificity of bat–fly interactions in these networks, we applied modularity and specialization metrics. We first tested for modularity (Qw) using the weighted DIRTLPAwb + algorithm (Beckett, 2016) and then evaluated low-level nestedness (within modules) using the WNODAsm metric (Pinheiro et al., 2019). Following Pinheiro et al. (2022), we did not test for nestedness in the overall network, as all networks were significantly modular (see Results). Modularity measures the extent to which species and their interactions can be divided into subgroups (modules) that are more interconnected within themselves than with others (Newman, 2006). Nestedness reflects the pattern in which interactions of species with fewer connections (specialists) form subsets of the interactions of species with more connections (generalists) (Mariani et al., 2019). Additionally, specialization was quantified using the H2’ metric (complementary specialization), which captures how selective the network is beyond what would be expected based on species relative abundances, approximated by the matrix’s marginal totals (Blüthgen et al., 2006). To test the significance of these metrics, we used the equiprobable (preserving species richness and total number of interactions) and proportional (same as equiprobable but also preserving marginal sums) null models described in Pinheiro et al. (2022). Specifically, the restricted version of the equiprobable null model, which also maintains the modular structure during randomizations, was used to test WNODAsm, while the proportional null model was employed for Qw and H2’. Species-level metrics were used to explore variation in the specialization of bat and fly species across sites, focusing only on species present at all sites. Species-level specialization (d’) was employed to describe how selectively a bat or fly interacts with available species from the opposite group within the network, based on the frequency of their interactions (Blüthgen et al., 2006). For flies, higher d′ values indicate higher host specificity. For bats, which do not choose their parasites, higher d′ values indicate that their assemblage of flies is composed of more host–specific parasites, whereas lower values indicate association with more generalist parasites. Additionally, species strength was calculated as the total sum of interaction proportions across all partners for a given species, reflecting how dependent bats or flies are on that species (Bascompte et al., 2006). These species-level metrics, along with the network-level metrics mentioned above, were calculated using the package ‘bipartite’ (Dormann et al., 2008) in the software R 4.4.1. Sampling coverage of networks was also analysed following the suggestion of Chiu et al. (2023). This metric indicates the proportion of the total number of interaction events represented by the detected interactions. For this assessment, the ‘iNext.link’ package (Hsieh et al., 2016) was used in the software R 4.4.1. Finally, to quantify differences between interaction networks among sites, we followed the approach of Fründ (2021), which decomposes total link dissimilarity into additive components. For each pair of sites, we calculated: βWN, the overall dissimilarity between the two interaction networks; βOS, the dissimilarity attributable to changes in interactions among species shared between sites (rewiring); βST, the dissimilarity attributable to species turnover, i.e. interactions that differ because one or both interacting species are present at only one site. Following Novotny (2009), βST was further partitioned into turnover caused exclusively by flies (βST.f), exclusively by bats (βST.b), or jointly by both (βST.fb). This analysis was performed using the betalinkr_multi function of the package ‘bipartite’ (Dormann et al., 2008) in the software R 4.4.1., specifically with the ‘commondenom’ partition method (Fründ, 2021). Results Bats and flies A total of 160 bats were captured, including 152 individuals from the family Phyllostomidae and 8 from Vespertilionidae. The bats belonged to 23 species, of which 71 individuals from 14 species were parasitized by bat flies (Table 1). The most abundant bat species were Carollia brevicauda (n = 48), C. perspicillata (n = 36) and Sturnira oporaphilum (n = 19). The species with the highest number of parasitized individuals were C. brevicauda (n = 30) and C. perspicillata (n = 18). Among the three sampling areas, Numparket had the highest number of captured bats (n = 73) and parasitized individuals (n = 33) (Table 1). C. brevicauda presented the highest abundance of ectoparasites across all three sites. In Numparket and Higuerón, C. perspicillata ranked second in parasite abundance, while in Chontas, the second most parasitized species was Myotis nigricans. Table 1.Bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru, and specific characterization based on their ectoparasitic fliesTable 1 long description.Across three elevations, Numparket had the most bats examined (73) but the lowest mean fly intensity (1.7), while Chontas had the highest intensity (2.8). See long description.HiguerónChontasNumparket(1480 m)(1560 m)(1800 m)Chiroptera speciesE (P)BfP%MIE (P)BfP%MIE (P)BfP%MIPyllostomidaeAnoura aequatoris2 (0)00–––––4 (2)2501Anoura caudifer1 (0)00–––––––––Anoura peruana––––––––1 (1)61006Artibeus glaucus––––1 (1)210024 (0)00–Artibeus planirostris1 (1)610061 (0)00–––––Carollia brevicauda16 (11)2068.71.819 (10)2053213 (9)1969.22.1Carollia perspicillata6 (4)966.62.28 (2)625322 (12)2054.51.6Choeroniscus minor––––––––1 (1)11001Desmodus rotundus3 (0)00–3 (0)00–2 (0)00–Platyrrhinus fusciventris––––––––1 (1)11001Platyrrhinus umbratus––––1 (0)00–1 (0)00–Sturnira bidens1 (0)00–2 (0)00–3 (1)133.31Sturnira erythromos1 (0)00–2 (0)00–3 (0)00–Sturnira magna––––––––2 (0)00–Sturnira oporaphilum2 (0)00–5 (2)640312 (4)433.31Sturnira tildae––––4 (2)6503––––Vampyressa thyone1 (0)00–––––––––Vampyrodes caraccioli1 (1)31003––––2 (1)1501VespertilionidaeEptesicus chiriquinus1 (0)00–––––––––Myotis caucensis1 (0)00–––––––––Myotis nigricans––––2 (2)91004.51 (0)00–Myotis riparius1 (1)510051 (1)710071 (1)11001Total38 (18)4346.12.349 (20)5640.82.873 (33)5645.21.7Bf, bat fly abundance; E(P), no. of examined bats (parasitized); MI, mean intensity; P%, prevalence; S, bat fly species richness. Table 1.Bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru, and specific characterization based on their ectoparasitic fliesTable 1 long description.Across three elevations, Numparket had the most bats examined (73) but the lowest mean fly intensity (1.7), while Chontas had the highest intensity (2.8). See long description.HiguerónChontasNumparket(1480 m)(1560 m)(1800 m)Chiroptera speciesE (P)BfP%MIE (P)BfP%MIE (P)BfP%MIPyllostomidaeAnoura aequatoris2 (0)00–––––4 (2)2501Anoura caudifer1 (0)00–––––––––Anoura peruana––––––––1 (1)61006Artibeus glaucus––––1 (1)210024 (0)00–Artibeus planirostris1 (1)610061 (0)00–––––Carollia brevicauda16 (11)2068.71.819 (10)2053213 (9)1969.22.1Carollia perspicillata6 (4)966.62.28 (2)625322 (12)2054.51.6Choeroniscus minor––––––––1 (1)11001Desmodus rotundus3 (0)00–3 (0)00–2 (0)00–Platyrrhinus fusciventris––––––––1 (1)11001Platyrrhinus umbratus––––1 (0)00–1 (0)00–Sturnira bidens1 (0)00–2 (0)00–3 (1)133.31Sturnira erythromos1 (0)00–2 (0)00–3 (0)00–Sturnira magna––––––––2 (0)00–Sturnira oporaphilum2 (0)00–5 (2)640312 (4)433.31Sturnira tildae––––4 (2)6503––––Vampyressa thyone1 (0)00–––––––––Vampyrodes caraccioli1 (1)31003––––2 (1)1501VespertilionidaeEptesicus chiriquinus1 (0)00–––––––––Myotis caucensis1 (0)00–––––––––Myotis nigricans––––2 (2)91004.51 (0)00–Myotis riparius1 (1)510051 (1)710071 (1)11001Total38 (18)4346.12.349 (20)5640.82.873 (33)5645.21.7Bf, bat fly abundance; E(P), no. of examined bats (parasitized); MI, mean intensity; P%, prevalence; S, bat fly species richness. Bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru, and specific characterization based on their ectoparasitic flies Table 1 long description. Higuerón Chontas Numparket (1480 m) (1560 m) (1800 m) Chiroptera species E (P) Bf P% MI E (P) Bf P% MI E (P) Bf P% MI Pyllostomidae Anoura aequatoris 2 (0) 0 0 – – – – – 4 (2) 2 50 1 Anoura caudifer 1 (0) 0 0 – – – – – – – – – Anoura peruana – – – – – – – – 1 (1) 6 100 6 Artibeus glaucus – – – – 1 (1) 2 100 2 4 (0) 0 0 – Artibeus planirostris 1 (1) 6 100 6 1 (0) 0 0 – – – – – Carollia brevicauda 16 (11) 20 68.7 1.8 19 (10) 20 53 2 13 (9) 19 69.2 2.1 Carollia perspicillata 6 (4) 9 66.6 2.2 8 (2) 6 25 3 22 (12) 20 54.5 1.6 Choeroniscus minor – – – – – – – – 1 (1) 1 100 1 Desmodus rotundus 3 (0) 0 0 – 3 (0) 0 0 – 2 (0) 0 0 – Platyrrhinus fusciventris – – – – – – – – 1 (1) 1 100 1 Platyrrhinus umbratus – – – – 1 (0) 0 0 – 1 (0) 0 0 – Sturnira bidens 1 (0) 0 0 – 2 (0) 0 0 – 3 (1) 1 33.3 1 Sturnira erythromos 1 (0) 0 0 – 2 (0) 0 0 – 3 (0) 0 0 – Sturnira magna – – – – – – – – 2 (0) 0 0 – Sturnira oporaphilum 2 (0) 0 0 – 5 (2) 6 40 3 12 (4) 4 33.3 1 Sturnira tildae – – – – 4 (2) 6 50 3 – – – – Vampyressa thyone 1 (0) 0 0 – – – – – – – – – Vampyrodes caraccioli 1 (1) 3 100 3 – – – – 2 (1) 1 50 1 Vespertilionidae Eptesicus chiriquinus 1 (0) 0 0 – – – – – – – – – Myotis caucensis 1 (0) 0 0 – – – – – – – – – Myotis nigricans – – – – 2 (2) 9 100 4.5 1 (0) 0 0 – Myotis riparius 1 (1) 5 100 5 1 (1) 7 100 7 1 (1) 1 100 1 Total 38 (18) 43 46.1 2.3 49 (20) 56 40.8 2.8 73 (33) 56 45.2 1.7 Bf, bat fly abundance; E(P), no. of examined bats (parasitized); MI, mean intensity; P%, prevalence; S, bat fly species richness. A total of 155 ectoparasitic flies were collected, representing 19 species from the families Streblidae (17 species) and Nycteribiidae (2 species) (Figures 2 –3). The most abundant fly species were Paraeuctenodes similis Wenzel, 1976 (n = 44), Trichobius joblingi Wenzel, 1966 (n = 43) and Megistopoda proxima (Séguy, 1926) (n = 13). Numparket and Higuerón exhibited the highest bat fly species richness (s = 11) and abundance (n = 56), followed by Chontas (s = 8, n = 43). Only P. similis and T. joblingi were recorded in all three areas. In terms of host specificity, most bat fly species were classified as monoxenous (n = 10), followed by oligoxenous (n = 8), and pleioxenous (n = 1) (Table 2).Figure 2.Species of ectoparasitic diptera from bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru (first part). (A) Anastrebla caudiferae, (B) Aspidoptera falcata, (C) Exastinion oculatum, (D) Megistopoda proxima, (E) Metelasmus pseudopterus, (F) Neotrichobius bisetosus, (G) Paraeuctenodes similis, (H) Paratrichobius longicrus complex, (I) Strebla guajiro.Figure 2 long description.Composite micrograph with nine sub-images labeled A) to I) of amber-stained insect bodies with bristles on a white background. See long description.Figure 3.Species of ectoparasitic diptera from bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru (second part). (A) Anatrichobius scorzai, (B) Basilia anceps, (C) Anatrichobius sp., (D) Speiseria ambigua, (E) Trichobius joblingi, (F) Paratrichobius salvini complex.Figure 3 long description.Composite figure with six labeled insect micrographs (A–F) in orange-brown tones on a white background, each with a scale bar. See long description.Table 2.Associations and characterization of ectoparasitic flies from bat captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, PeruTable 2 long description.Across three sites, 11 ectoparasitic fly taxa were recorded at H and N (8 at C); Carollia bats hosted the highest counts, especially Paraeuctenodes similis and Trichobius joblingi. See long description.Diptera speciesChiroptera speciesHCNSpeNycterebiidaeBasilia anceps*OliMyotis nigricans–4–Myotis riparius–5–Basilia sp. (group juquiensis)MonMyotis riparius3––StreblidaeAnastrebla caudiferae**OliCarollia perspicillata––1Carollia brevicauda1––Anastrebla sp.MonChoeroniscus minor––1Anatrichobius scorzai*OliMyotis nigricans–5–Myotis riparius22–Anatrichobius sp.MonMyotis riparius––1Aspidoptera falcata*MonSturnira tildae–4–Aspidoptera phyllostomatis*MonArtibeus planirostris––5Exastinion deceptivum**OliAnoura aequatoris––2Anoura peruana––1Exastinion oculatum**MonAnoura peruana––5Megistopoda proxima*OliSturnira bidens––1Sturnira oporaphilum–64Sturnira tildae–2–Metelasmus pseudopterus*MonArtibeus planirostris1––Neotrichobius bisetosus*PleArtibeus glaucus–2–Vampyrodes caraccioli2––Paraeuctenodes similis**OliCarollia brevicauda9813Carollia perspicillata158Paratrichobius longicrus complex*MonPlatyrrhinus fusciventris––1Speiseria ambigua*MonCarollia perspicillata11–Strebla guajiro*OliCarollia brevicauda1––Carollia perspicillata1–1Trichobius joblingi*OliCarollia brevicauda9126Carollia perspicillata6–10Paratrichobius salvini complex**MonVampyrodes caraccioli1–1Abundance435656Richness11811C, Chontas (1560 m); H, Higuerón (1480 m); N, Numparket (1800 m); P%, Prevalence; Spe, Specificity (Mon, Monoxenous; Oli, Oligoxenous; Ple, Pleioxenous).**New species record for Peru; *New record only for Amazonas. Species of ectoparasitic diptera from bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru (first part). (A) Anastrebla caudiferae, (B) Aspidoptera falcata, (C) Exastinion oculatum, (D) Megistopoda proxima, (E) Metelasmus pseudopterus, (F) Neotrichobius bisetosus, (G) Paraeuctenodes similis, (H) Paratrichobius longicrus complex, (I) Strebla guajiro. Figure 2 long description. Species of ectoparasitic diptera from bats captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru (second part). (A) Anatrichobius scorzai, (B) Basilia anceps, (C) Anatrichobius sp., (D) Speiseria ambigua, (E) Trichobius joblingi, (F) Paratrichobius salvini complex. Figure 3 long description. Table 2.Associations and characterization of ectoparasitic flies from bat captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, PeruTable 2 long description.Across three sites, 11 ectoparasitic fly taxa were recorded at H and N (8 at C); Carollia bats hosted the highest counts, especially Paraeuctenodes similis and Trichobius joblingi. See long description.Diptera speciesChiroptera speciesHCNSpeNycterebiidaeBasilia anceps*OliMyotis nigricans–4–Myotis riparius–5–Basilia sp. (group juquiensis)MonMyotis riparius3––StreblidaeAnastrebla caudiferae**OliCarollia perspicillata––1Carollia brevicauda1––Anastrebla sp.MonChoeroniscus minor––1Anatrichobius scorzai*OliMyotis nigricans–5–Myotis riparius22–Anatrichobius sp.MonMyotis riparius––1Aspidoptera falcata*MonSturnira tildae–4–Aspidoptera phyllostomatis*MonArtibeus planirostris––5Exastinion deceptivum**OliAnoura aequatoris––2Anoura peruana––1Exastinion oculatum**MonAnoura peruana––5Megistopoda proxima*OliSturnira bidens––1Sturnira oporaphilum–64Sturnira tildae–2–Metelasmus pseudopterus*MonArtibeus planirostris1––Neotrichobius bisetosus*PleArtibeus glaucus–2–Vampyrodes caraccioli2––Paraeuctenodes similis**OliCarollia brevicauda9813Carollia perspicillata158Paratrichobius longicrus complex*MonPlatyrrhinus fusciventris––1Speiseria ambigua*MonCarollia perspicillata11–Strebla guajiro*OliCarollia brevicauda1––Carollia perspicillata1–1Trichobius joblingi*OliCarollia brevicauda9126Carollia perspicillata6–10Paratrichobius salvini complex**MonVampyrodes caraccioli1–1Abundance435656Richness11811C, Chontas (1560 m); H, Higuerón (1480 m); N, Numparket (1800 m); P%, Prevalence; Spe, Specificity (Mon, Monoxenous; Oli, Oligoxenous; Ple, Pleioxenous).**New species record for Peru; *New record only for Amazonas. Associations and characterization of ectoparasitic flies from bat captured along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru Table 2 long description. Diptera species Chiroptera species H C N Spe Nycterebiidae Basilia anceps* Oli Myotis nigricans – 4 – Myotis riparius – 5 – Basilia sp. (group juquiensis) Mon Myotis riparius 3 – – Streblidae Anastrebla caudiferae** Oli Carollia perspicillata – – 1 Carollia brevicauda 1 – – Anastrebla sp. Mon Choeroniscus minor – – 1 Anatrichobius scorzai* Oli Myotis nigricans – 5 – Myotis riparius 2 2 – Anatrichobius sp. Mon Myotis riparius – – 1 Aspidoptera falcata* Mon Sturnira tildae – 4 – Aspidoptera phyllostomatis* Mon Artibeus planirostris – – 5 Exastinion deceptivum** Oli Anoura aequatoris – – 2 Anoura peruana – – 1 Exastinion oculatum** Mon Anoura peruana – – 5 Megistopoda proxima* Oli Sturnira bidens – – 1 Sturnira oporaphilum – 6 4 Sturnira tildae – 2 – Metelasmus pseudopterus* Mon Artibeus planirostris 1 – – Neotrichobius bisetosus* Ple Artibeus glaucus – 2 – Vampyrodes caraccioli 2 – – Paraeuctenodes similis** Oli Carollia brevicauda 9 8 13 Carollia perspicillata 1 5 8 Paratrichobius longicrus complex* Mon Platyrrhinus fusciventris – – 1 Speiseria ambigua* Mon Carollia perspicillata 1 1 – Strebla guajiro* Oli Carollia brevicauda 1 – – Carollia perspicillata 1 – 1 Trichobius joblingi* Oli Carollia brevicauda 9 12 6 Carollia perspicillata 6 – 10 Paratrichobius salvini complex** Mon Vampyrodes caraccioli 1 – 1 Abundance 43 56 56 Richness 11 8 11 C, Chontas (1560 m); H, Higuerón (1480 m); N, Numparket (1800 m); P%, Prevalence; Spe, Specificity (Mon, Monoxenous; Oli, Oligoxenous; Ple, Pleioxenous). New species record for Peru; *New record only for Amazonas. Specialization of bat–fly interactions Sampling coverage of bat–fly networks was always above 0.85 (Table 3), indicating that they are a good representation of bat–fly interactions at each site as well as at the regional scale. All networks exhibited a modular topology, but nestedness within modules was observed only in the regional network (Table 3). Specificity of interactions was intermediate to high across all networks, as indicated by Qw (≥0.48) and H2’ (≥0.75). The highest values for these metrics were observed in Chontas (middle elevation), while Higuerón and Numparket showed similar values, still reflecting high specificity. The modular structure showed a clear partition based on the phylogenetic relationships of bats, with modules never including unrelated taxa of bats (Figure 4). Anoura, Carollia, Myotis and Sturnira species were always grouped within the same module, except for Carollia at higher altitudes (Numparket). In this latter case, C. brevicauda and C. perspicillata formed their own module, although they still shared the same flies.Figure 4.Modular structure of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. The regional network is the result of the aggregation of interactions of the three other local networks. Interactions and species of the same module share specific colours and interactions between species of different modules are in grey.Figure 4 long description.A diagram showing modular structures of bat-fly interaction networks at different elevations in Peru. See long description.Abbreviations: Anoura aequatoris = Anaeq, Anoura peruana = Anper, Artibeus glaucus = Argla, Artibeus planirostris = Arpla, Carollia brevicauda = Cabre, Carollia perspicillata = Caper, Choeroniscus minor = Chmin, Platyrrhinus fusciventris = Plfus, Sturnira bidens = Stbid, Sturnira oporaphilum = Stopo, Sturnira tildae = Sttil, Vampyrodes caraccioli = Vacar, Myotis nigricans = Mynig, Myotis riparius = Myrip. Basilia anceps = Baanc, Basilia sp = Basp, Anastrebla caudiferae = Ancau, Anastrebla sp = Ansp, Anatrichobius scorzai = Ansco, Anatrichobius sp = Ansp, Aspidoptera falcata = Asfal, Aspidoptera phyllostomatis = Asphy, Exastinion deceptivum = Exdec, Exastinion oculatum = Exocu, Megistopoda proxima = Mepro, Metelasmus pseudopterus = Mepse, Neotrichobius bisetosus = Nebis, Paraeuctenodes similis = Pasim, Paratrichobius longicrus = Palon, Paratrichobius salvini = Pasal, Speiseria ambigua = Spamb, Strebla guajiro = Stgua, Trichobius joblingi = Trjob.Table 3.Sampling coverage and structural properties of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. The regional network is the result of the aggregation of interactions of the three local networksTable 3 long description.Chontas has the highest modularity (0.66) and specialization (H2’ 0.92), while Numparket shows the highest within-module nestedness (75) among the networks. See long description.NetworkSCQwWNODAsmH2’Regional0.960.5960.740.75Higuerón (1480 m)0.860.51700.75Chontas (1560 m)0.980.6631.250.92Numparket (1800 m)0.880.48750.65SC, sampling coverage; Qw, weighted modularity; WNODAsm, within-module nestedness, H2’: complementary specialization. Statistically significant values (P < 0.05) are in bold. Modular structure of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. The regional network is the result of the aggregation of interactions of the three other local networks. Interactions and species of the same module share specific colours and interactions between species of different modules are in grey. Figure 4 long description. Table 3.Sampling coverage and structural properties of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. The regional network is the result of the aggregation of interactions of the three local networksTable 3 long description.Chontas has the highest modularity (0.66) and specialization (H2’ 0.92), while Numparket shows the highest within-module nestedness (75) among the networks. See long description.NetworkSCQwWNODAsmH2’Regional0.960.5960.740.75Higuerón (1480 m)0.860.51700.75Chontas (1560 m)0.980.6631.250.92Numparket (1800 m)0.880.48750.65SC, sampling coverage; Qw, weighted modularity; WNODAsm, within-module nestedness, H2’: complementary specialization. Statistically significant values (P < 0.05) are in bold. Sampling coverage and structural properties of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. The regional network is the result of the aggregation of interactions of the three local networks Table 3 long description. Network SC Qw WNODAsm H2’ Regional 0.96 0.59 60.74 0.75 Higuerón (1480 m) 0.86 0.51 70 0.75 Chontas (1560 m) 0.98 0.66 31.25 0.92 Numparket (1800 m) 0.88 0.48 75 0.65 SC, sampling coverage; Qw, weighted modularity; WNODAsm, within-module nestedness, H2’: complementary specialization. Statistically significant values (P < 0.05) are in bold. Species showed different patterns across sites in terms of species-level specialization and species strength (Figure 5, Supplementary table S1-S2). Carollia species exhibited higher specialization at middle elevations (Chontas), with C. brevicauda consistently showing higher values than C. perspicillata. Myotis riparius displayed the highest specialization values at lower (Higuerón) and higher (Numparket) elevations. However, species strength for C. brevicauda decreased continuously from lower to higher elevations, while values for C. perspicillata increased at higher elevations, eventually surpassing C. brevicauda. Myotis riparius showed the same pattern observed in specialization, with lower species strength values at middle elevations. Among flies, T. joblingi was more specialized than P. similis at lower and middle elevations, but roles reversed at higher elevations. However, T. joblingi had higher species strength values than P. similis at the lower-elevation site, while the opposite occurred at middle- and higher-elevation sites.Figure 5.Species-level metrics of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. Species-level specialization d’ (A and B) and species strength (C and D). Only species present in the three evaluated sites are assessed.Figure 5 long description.Figure: four line plots of d specialization and species strength across three sites, comparing bats vs flies and changes from Higuerón to Chontas to Numparket. See long description. Species-level metrics of bat–fly interaction networks along the Nueva Esperanza Trail to Numparket Falls, Amazonas, Peru. Species-level specialization d’ (A and B) and species strength (C and D). Only species present in the three evaluated sites are assessed. Figure 5 long description. Network dissimilarity between elevations was moderate to high, with βWN values ranging from 0.60 (Higuerón–Chontas) to 0.70 (Chontas–Numparket) (Table 4). Across all pairwise comparisons, the contribution of rewiring among shared species (βOS) was consistently low (0.03–0.04), representing only 5–7% of total link dissimilarity. In contrast, species turnover (βST) accounted for the vast majority of network differences (93–95%). Within the turnover component, co-turnover of both bats and flies (βST.fb) was consistently the dominant factor, contributing approximately 50–60% of βST in all comparisons. Turnover restricted to flies (βST.f) also made a substantial contribution (33–40%), whereas turnover restricted to bats (βST.b) was smaller (0–21%) and, in one comparison, absent.Table 4.Pairwise network dissimilarity between sites. ΒOS represents the dissimilarity attributable to rewiring among species shared between sites. ΒWN is the overall dissimilarity between interaction networks. ΒST corresponds to the component of dissimilarity explained by species turnover. ΒST.F, βST.B and βST.Fb indicate the portions of βST attributable to turnover restricted to fly species, restricted to bat species, and jointly to both trophic levels, respectivelyTable 4 long description.Overall network dissimilarity is highest for Chontas–Numparket (βWN 0.70) and lowest for Higueron–Numparket (0.57), with rewiring constant and small (βOS 0.04) across pairs. See long description.NetworksβOSβWNβSTβST.fβST.bβST.fbHigueron – Chontas0.040.60.560.160.120.28Higueron – Numparket0.040.570.540.2100.32Chontas – Numparket0.040.70.670.220.070.37 Table 4.Pairwise network dissimilarity between sites. ΒOS represents the dissimilarity attributable to rewiring among species shared between sites. ΒWN is the overall dissimilarity between interaction networks. ΒST corresponds to the component of dissimilarity explained by species turnover. ΒST.F, βST.B and βST.Fb indicate the portions of βST attributable to turnover restricted to fly species, restricted to bat species, and jointly to both trophic levels, respectivelyTable 4 long description.Overall network dissimilarity is highest for Chontas–Numparket (βWN 0.70) and lowest for Higueron–Numparket (0.57), with rewiring constant and small (βOS 0.04) across pairs. See long description.NetworksβOSβWNβSTβST.fβST.bβST.fbHigueron – Chontas0.040.60.560.160.120.28Higueron – Numparket0.040.570.540.2100.32Chontas – Numparket0.040.70.670.220.070.37 Pairwise network dissimilarity between sites. ΒOS represents the dissimilarity attributable to rewiring among species shared between sites. ΒWN is the overall dissimilarity between interaction networks. ΒST corresponds to the component of dissimilarity explained by species turnover. ΒST.F, βST.B and βST.Fb indicate the portions of βST attributable to turnover restricted to fly species, restricted to bat species, and jointly to both trophic levels, respectively Table 4 long description. Networks βOS βWN βST βST.f βST.b βST.fb Higueron – Chontas 0.04 0.6 0.56 0.16 0.12 0.28 Higueron – Numparket 0.04 0.57 0.54 0.21 0 0.32 Chontas – Numparket 0.04 0.7 0.67 0.22 0.07 0.37 Discussion Bats and flies This study expands the current knowledge of bat–ectoparasite interactions in the montane forests of northern Peru by documenting 19 species of parasitic flies associated with 14 bat species. All bat fly species reported represent new records for Amazonas (Minaya et al., 2021), and Anastrebla caudiferae, Exastinion deceptivum, Exastinion oculatum, P. similis and the Paratrichobius salvini complex are documented for the first time in Peru. An important observation concerns the record of T. joblingi. Although this species was previously reported from Condorcanqui, Amazonas, by Ibáñez and Jara (2008), we noted inconsistencies between their figure and the diagnostic characters of T. joblingi. Based on our specimens, we provide the first confirmed record of T. joblingi for the department of Amazonas. Two bat fly species, P. similis and T. joblingi, represented the dominant and most abundant core ectoparasites in the bats sampled in Amazonas. Both species parasitized more than 50% of the C. brevicauda and C. perspicillata individuals examined. A similar pattern was observed in the Magdalena River basin (López-Rivera et al., 2024) and in Caldas (Raigosa et al., 2020), both in Colombia, where approximately 50% of individuals were primarily parasitized by T. joblingi and P. similis, consistent with our observations in Amazonas. These findings suggest that both dipteran species exhibit strong host specificity toward Carollia bats, maintaining a stable host–parasite association across distinct Neotropical ecosystems. This stability may be further reinforced by the high sociality and frequent sharing of roosts and foraging resources among Carollia bats, which facilitate parasite transmission (Altizer et al., 2003; McLellan and Koopman, 2008; Rifkin et al., 2012; Webber et al., 2015; Medina and Torres, 2018). Among other streblid flies recorded, M. proxima and N. bisetosus stand out for exhibiting the broadest host associations, though for distinct biological reasons. M. proxima was the only species parasitizing more than one congeneric host within Sturnira, whereas N. bisetosus exploited two hosts from different genera – A. glaucus and V. caraccioli. Under classical host-specificity categories (Seneviratne et al., 2009), M. proxima qualifies as oligoxenous and N. bisetosus as pleioxenous, the latter being an uncommon pattern in Streblidae, a group known for strong phylogenetic fidelity (Dick and Patterson, 2006; Autino et al., 2011). These broader host associations likely reflect ecological opportunities for cross–host transmission. Although direct evidence for multispecies roost sharing among Sturnira in Andean forests is limited, phyllostomid bats commonly use diverse natural shelters, where mixed-species roosts can occur (Kunz and Lumsden, 2003; Patterson et al., 2007). Such conditions plausibly increase contact opportunities among sympatric hosts and may facilitate the movement of M. proxima among closely related Sturnira species. Similarly, rare pleioxenous patterns like that of N. bisetosus have been reported in other Neotropical systems (e.g. Neotrichobius delicatus in Loreto, Peru; Autino et al., 2011), typically involving flies associated with ecologically overlapping phyllostomid bats (Fagundes et al., 2017). Host–parasite associations between bat flies and their chiropteran hosts are generally characterized by strong specificity, as seen in Basilia and Anatrichobius, which primarily parasitize Myotis species (Guerrero, 1995b; Ospina-Pérez et al., 2023), or Exastinion, apparently restricted to Anoura (Guerrero, 1995b). Similar patterns have been documented in Peru (Minaya et al., 2021), and our findings corroborate these associations while extending their known geographic distributions into the montane forests of northern Peru. Specialization of bat–fly interactions This study indicates a great specialization at the community level among parasitic bat–flies and their hosts in the montane forests near Nueva Esperanza, Amazonas. These specific associations are well documented in different parts of the world (Lim et al., 2020; Poon et al., 2023); the Neotropical region (Guerrero, 2019; Ospina-Pérez et al., 2023; Ramírez-Martínez and Tlapaya-Romero, 2023; França et al., 2024) and specifically Peru are no exception (Autino et al., 2011). Bat–fly interaction networks showed high specificity (high Qw and H2’), a pattern frequently observed in bat–fly interactions at other locations (Fagundes et al., 2017; Urbieta et al., 2022; Hiller et al., 2021; Ramalho et al., 2021; Ospina-Pérez et al., 2023; Ramírez-Martínez and Tlapaya-Romero, 2023). This high specificity is mainly driven by the parasitic nature of these interactions, where parasites typically depend strongly on specific hosts to maximize their fitness (Runghen et al., 2021). Such a high degree of dependency often results in parasite–host networks forming modules composed of phylogenetically related species (Felix et al., 2022), as was also observed in all our bat–fly networks. The regional network showed internally nested modules due to the aggregation of interactions that were uniquely observed at specific sites. For example, this pattern is evident in the module of Myotis species: at Higuerón, only M. riparius is present, interacting with A. scorzai and Basilia sp.; at Chontas, M. nigricans appears along with B. anceps; and at Numparket, Anatrichobius sp. is present. When aggregating all these interactions, nestedness increases within the Myotis module, and a similar pattern occurs in other modules, resulting in a compound structure (internally nested modules). This is consistent with the integrative hypothesis of specialization proposed for parasitic networks, which suggests that at larger scales, internally nested modules are more likely to emerge due to the aggregation of different allopatric species and interactions (Felix et al., 2022). In addition to the evolutionary component behind interactions between bats and flies, these associations have also been particularly discussed in the context of roost-sharing among bat species (Reckardt and Kerth, 2006; Patterson et al., 2007; Fagundes et al., 2017; Urbieta et al., 2022). Logically, species that share roosts are more susceptible to sharing flies, as mentioned in the previous section. This could also be a factor driving the interactions observed in this study, although specific information on roost-sharing is not available for most species. Various bat species share roost with congeneric species, especially in caves (Tanalgo et al., 2022), which could have contributed to the independent module aggregation observed in our study for Anoura, Carollia, Myotis and Sturnira, which have been reported to roost in caves frequently (Tanalgo et al., 2022). The Stenodermatini species recorded in our study apparently prefer different kinds of roosts (Garbino and Tavares, 2018), and as far as we know, there are no records of them roosting together. However, A. glaucus and V. caraccioli were both hosts of P. salvini complex. Artibeus glaucus is a strictly tent-making bat (Ortega et al., 2015), and V. caraccioli is also suggested to be a tent-making bat (Page and Dechmann, 2022). This may suggest they could potentially share roosts; however, tents are usually inhabited by only one species (Rodríguez-Herrera et al., 2007). Nevertheless, considering that bats can colonize a tent previously used (Rodríguez-Herrera et al., 2007) or may possibly try to exclude bats from an existent tent (Kunz and McCracken, 1996), there is a possibility that flies can be transmitted through tents. This could represent a strategy by flies to spread among populations and species, taking advantage of the complex roosting dynamics of tent-making bats (Chaverri and Kunz, 2006, 2010; Fernandez et al., 2021). In summary, roosting behaviour of bats may be closely related to their interactions with flies and should be explored in detail to better understand these relationships. Specialization and species strength of species varied across sites but followed different patterns, which may indicate that bat–fly relationships change according to specific properties of each site, even though the overall network structure can remain similar (Nielsen and Totland, 2014). The observed changes in specialization and species strength for both Carollia species, T. joblingi and P. similis, are consistent with the module separation observed for Carollia at Numparket. At this site, fly species associated with the genus Carollia were much more dependent on C. perspicillata (as indicated by a disproportionately high species strength), while the specialization of P. similis at this site surpassed that of T. joblingi. Although these metrics do not causally drive modularity, they highlight complementary information about how these species structure the network. This suggests that not only species turnover or richness differences can modify networks, but also that changes in how bat and fly species interact can be a driver of subtle network structural variations (Jordán et al., 2008; Fründ, 2021). Despite of the evident species-level variations, rewiring contributed only a small fraction of total dissimilarity in all pairwise comparisons, indicating that species occurring at multiple sites tended to retain similar partners. In contrast, species turnover accounted for more than 90% of link dissimilarity, with the largest contribution coming from the joint turnover of bats and flies, followed by turnover restricted to flies. This pattern reflects both the inherent specificity of flies (Runghen et al., 2021) and the considerable variation in bat and fly assemblages across elevations. These results show that differences in community composition were the main driver of the to the observed variation in interactions, although network structure remained broadly similar among sites as has been observed in other studies (e.g. Kemp et al., 2017; White et al., 2022). In conclusion, we provide novel insights into the diversity and structure of bat–fly interactions in the montane forests of northern Peru and represents the first in the country to apply a network-based approach to these associations. Our records reveal new distributional records at both local and national levels. Bat–fly relationships were highly specialized at both local and regional scales, with slight structural variation across sites. Network structure appears to be shaped by phylogenetic constraints and the roosting behaviour of bat hosts. Species turnover was the major factor behind interaction differences along the elevational gradient. However, species-level roles of bats and flies varied across sites, suggesting that specific interaction dynamics, rather than species turnover alone, contributed to the observed differences in interactions. This also points to a possible interplay between environmental factors and bat–fly relationships. Overall, our findings in this important but previously unexplored region of the Peruvian Andes contribute substantially to the broader ecological understanding of bat–fly interactions in Neotropical ecosystems. Supporting information Minaya et al. supplementary material Minaya et al. supplementary material