Endemic bats (Mammalia: Chiroptera) of Colombia: State of knowledge, distribution, and conservation

Colombia, with 209 species, is one of the richest countries in terms of bat diversity. This high bat diversity is comprised in nine families and 72 genera. A total of eight species of the families Emballonuridae (n = 1) and Phyllostomidae (n = 7) are listed as endemic to the country. In spite the relevance of Colombia in bat diversity, little is known of these endemic species which are mostly known from the type locality (n= 4), whereas for others (n= 2), their taxonomic status is uncertain. Here, available information of endemic bats from Colombia is compiled, and new information on their distribution and conservation is provided. The most included species in publications including the original description (n = 15) was Lonchorhina marinkellei. The only distribution pattern observed for the species evaluated was for Carollia monohernandezi and Vampyressa sinchi that might overlap distributions in the eastern slopes of the Eastern Cordillera. Most endemic species are from the Andean and inter-Andean regions of the country (n = 6). At national level, there are no endemic species in any threatened category; however, at least one species (Lonchorhina marinkellei) is considered as Vulnerable (VU) and one (Saccopteryx antioquensis) as Endangered (EN) by the International Union for Conservation of Nature IUCN


Introduction
Species restricted to geographic units are recognized as endemic (Stattersfield et al., 1998), and areas of endemism are diagnosed by shared distributional boundaries of at least two endemic species (Morrone 1994). The characterization of endemic species is a priority, due to that their persistence along the time requires immediate conservation actions. Bat (Mammalia: Chiroptera) diversity in Colombia comprises almost 40 % of the total of the mammalian species recorded in the country (Ramírez-Chaves et al., 2016). Probably, bats are the most frequently studied mammalian group in Available information on bats of Colombia is scarce, and this is especially true for endemic species. The information on endemic bats of Colombia is disperse in the literature, and little attention has been paid to these species, despite their restricted distribution, and that almost nothing is known about their biology or systematics. For example, four endemic bats are known only from the type locality (C. monohernandezi, Lonchorhina mankomara, L. marinkellei, S. mistratensis: Phyllostomidae), and two of them (C. monohernandezi and S. mistratensis) are currently known only from the holotype (Contreras Vega & Cadena, 2000;Zurc & Velazco, 2010). In another case, some species are only known from less than five specimens (e.g. Vampyressa sinchi; Tavares et al., 2014). The aim of this manuscript is to present updated information of the state of knowledge, distribution, research priorities and conservation of the endemic bats of Colombia.

State of knowledge
Information from published literature is gathered since the first Colombian endemic bat species description in 1978 until 2018. In each study, information on the following categories used to establish the state of knowledge of endemic mammals of Colombia was searched: Publication type (Journal article, Conference proceedings, Book chapter, Book, Thesis, Conference paper), Relevance in the study (Main, Secondary), Anatomy and morphology, Taxonomy and Systematics, Biogeography/distribution and records, Threats and relationships with humans/Conservation (Threats, Conservation, Cultural aspects), Ecology and Natural History (Habitat use, Diet, Behavior, Abundance, Reproduction), Genetics, and Generalities (Compilation). The key words used for the searches included the names of each endemic species registered (Anoura carishina, Anoura cadenai, Carollia monohernandezi, Lonchorhina marinkellei, Lonchorhina mankomara, Saccopteryx antioquensis, Sturnira mistratensis, and Vampyressa sinchi) and "Colombia". When a paper covered different topics, each topic was considered separately.

Distribution
Specimens housed in Colombian museums and natural history collections were reviewed, including the Instituto de Investigaciones Biológicas Alexander von Humboldt, Villa de Leyva (IAvH), Instituto de Ciencias Naturales, Universidad Nacional de Colombia, Bogotá (ICN), Colección Teriológica, Universidad de Antioquia, Medellín (CTUA, previously known as MUA in reference to Javier Muñoz Arango, the former professor in charge of those collections at the Universidad de Antioquia), and Museo de Historia Natural, Universidad del Cauca, Popayán (MHNUC). Information from publications was also obtained. The area of occupancy (AOO; IUCN 2012) was calculated for each species overlapping the records by a four-square kilometer grid, and the proportion of endemic bats by municipality, department and ecoregion was reported.
The identification of areas of endemism was explored based in these eight bat species. An area of endemism is a group of geographic units that share at least two species (geographic synapomorphy; Morrone, 1994). A method used to identify areas of endemism is the Parsimony Analysis of Endemicity (PAE; Rosen, 1988), which joins sets of geographic units based on geographic synapomorphies (Morrone, 1994). Parsimony Analysis of Endemicity is an insightful method for inferring historical patterns as the areas of endemism (Nihei, 2006). Therefore, a binary matrix was built by coding the presence (1) or absence (0) of the endemic species in ecoregions (Olson et al., 2001), intersecting the records with the eco-region maps. A hypothetical area coded by zeros was included to root the cladogram (Morrone, 1994). The matrix was processed applying a PAE (Morrone, 1994) using a heuristic search option, 100 trees to keep, random seed 20, Multiple TBR+TBR and Fast optimization in NONA (Goloboff, 1998), and Winclada (Nixon, 2000).
Additionally, ecological niche models of four endemic bats were proposed (Anoura cadenai, A. carishina, L. marinkellei and V. sinchi), which were used to identify their potential distribution. Potential distributions of species with low sample sizes can be used to identify unknown populations and distributional areas (Pearson et al., 2007). The ecological niche models were built using Maxent 3.4.0 (Phillips & Dudík, 2008;Phillips et al., 2017) as implemented by the 'dismo' package (R Core Team, 2016;Hijmans et al., 2017) using random seed. The four remaining endemic bats were characterized by the small number of records (n ≤ 2); therefore, the delete-one jackknife modeling approach by Pearson et al. (2007), called "n -1 jackknife", was used. Only occurrence data identified at the specific level and with longitude-latitude data was used. All occurrence data were verified geographically and taxonomically following the suggestions by Chapman (2005). As predictors, the set of 19 bioclimatic variables (30 arc-seconds (∼ 1 km) was used, obtained from the WorldClim database (www.worldclim.org ver. 2.0; Fick & Hijmans, 2017). Logistic output format was used to describe the probability of presence (Phillips & Dudík, 2008), it is a continuous habitat suitability range between 0 (unsuitable) and 1 (the most suitable). Lineal, quadratic and product features were applied as well as other Maxent settings by default and made multiple predictions per species with one of the records excluded in each case, assessing the model's performance by testing the ability of the model to predict the excluded locality under a statistical significance (Binomial test) (Pearson et al., 2007). The lowest predicted value (lowest presence threshold, LPT) associated with any of the observed presence records was used to test the models (Pearson et al., 2007; P < 0.05). It represents the lowest value of the prediction for any of the pixels that correspond to the localities used in calibrating the model. The threshold to convert the ecological niche models maps from suitability indices to presence/absence was applied.

Results
Endemic bats of Colombia have been poorly studied. The species that appears in a larger number of publications is Lonchorhina marinkellei (n = 15) which is the oldest described endemic species of the country (Figs. 1a, 1b); however, most of these publications are compilations. In contrast, Lonchorhina mankomara and V. sinchi (Figs. 1c,1d) are only documented in two or one publication (the original description), respectively, that together with A. carishina (three publications) are the endemic species of Colombia described in the last decade. The localities of each species, related in the content about distributions, are indicated with numbers ( Fig. 2).

Distribution
It is only known from two localities; the first is the inter-Andean part of the nor thern section of the

Conservation
S. antioquensis is currently listed as Endangered by the International Union for Conservation of Nature -IUCN (Solari, 2016a), because it is only known from two localities, the extent of occurrence is less than 5 000 km 2 and its area of occupancy (AOO) is estimated to be 8 km 2 , and there is no information on aspects of the biology of this species.

Comments
S. antioquensis is morphologically closer to S. gymnura. The skull of the holotype (CTUA 617) is shown in Fig. 1. The morphological traits forearm length (36 mm), zygomatic arch width (8 mm), and absence of dorsal lines can be used to separate this taxon from congenerics, but the morphology of this taxon is poorly known. Molecular information is needed to corroborate whether this taxon is a valid species because it is the only species of the genus lackingof phylogenetic analyses (Lim, 2010). The paratype was not found (DMM) in 2016 at CTUA and it is probably lost.
State of knowledge: It is the endemic bat species that has been included in more publications (Marinkelle, 1976

Comments
Genetic analyses are needed to clarify whether the morphological differences with L. marinkellei are not due to clinal variation. Up to date, all documented specimens assigned to L. mankomara from the new locations presented herein fit within measurements in the description of the species. Based on captures, the species is more active late in the night after 21:00 h and early in the morning between 4:00 and 5:00 h. Many individuals were captured at the same time and mist-net, suggesting a gregarious foraging behavior, with larger numbers of females. All captures of large Lochorhina (L. marinkellei and L. mankomara) both at the PNN Chiribiquete (central and northern portions) and Serrania de La Lindosa, in the departments of Caquetá and Guaviare, Colombia, were registered at rocky formations near or on water courses. The only material recovered in the feces of the species corresponded to a soft and sometimes liquid mass of an apparent animal origin. Four specimens on which the species description was based on were previously reported as L. marinkellei by Montenegro & Romero-Ruiz (1999 4).

State of knowledge
It is only known from the holotype and the original description. The species description presents limited information on morphology and the taxonomic and distributional descriptions (Contreras Vega & Cadena, 2000). Three additional papers and one congress abstract provided information on its taxonomy (Gardner, 2008b;Solari et al., 2013;Castaño et al., 2017;Torres-Arboleda et al., 2019), and two on threats/conservation (Simmons, 2005;Mantilla-Meluk, 2015).

Conservation
The species has been listed as Data Deficient by the IUCN (Mantilla-Meluk, 2015). It was not included in the in IUCN/SSC Action Plan -2001 (Simmons, 2005). Its area of occupancy (AOO) is estimated to be 4 km 2 .

Comments
This taxon probably represents a junior synonym of S. parvidens (Solari et al., 2013;Castaño et al., 2017). No additional records apart of the holotype, collected in August 26, 1991 are known and the molar differences in the lower molars (presence of an entoconid plus a paraconulid in the lower molars) are perhaps an abnormality (Solari et al., 2013;Castaño et al., 2017). Genetic analyses using crusties from the holotype might clarify the controversial specific status of this taxon. It is only known from the original description (Tavares et al., 2014). The species description presents limited information on ecology and natural history (diet and distribution) and threats /conservation, associated with the taxonomic and distributional descriptions.   Fig. 2).

Conservation
There are not current assessments of its threaten category. Due the limited extend of occurrence, and its rarity (even rarer than V. melissa which is considered as VU by the IUCN; Ramírez-Chaves et al., 2015), this species might be considered in the Endangered category. Its area of occupancy (AOO) is estimated to be 16 km 2 .

Comments
The species was separated from V. melissa by its larger size. However, genetic analyses are needed to corroborate that size differences are associated with linage diversification. The holotype was collected on May 10, 1971. The most recent record from department of Cauca was collected on May 5, 2005 ( Fig. 1d and Fig. 5; Tavares et al., 2014). One specimen on which the description was based (paratype IAvH 2282) collected in August 3, 1976, was previously reported as V. melissa (Lemke et al., 1982).  Fig. 2).

Conservation
The species has been listed as Data Deficient by the IUCN . Its area of occupancy (AOO) is estimated to be 20 km 2 .

Comments
The

Conservation
The species has not been included in any assessment. Its area of occupancy (AOO) is estimated to be 16 km 2 . The species has been recently reported at the Department of Quindio, vereda La Virginia, Calarcá.

Comments
Considering that A. carishina was separated from the A. geoffroyi complex, genetic analysis are needed to provide further evidence of the species validity.

Conservation
The species has not been included in any assessment. Its area of occupancy (AOO) is estimated to be 4 km 2 .

Distribution
The distribution of the endemic bats includes 10 departments from which departments of Cauca, Guaviare, Nariño and Valle del Cauca have the greater number of endemic bats (2 species; Fig. 7a). At municipal level, the endemic bats are distributed in zones from 16 municipalities. Cali (2 species) supports the greater numbers of endemic bats (Fig. 7b). The endemic bats inhabit nine Colombian ecosystems ( Table 1), but the most representative ecosystem was the Cauca Valley  montane forests with four species. One area of endemism composed by two ecosystems (Caquetá moist forests and Purus varzeá) and supported by L. mankomara and L. marinkellei was identified in Colombia (Fig. 8). The potential distribution of A. cadenai showed higher values of suitability in the departments of Cauca, Huila, Risaralda Tolima and Caldas, on the Central Cordillera (Fig. 9, Suppl. 1). The potential distribution of A. carishina extends throughout the three Andean mountains (Western, Central and Eastern mountains), from north of Ecuador to Antioquia and Norte de Santander in Colombia (Fig. 10, Suppl. 1). The potential distribution of L. marinkellei included zones on the eastern Llanos Basin of Colombia and Amazonia in the departments of Arauca, Casanare, Guaviare, Guainía, Meta, Vaupes, and Vichada (Fig. 11, Suppl. 1). V. sinchi is potentially distributed on the Western Cordillera, and it includes zones from the north of Ecuador and South of Colombia (Nariño) to Caldas and Antioquia (Fig. 12, Suppl. 1).

Discussion
The definition of endemic bat species for Colombia has been a t opic that was little explored in the last 20 years. For 2000 no bat species were listed as endemic of Colombia (Alberico et al., 2000), however, for 2013 the number of endemic species for the country was seven (Solari et al., 2013), but of these Hsunycteris cadenai (Phyllostomidae) was reported in Ecuador (Tirira, 2012).
The two-additional endemic species, in comparison with the information of Solari et al. (2013) are V. sinchi and L. mankomara that were described in 2014 and 2018, respectively. It is possible that some of the endemic species of Colombia inhabit in neighboring countries. Considering the potential distributions maps, A. carishina and V. sinchi are likely distributed in Ecuador. Similarly, L. marinkellei could be present in Venezuela; however, some Lonchorhina have very specific habitat requirements that might limit their real distribution (Morales-Martínez & López-Arévalo, 2018). Although, the ecological niche models using small number of samples cause overprediction in the potential distributions or no significative model (v.g. Potential distribution of A. carishina; p-value 0.1), the maps of the endemic species are initial hypotheses that provide valuable biogeographical information to identify regions that have similar environmental conditions to where the species currently inhabits (Pearson et al., 2007;Proosdij et al., 2016), which could be used to select places to search new populations (Pearson et al., 2007). The number of endemic species found in Colombia is low; however, this seems to be the general pattern for other South American countries such as Ecuador and Peru ( six and seven endemic species respectively; Pacheco et al., 2009;Tirira, 2015). Larger countries as Brazil holds 14 endemic bats, mostly phyllostomids (Paglia et al., 2012). Hernández-Camacho et al. (1992) proposed 58 centers of endemism in Colombia, based on the identification o f r e stricted r a nge species (< 50 000 km 2 ). Comparing the known distribution of the endemic bats with the centers of endemism (Hernández-Camacho et al., 1992), the endemic bats support nine centers. Under actual knowledge, five endemic bats are restricted to Colombian endemism centers proposed by Hernández-Camacho et al. (1992): S. antioquensis (Nechi-Nare Center) L. marinkellei (Vaupés Center), L. mankomara (Chiribiquete Center), A. cadenai (Biogeographic Chocó Center), C. monohernandezi (Piedemonte de Florencia Center), and two species are reported in two endemic centers V. sinchi (Piedemonte de Florencia Center and Cofane Center) and A. carishina (Chocó Center and Sierra Nevada de Santa Marta Center). By identifying the coincidence of the distribution bats with endemism centers of Colombia, this study adds eight species of bats to the list of 45 species and subspecies of endemic mammals that support the endemism centers (Hernández-Camacho et al., 1992), and also revalidates the importance of bats in the identification of geographic zones. For example, Cauca Valley montane forests could be a prime ecosystem to analyze the ecology of four endemic bats, whereas departments as Valle del Cauca and Caquetá, and the municipalities of Calamar, Cali or Mitú comprise zones that could be important to develop conservation strategies (i.e., AICOM's: Áreas Importantes para la Conservación de Murciélagos/Important Areas for Bat Conservation, and SICOM's: Sitios Importantes para la Conservación de Murciélagos/Important Sites for Bat Conservation) around the endemic bats.
Parsimony Analysis of Endemicity based on endemic bats resulted in the identification of one area of endemism in the Amazon basin. Although it is comprised of only two species (L. mankomara and L. marinkellei), the Amazon basin has been identified as an important center of endemism (Kier et al., 2009), and it is recognized as an important biogeographic pattern to others biological groups (Costa et al., 2000;Noguera-Urbano & Escalante, 2015), due to multiple neotropical evolutionary and geological process that occurred in that area (Hoorn et al., 2010). Even though endemic species represent a national natural patrimony that should be priority for research activities and conservation, none of the species evaluated have been deeply studied or included in the national red lists or government resolutions (Rodríguez-Mahecha et al., 2006;Ministerio de Ambiente y Desarrollo Sostenible, 2017). It is an urgent activity to clarify the taxonomic status of at least four of these endemic bats of Colombia, thus, the risk of extinction of these lineages could be properly assessed. The inclusion of these endemic species in evolutionary studies is an urgent activity considering that they are missing from phylogenies (Lim, 2007). Further study of this topic should contribute both to increased understanding of the endemic species in Colombia, and to propose strategies and policies of conservation of the endemic species. Finally, endemic species (specialist species) probably are expected to have smaller tolerances to climate change in comparison whit generalist species (Brown, 1995). Therefore, negative factors as climate change (Thuiller et al., 2005), and the loss of biological and socioeconomic functions of ecosystems, affect the endemic species too (Pinilla-Buitrago et al., 2018); thus, they require to be treated as targets of a global, regional, and national conservation strategies.

Conclusions
Even though the first endemic bat of Colombia was described nearly 30 years ago, little is known on the species under this category. In general, the pattern is the absence of knowledge in all the topics evaluated. However, the distributional information is the most recurrent research topic, but the records for all endemic species come from only 21 localities. Field work to obtain new records of endemic bat species of Colombia is needed, and molecular tools should be used to address controversial issues on taxonomy and systematics. With this novel information, the threaten category for all the endemic species should be re-assessed.

Héctor E. Ramírez-Chaves
Is interested in mammalogy, evolution and conservation. He currently works at the Biological Department of Universidad de Caldas, and is the Vertebrate Curator at the Centro de Museos, Museo de Historia Natural, Universidad de Caldas.

Elkin A. Noguera-Urbano
Is a biogeographer, and his main research interests are in mammalogy, biodiversity, systematics, conservation planning and spatial modelling.

Darwin M. Morales Martínez
Is a mammalogist and his main research interest is taxonomy, ecology, and systematics of Neotropical bats.

Danny Zurc
Is currently a Ph.D. student, with a Master in Biology at the Universidad de Antioquia. She is the Curator of the biological collections of the Museo de Ciencias Naturales de La Salle, Instituto Tecnológico Metropolitano (Medellín-Antioquia). She has published several papers on the taxonomic status and acoustic identification of bat species, appropriation of science knowledge, and software development.