The Magdalena River forms the largest inter-Andean basin of Colombia covering 257,438 km² of national territory. It originates at the head of the Colombian Massif at an elevation of 3,865 m in the Puracé National Natural Park and flows into the Caribbean Sea in Bocas de Ceniza in the Department of Atlántico. This basin exhibits a great diversity of geological, edaphic, climatic, hydraulic, sedimentological, and morpho-dynamic conditions, forming a highly complex socio-ecological system (Gallo-Vélez et al. 2023). It crosses 1,540 km from south to north along 13 departments of Colombia, where 79% of the country’s population resides, making it an area of economic importance since 80% of Colombia’s GDP is generated there. The Magdalena River basin is significantly influenced by human activity, including deforestation, poor soil conservation and mining practices (Restrepo and Syvitski 2006).

Due to its geographical location, the climate of the region is tropical, primarily determined by altimetric variations, the relief topography and the influence of the Intertropical Confluence Zone, which generates two wet and two dry periods that occur interspersed throughout the year (León et al. 2000). Other factors that influence the climatic characteristics of the Magdalena River basin are precipitation, temperature, relative humidity, sunlight, and wind, which can create microclimates around the basin (IDEAM 2001; Nardini et al. 2020). According to Holdridge (1978), the Magdalena River basin encompasses five altitudinal zones: the tropical zone (0–1,000 m), the premontane zone (1,000–2,000 m), the submontane zone (2,000–3,000 m), and the Andean zone (3,000–4,000 m). The Magdalena River Basin supports ecosystems of Andean forests (26.36%), paramo (1.96%), xerophytic vegetation (3.01%), and wetlands (2.56%).

We conducted field trips in 12 localities in the Magdalena River basin between March, July and September 2021, April and November 2022, and January and March 2023. Specific dates for each locality are shown in Table 1. Four sampling sites were in the Department of Caldas at elevations between 170 and 650 m (Table 1). Five localities were in the Department of Cundinamarca, with elevations between 800 and 1,900 m. Three localities were within the Department of Cesar, with elevations between 50 and 200 m (Table 1; Fig. 1).

Figure 1. 

Locality records of bat flies (Nycteribiidae and Streblidae) in the inter-Andean Magdalena River basin, Colombia. Yellow circles indicate localities where field trips were conducted, blue circles are localities reported in the literature, and white circles indicate records of specimens in the ectoparasite collection of the Museo de Historia Natural de la Universidad de Caldas (MHN-UCa-Ec). The Magdalena River basin is indicated in green.

To capture bats, we installed 5 nylon mist nets (12.0 m × 2.5 m, and mesh size 36 mm) for five nights at each sampled location. Mist nets were randomly placed and operated between 18:00 and 22:00 hours. We placed bats individually in cotton bags and identified them using taxonomic keys (i.e., Gardner 2008; Díaz et al. 2021). To mitigate possible contamination between samples, the cloth bags used for bat sampling were cleaned and employed only once each night. We collected some specimens from the captured individuals to confirm identifications and deposited them in the Mammals (M) and Ectoparasites (Ec) collections of the Museo de Historia Natural, Universidad de Caldas (MHN-UCa), Colombia.

We manually collected bat flies using entomological tweezers. The collected bat flies were placed in Eppendorf tubes with 70% ethanol to prevent desiccation during transport to the laboratory. For the identification of Streblidae and Nycteribiidae specimens, we used the dichotomous keys and descriptions from Wenzel et al. (1966), Wenzel (1976), Guerrero (1994, 1995, 1998), Autino et al. (1999), and Graciolli (2004). All the collections were conducted under permits granted by the National Authority for Environmental Licenses (ANLA) to the Universidad de Caldas, as stipulated in Resolution 02497 of 2018, and to the National University of Colombia through Resolution 01435 of September 3, 2018, and to the Universidad de Caldas as stipulated in resolution 854 of 20 May 2019, modified by resolution 519 of 3 March 2022.

To compile additional records within the study area, we reviewed and identified bat and associated bat flies from different locations in the Magdalena River basin encompassing the departments of Caldas, Cundinamarca, Huila, Santander, and Tolima, deposited in the MHN-UCa-Ec collection. Additionally, we conducted searches for studies on ectoparasitic flies associated with bats in the Magdalena River Basin region. We reviewed the available information retrieved from search engines such as Science Direct, Web of Science, SciELO, Scopus, and Google Scholar, using the keywords ((fly*) OR (flies) AND (Streblidae*) OR (Nycteribiidae) AND (bat*) AND (Colombia*)). The last search was performed in October 2023. We also reviewed references and sources cited in the publications to obtain as much information as possible for creating interaction networks. We considered articles that included records of interactions between bat flies and bats in the Magdalena River Basin region, with no temporal restrictions. This approach enabled us to consolidate a more comprehensive data set for our study. We updated the taxonomic names of bat and bat fly’s species reported in the literature. In the case of the fly reported as Paratrichobius cf. longicrus by Tamsitt and Fox (1970) we listed these records as P. longicrus when constructing the interaction network. Similarly, we considered the records of the bats Sturnira lilium and S. parvidens as part of S. cf. giannae, while S. lilium is restricted to the South Cone in Argentina, Brazil, Bolivia, Paraguay, and Uruguay; and S. parvidens is restricted to Mexico, Guatemala, Belize, Honduras, El Salvador, Nicaragua and Costa Rica (Mammal Diversity Database, 2023).

We analyzed the coexistence of bat fly species and their hosts, using the Kendall correlation, which is suitable for small samples and allows for the observation of negative relationships. We excluded infracommunities reported in only one individual as analysis was not possible in such cases. The analyses were performed using the Bipartite package v. 2.20 (Dormann et al. 2008), and the network graphics were created using the “plotweb” function and the “plotModuleWeb” function from the same package, implemented in R software v. 4.3.2 (R Core Team 2022).

To construct the interaction network between bats and bat flies, we classified the associations as primary, non-primary, or accidental, following the criteria established by Dick (2007). Primary associations are defined as those host species infested by ≥ 5% of the total number of individuals of a species of parasite. Additionally, we reviewed the literature to check if associations that were below 5% had been reported previously and if so, we included them in the network. For the data obtained from the review of the MHN-UCa-Ec, we only considered the associations that have been previously reported in the literature. To evaluate host-ectoparasite interactions in the Magdalena River Basin, we unified the data obtained from the different sources mentioned earlier, constructing a two-dimensional matrix through the quantitative summation of the three data sets (fieldwork, literature, and collection specimens).

We also performed bipartite interaction networks, in which bat and ectoparasite species are represented by nodes, and interacting species are linked by lines, with line width proportional to the frequency of each interaction (Blüthgen et al. 2006). Additionally, we evaluated network properties such as complementary specialization (H₂’), specialization at the species level (d’), connectance (C), and modularity (M) (Dormann et al. 2009; Fortuna et al. 2010; Mello et al. 2016). The complementary specialization index (H₂’) measures both the degree of niche complementarity between species and specialization at the species level (Blüthgen et al. 2006; Blüthgen 2010). This index ranges from 0 (unspecialized network) to 1 (perfectly specialized network). The variation of species-level specialization measures (standardized Kullback-Leibler distance or divergence, d’) provides valuable information about the structural properties of a network. The C index represents the number of interactions or links observed in the network, between bats and their ectoparasitic flies considering the total number of potential interactions. It takes values from 0 to 1 where 0 indicates that there are no connections and 1 that denotes that most of the nodes in the network interact with each other (Blüthgen et al. 2006). We calculated Modularity (M) to identify subgroups of species that are more connected to each other than to the rest of the network (modules) (Fortuna et al. 2010). Modularity ranges from 0 to 1, with a value of 1 indicating a highly modular network and 0 a non-modular network. We use the DIRTLPAwd+ algorithm to compute modularity (Dormann and Strauss 2014; Beckett 2016). In addition, we used a null model to test the significance of specialization (H2’) and modularity (M) based on 1000 randomly generated matrices based on a Patefield null model (Dormann et al. 2009). Finally, we standardized modularity by calculating the ZQ score (ZQ), where values greater than 1.96 represent differences from the null model (Carstensen et al. 2016).

During the field work, we captured 376 bats belonging to 31 species, 22 genera, and four families. Of these, 285 bats of 25 species of Phyllostomidae and one species of Noctilionidae carried bat flies. In total we collected 588 bat flies belonging to 23 species, 10 genera and a single family (Streblidae). The most common bat species captured were Carollia perspicillata (n = 176), Carollia brevicauda (n = 38), Glossophaga soricina (n = 23), and Artibeus lituratus (n = 18). The most abundant species of bat flies were Trichobius joblingi (n = 301) and Speiseria ambigua (n = 50), mainly associated with species of the genus Carollia (Table 2). The literature review added records of 107 bats of eight species of the family Phyllostomidae, of which 51 were being parasitized by 170 bat flies belonging to 14 species of Streblidae. The most frequently reported bat species in the literature were Carollia perspicillata (49), Artibeus planirostris (20), and Desmodus rotundus (16). Similarly, the most abundant bat fly’s species recorded in the literature were T. joblingi (n = 114) and T. costalimai (n = 82), mainly associated with Carollia perspicillata and Phyllostomus discolor, respectively (Table 3). The review of specimens housed at the MHN-UCa-Ec added 145 bat flies belonging to 20 species, 11 genera, and two families (Streblidae and Nycteribiidae), linked to 67 bats of 19 species of Phyllostomidae, and two species of Vespertilionidae. Carollia perspicillata (n = 23) and T. joblingi (n = 48) were the most abundant bat and bat fly species (Table 3). For Nycteribiidae, all specimens recorded belong to the genus Basilia. The male specimens of Basilia deposited at the MHN-UCa-Ec could not be identified to the species level, since the majority of identification keys available correspond to females. We identified Basilia juquiensis males because they were collected with females (Fig. 2), and the latter are characterized by having the sternite almost twice as long as it is wide, sternite III covered by sternite II, and sternite VI divided longitudinally.

Figure 2. 

Micrographs of Basilia juquiensis (MHN-UCa-Ec 849), female (A, B) ventral view and (C) dorsal view; male (D) ventral (F) dorsal view. Scale bars: 0.5 mm.

The bat-fly bat interaction network for the Magdalena River Basin exhibited high specialization (H₂’ = 0.74) and low connectance (C = 0.06). Likewise, the specialization index by bat fly species indicating that 23.68% of species were highly specialized (d’ = 0.936–1). The results obtained from the reciprocal specialization index at the species level (d’) revealed that the species Anastrebla modestini, Exastinion clovisi, Paratrichobius sanchezi, Strebla alvarezi, S. christinae, Trichobius lonchophyllae, and T. parasiticus each had a value of 1, indicating high reciprocal specialization. These species were followed by Exastinion decepticum (d’ = 0.97) and Paradyschira parvuloides (d’ = 0.93). In contrast, the species with the lowest values in the specialization index were Strebla guajiro (d’ = 0.11) and S. consocius (d´ = 0.11), S. hertigi (d´ = 0.19), and Speiseria ambigua (d’ = 0.28), suggesting lower specialization compared to the aforementioned species (Suppl. material 1: table S1).

Of the 38 species of bat flies included in our interaction network, 19 were associated with a single bat species: A. mattadeni, A. modestini, B. fuquiensis, E. clovisi, P. parvuloides, Paraeuctenoides longipes, Paratrichobius sanchezi, S. alvarezi, S. christinae, S. consocius, S. tonatidae, Trichobioides perspicillatus, Trichobius affinis, T. dugesii, T. dugesioides, T. lonchophyllae, T. longipes, T. mendezi, and T. parasiticus. The bat fly species with the greatest number of interactions were T. joblingi (9) and Megistopoda proxima (6). The bat species with the most associations were C. perspicillata (8), A. lituratus (5), and Lophostoma nicaraguae, Phyllostomus discolor and P. hastatus (5). Additionally, 16 of the 37 bat species used to create the interaction network were associated with only one bat fly species (Suppl. material 1: fig S1). Moreover, we recorded high modularity for the interaction network (M = 0.64), forming 13 modules of related species. Most modules exhibited medium (the fly interacts with the host bat species with a noticeable, but not constant, frequency) to low (the fly interacts with the host bat species on rare occasions or under specific conditions) interaction strength (Fig. 3), except for the first module, which showed high interaction strength (the bat fly interacts with the host species frequently and regularly). This indicates a strong dependence of the fly on that particular bat for its survival and reproduction, specifically between C. perspicillata and T. joblingi. Most modules were formed by phylogenetically similar host, either from the same genus (modules 3 and 11 in Fig. 3) or from the same family (module 1 and 10 of Fig. 3). Comparisons with null models showed differences (ZQ = 83.51) indicating that the observed indices result from ecological processes rather than random chance. Furthermore, the comparison between the observed H₂’ value (0.74) and the null model values revealed significant differences (p-value = 0.00) (Suppl. material 1: fig. S2).

Figure 3. 

Modularity network generates in the bipartite bat-bat fly network in the inter-Andean Magdalena River basin of Colombia. The intensity of the blue box colors indicates the strength (intensity or frequency) of the interaction, due to the number of fly individuals involved.

We identified infracommunities associated with six bat species within Phyllostomidae: A. aequatorialis, C. brevicauda, C. perspicillata, P. discolor, P. hastatus, and S. giannae (Table 4). However, co-occurrence analyses of bat-bat fly species could be conducted for only five bat species (Table 4) due to the limited sample size of A. aequatorialis. The interaction between T. joblingi and S. ambigua on C. perspicillata showed negative interactions (z = -2.867, -0.311, p-value < 0.05), while the remaining five pairs showed positive or null density correlations (Table 4; Fig. 4). Notably, Carollia perspicillata was parasitized by two species of bat flies belonging to two different ecomorphological groups: wing crawlers and fur runners (Fig. 5).

Figure 4. 

Scatterplots of densities for all bat fly species pairs occurring on their respective host bat species along the inter-Andean Magdalena River basin of Colombia.

Figure 5. 

Carollia perspicillata parasitized by two species of bat flies belonging to two different ecomorphological groups Trichobius joblingi (wing crawler) and Speiseria ambigua (fur runner). Photograph: Carlos González Salazar.

The authors have declared that no competing interests exist.

No ethical statement was reported.

This work was supported by the program “Relación, distribución, taxonomía de especies de garrapatas asociadas a mamíferos silvestres en zonas endémicas de rickettsiosis en Colombia. Un acercamiento a la comprensión de la relación vectores patógenos-reservorios” (Code: 120385270267), specifically to the project “Mamíferos silvestres y su relación con rickettsias asociadas a garrapatas en dos zonas del Departamento de Cundinamarca: aproximación eco-epidemiológica y genómica” (Code: 71800), granted by the Ministerio De Ciencia, Tecnología e Innovación - Minciencias (CTO 80740- 200-2021).

CLR and HERC, conceptualization. CL-R, LNRS, ARH, JACV, JACS, JRC, JDV, FARP, PAOL, JJHO, ACG, EMOP, DMMM, MERP, MH, HERC revised the manuscript, contributed critically to the drafts, and approved the final version for publication. CLR, LNRS, ARH, JACS, JDV, JJHO, JRC, ACG, HERC carried out field trips. CLR, EMOP and HERC created and organized the figures of the manuscript. ARH, JACV, FARP, PAOL, MH, HERC, searched for funding of this project.

Camila López-Rivera https://orcid.org/0009-0003-5414-7956

Laura Natalia Robayo-Sánchez https://orcid.org/0000-0002-4074-1001

Alejandro Ramírez-Hernández https://orcid.org/0000-0001-8689-5930

Jerson Andrés Cuéllar-Saénz https://orcid.org/0000-0003-2257-8981

Juan Diego Villar https://orcid.org/0009-0006-8820-1908

Jesús Alfredo Cortés-Vecino https://orcid.org/0000-0003-2641-604X

Fredy Arvey Rivera-Páez https://orcid.org/0000-0001-8048-5818

Paula Andrea Ossa-López https://orcid.org/0000-0002-9079-4988

Erika M. Ospina-Pérez https://orcid.org/0000-0001-5784-6216

Jose J. Henao-Osorio https://orcid.org/0000-0002-8618-8539

Alexandra Cardona-Giraldo https://orcid.org/0000-0002-7534-994X

Javier Racero-Casarrubia https://orcid.org/0000-0001-5989-4174

Darwin M. Morales-Martinez https://orcid.org/0000-0001-5786-4107

Marylin Hidalgo https://orcid.org/0000-0001-8960-0616

Héctor E. Ramírez-Chaves https://orcid.org/0000-0002-2454-9482

All of the data that support the findings of this study are available in the main text or Supplementary Information.