INTRODUCTION

Databases on the occurrence and abundance of species aim at compiling and making taxonomic data available, usually scattered throughout the literature, museum collections and other non-published sources, such as monographs, dissertations and theses (^{Hortal 2008}). It is undeniable that the availability of biodiversity databases has greatly advanced and improved studies on ecological and biogeographical patterns (^{Santos et al. 2010}). However, these online databases (e.g. GBIF and SpeciesLink) are based on the digitalization of natural history collections of different museums, where a formal control process is very often nonexistent, and the amount of detail and quality of data are strongly dependent on the data supplier (^{Maldonado 2015}). Therefore, it is not uncommon to find species occurrence records with erroneous georeferencing or mistakes in their taxonomic identification, and verifying the data with substantial taxonomic knowledge before applying biodiversity analysis is recommended.

Limitations on the accuracy and precision of the data (e.g., erroneous georeferencing and taxonomic misidentification) may lead to two types of errors on species distribution (^{Hortal 2008}). Error 1 leads to false positive, or the attribution of a species to a site where it does not occur; error 2 leads to false negative, or the omission of a species on a site where it actually occurs. These errors address issues in more complex ecological, biogeographic and conservation studies. For example, distribution or niche models use species occurrence and predictor variables of occurrence site to build their projections (i.e. ^{Dobrovolski et al. 2012}; ^{Aguiar et al. 2016}; ^{Delgado-Jaramillo et al. 2020}; ^{Batista et al. 2020}). Furthermore, knowledge on the biodiversity of forest remnants is crucial for the identification of new Priority Areas and for maintaining existing ones (i.e. ^{Sigrist & Carvalho 2008}; ^{Campos & Lourenço-De-Moraes 2017}). Consequently, databases should be developed by experts to guarantee that the information is reliable, of good quality, and systematized. Compiling local inventories of species, and making this information published and accessible, is important to serve as a basis for the most accessed global databases. That is why local lists are important to maintain the quality of global databases and of the more complex studies developed based on them.

There are 182 bat species recorded in Brazil (^{Garbino et al. 2020}) and, at least, 63 in the state of Paraná (^{Passos et al. 2010}; ^{Garbino & Nogueira 2017}). Bat fauna from the region of Londrina, northern Paraná, has been investigated since 1982 by a research group at the Laboratório de Ecologia de Mamíferos of the Universidade Estadual de Londrina, being one of the most well studied regions for these taxa in the state (^{Miretzki 2003}). These almost four decades of sampling efforts were largely supported by public financial investment (i.e. CAPES, CNPQ and Fundação Araucária), which further highlights the importance of retrieving this information and making it public. In fact, about 44% of our data primary source comes from unpublished data. Therefore, in this study we compile and organize all available information on published and non-published data of bat species for the region of Londrina to provide a systematized and taxonomically verified database. We also detect and point out information gaps to direct where and how new inventories should be undertaken. These data are particularly relevant for the state of Paraná, which is in third place in the deforestation ranking of the Atlantic Forest (^{SOS Mata Atlântica 2018})—with the north of Paraná being one of the most affected and fragmented regions. Efforts for cataloging current and new species occurrences in Londrina are, therefore, urgent.

MATERIALS AND METHODS

Londrina region and Compilation of the species list

The municipality of Londrina is located in the northern region of the state of Paraná, Brazil, with an area of approximately 1 650 809 km² (^{IBGE 2010}). The climate is humid subtropical (Cfa according to Köppen-Geiger classification; ^{Alvares et al. 2014}), with hot summers, and rainfall throughout the year, decreasing in the winter. The region belongs to the Atlantic Forest biome within the Alto Paraná ecoregion, covered by Semi-deciduous Seasonal Forest. Mean annual rainfall is 1888 mm and mean annual temperature is 21.6 °C (^{IAPAR 2019}). The Londrina region is home to 24 urban forest fragments, two protected areas (i.e., Conservation Units) and Legal Reserves within farms in rural areas, some of which have been subject to short or long-term sampling for bats between 1982 and 2018 (Supplementary Mat. 1, Fig. 1).

Fig. 1 Location of the Londrina region, Paraná state, Southern Brazil. Black dots: areas sampled for bats PMAT – Parque Municipal Arthur Thomas (Latitude: -23.344167, Longitude: -51.133500); PEMG – Parque Estadual Mata dos Godoy (Latitude:-23.448056, Longitude: -51.255833); UEL –Universidade Estadual de Londrina campus and Horto Florestal (Latitude: -23.321944, Longitude: -51.201111); JBL – Jardim Botânico de Londrina (Latitude: -23.366000, Longitude: -51.177700); MB – Mata do Bule (Latitude: -23.403000, Longitude: -51.337000); FR – Fazenda Regina (Latitude: -23.183806, Longitude: -51.127694), FI, CK, CV – Fazenda Imbaúva, Chácara Kanashiro and Chácara Vitória (Latitude: -23.380144, Longitude: -51.076219). 

We listed the occurrences (species richness; presence) of bat species registered in Londrina between 1982 and 2018 throughout the scientific literature and non-published data. The latter were retrieved from field notes, monographs and dissertations at the Laboratório de Ecologia de Mamíferos of the Universidade Estadual de Londrina (Supplementary Mat. 1). The first step consisted in verifying all records and establishing filters for the exclusion of false occurrences. Exclusion criteria comprised lack of voucher specimen deposited in the Museu de Zoologia (MZUEL) or at the Laboratório de Ecologia de Mamíferos of the Universidade Estadual de Londrina; expected geographic distribution of species based on the literature; and the lack of precise site where the species was captured.

The species list for Londrina region followed the nomenclature and taxonomic arrangement proposed by ^{Garbino et al. (2020)} for the Brazilian bats. Conservation status of each species was based on international (^{IUCN 2020}), national (^{ICMBio/MMA 2018}) and the regional lists of endangered fauna (^{Margarido & Braga 2004}).

Data Analyses

A Venn-Euler diagram (^{Venn 1980}) was used to illustrate the occurrence of bat species in the region of Londrina. In order to verify the sample gaps in the region of Londrina, data were tabulated in a database (Supplementary Mat. 1 and 2), where we compiled information about the number of capture nights (the only common information among published and non-published data available for the sampling effort) and the environments where mist-nets were installed in each sample site. Therefore, we evaluate whether sampling effort was heterogeneously distributed in the different environments within the forest fragments.

Using ArcGIS 10.5 from ESRI (ArcGIS trial), a 30 km² buffer from downtown Londrina was constructed. The distance was large enough to comprise all of the sampled forest fragments for bats in the region. Based on this vector layer, the area (ha) of all fragments located within a 30 km2- radius were added. However, the percentage of sampled area was calculated using only the fragments that had an exact georeferenced location (namely, PEMG, PMAT, Mata do Bule, Jardim Botânico, Horto Florestal/UEL, and Fazenda Regina). In the case of non-sampled fragments, we employed the ArcGis option for the classification of the vector layer according to certain characteristics. The variable ‘fragment size’ coupled with the use of six classes was selected. Non-sampled fragments were divided into: smaller than 30 ha; between 30 and 48 ha; between 48 and 70 ha; between 70 and 114 ha; between 144 and 196 ha; between 196 and 504 ha. This allowed us to locate and determine the number of fragments in each class.

RESULTS

Forty-one bat species distributed into four families, nine subfamilies and 27 genera were listed for the Londrina region, in 577 sampling nights from 1982 to 2018. All bat trophic guilds (nectarivores, frugivores, insectivores, carnivores, omnivores, hematophages, and piscivores) are represented (Table 1). The Conservation Units PEMG and PMAT harbor, respectively, 68.3% and 41.50% of the bat fauna in the region of Londrina. Seven of the 12 species which were not recorded in these sites (Table 1, Fig. 2) were listed only in the urban forest areas of the metropolitan region; two species were registered on protected areas such as Legal Reserves within private farms, and other two species in the metropolitan region roosting in human buildings; there was no environmental information on the locality where Eumops perotis was captured in the metropolitan region of Londrina. Four species in our list have some degree of extinction threat according to red lists, with Diaemus youngii classified as “critically endangered” (^{Margarido & Braga 2004}) Diphylla ecaudata, Chrotopterus auritus and Chiroderma doriae listed as “vulnerable” (^{Margarido & Braga 2004}); and Myotis ruber as “near threatened” (^{IUCN 2020}).

Table 1 List of species for the Londrina region, Paraná state, Southern Brazil, based on literature and unpublished data from the Laboratório de Ecologia de Mamíferos, Universidade Estadual de Londrina. Trophic Guild: C- Carnivorous; I - Insectivorous; H - Hematophagous; O - Omnivorous; N - Nectarivorous; F - Frugivorous; P - Piscivorous. Capture sites Parque Estadual Mata dos Godoy (PEMG); Parque Municipal Arthur Thomas (PMAT); the Legal Reserves, Mata do Bule (MB), Fazenda Imbaúva (FI), Chácara Kanashiro (CK), Chácara Vitória (CV), Fazenda Regina (FR), Universidade Estadual de Londrina campus and Horto Florestal (UEL) ; Jardim Botânico de Londrina (JBL), banks of the Igapó Lake (IL), non-identified valleys and human building (UA). Threat categories based on a regional (RE, ^{Margarido and Braga, 2004}), National (NA, ^{ICMBio, 2014}) and international (IN, IUCN, 2020) lists: LC- Least concern; VU - Vulnerable; CR - Critically Endangered; DD - Data deficient; NT - Near threatened. 

Fig. 2 Venn-Euler diagram to illustrate occurrence of bat species in the Londrina region, Paraná state, Southern Brazil. – PEMG- Parque Estadual Mata dos Godoy; PMAT- Parque Municipal Arthur Thomas; Other sites - Mata do Bule, Fazenda Imbaúva, Chácara Kanashiro, Chácara Vitória, Fazenda Regina, Universidade Estadual de Londrina Campus/ Horto Florestal, Jardim Botânico de Londrina, banks of the Igapó Lake, non-identified urban areas (valleys and human building). The arrow indicates the Phyllostomus hastatus position in the diagram. 

Spatial analysis revealed that less than 4% of forest fragments in the region of Londrina, within a 30 km radius, were sampled for bats. Moreover, most of the sampling (with the exception of Fazenda Regina) took place in the southern part of the Londrina region. Most forest fragments remain unsampled for bats (Fig. 3). There are seven fragments with an area between 196 and 504 ha, of which five are greater than 300 ha; nine fragments between 196 and 114 ha; 21 fragments between 114 and 70 ha; 27 fragments between 70 and 48 ha; 37 between 48 and 30 ha; and several fragments with less than 30 ha.

DISCUSSION

Our review demonstrated that the region of Londrina, with 41 bat species (14 more species than indicated for the region in the current database for Atlantic Forest bats – (^{Muylaert 2017})), harbors at least 65% of the bat fauna found in the state of Paraná (^{Passos et al. 2010}; ^{Garbino & Nogueira 2017}), and 43% of the species found in the Atlantic Forest (^{Muylaert 2017}). However, most of the Londrina region has never been inventoried and the region may have a larger species richness than currently registered. We have excluded the record for Uroderma bilobatum, which was previously recorded in Londrina (^{Reis et al. 1998}). This was the only known record of this species in the Paraná state, but was not included in the present species list, because no voucher specimen has been deposited in the MZUEL or in the Laboratório de Ecologia de Mamíferos of the Universidade Estadual de Londrina, where the bats reported in ^{Reis et al. (1998)} were deposited (^{Garbino & Nogueira 2017}). Based on the species’ natural history and lack of additional voucher specimens in collections, all reports of U. bilobatum for Paraná state and southern Brazil should be disregarded (see ^{Garbino & Nogueira 2017}). Although there is no voucher of the Eumops perotis reported for Londrina by ^{Suckow et al. (2010)}, the occurrence of this species was confirmed by photographs of one individual (Fig. 4) which was captured on August 3, 2004 and forwarded by a city resident to the Laboratório de Ecologia de Mamíferos of the Universidade Estadual de Londrina for identification, and then released. It is possible to recognize the species based on the photographic record of the individual that can be seen in Fig. 4. The forearm of this individual measured 79.1 mm (^{Suckow et al. 2010}), corroborating the morphometric data of the species, the largest of the genus, with a forearm ranging from 70.2 to 83.4mm (^{Sartore et al. 2017}). Furthermore, according to the distribution of the species, it is expected to occur in northern Paraná (^{Suckow et al. 2010}).

Fig. 3 a) The 30 km² buffer from downtown Londrina, Paraná state, Southern Brazil. Green: fragments sampled for bats. Other colors: unsampled fragments according to size in hectares. b) Forest fragment where the Parque Estadual Mata dos Godoy (PEMG) and the Ribeirão dos Apertados stream are located. Bat capture sites: 1-Trilha dos Catetos; 2- Trilha das Perobas. 

The urban forest areas and the protected Legal Reserves in the rural area of the municipality are relevant for the maintenance of bat diversity in the Londrina region. In fact, when they were analyzed together, they harbored 95% of the bat species. The two Conservation Units (PEMG and PMAT) are sites with high species richness and may constitute suitable areas for species classified as threatened. The PEMG is the sole occurrence site for two threatened hematophagous species, D. ecaudata and D. youngii. Our results have shown that despite the great sampling effort in the region, there are still many sampling gaps. The region’s biggest forest fragment comprises 2000 ha, but samplings are limited to PEMG, which has an area of only 680 ha within this forest fragment (Fig. 3). PEMG is one of the main forest remnants in the state of Paraná and one of the main genetic forest reserves in south Brazil (^{IAP 2015}), with the presence of different micro-habitats (^{Willrich et al. 2019}). In spite of the great sampling effort along the years (238 capture nights) in the area, the sites where mist-nets were installed are mainly in the visitor’s tracks located in the northern part of the park (Fig. 3). The southern portion of the PEMG, in the banks of the Ribeirão dos Apertados stream has not been systematically sampled, there were only two sampling nights with zero specimen deposited in collections and without any field entries. The region lies along an elevated gradient associated with changes in vegetation structure (^{Willrich et al. 2019}), and houses several water bodies that provide a different set of resources for species such as Noctilio leporinus (a piscivore species, which has never been captured in the park), thus, increasing the probability of catching insectivorous species that forage above water. The above suggests that there are probably more bat species in the PEMG area which have never been registered.

Smaller fragments (including urban parks) and Legal Reserves can also be important for the maintenance of bat diversity in Londrina. However, these areas have low or no sampling, which suggests that Londrina may actually have a higher richness of bat species. A good example is the Mata do Bule Legal Reserve, where, with a sampling effort of only 14 nights in 2014 and 2015, a total of nine species were recorded, including M. ruber, listed as near threatened (^{Solari 2019}).

Furthermore, 96% of the area covered by forest fragments has never been sampled, while the remaining 4% has an uneven distribution of sampling nights. For example, only one forest fragment (Fazenda Regina) has been sampled in the northern region of Londrina; all the other forest fragments lie in the southwestern region of Londrina. Moreover, the southeastern region contains three of the five largest fragments (over 300 ha) in a 30 km radius that have never been sampled (Fig. 3), showing the huge sampling gap for the Londrina region.

There is still a methodological issue to be considered regarding potential sampling gaps for bats in Londrina. It is very common for bat inventories to be carried out using mist-nets only, which may limit the detectability of bat families that are less prone to be sampled by this method, such as Molossidae, Emballonuridae and Vespertilionidae (^{Simmons & Voss 1998}; ^{Carvalho et al. 2013}). For instance, there are six insectivore species (Eumops glaucinus, Eumops perotis, Nyctinomops macrotis, Tadarida brasiliensis, Eptesicus taddei, Rhogeessa io and Myotis levis), which have not been captured within the protected areas of Londrina where bat captures were performed mainly with mist-nets. Lack of precision on the localities where these insectivore species were captured only allows us to assert that the above species were captured within the urban area. Urban areas usually present an abundance of insects that can be easily caught (insects flying around illuminated objects) and roosting areas (human buildings), which attracts insectivore bats (^{Pacheco 2010}). Moreover, it is easy to capture insectivorous bats in urban environment by actively searching for day roosting areas and placing mist-nets close to their exit. It is highly probable that the absence of occurrence records for these species within protected areas such as Conservation Units and Legal Reserves of the region may be a type 2 error - the false absence of species in places where they may actually occur.

Bat capture undertaken only by mist-nets in forest fragments may increase type 2 error for insectivore bats of the families Molossidae, Vespertilionidae and Emballonuridae. Bats from these families usually fly very high and have a more efficient eco-localization system, making their capture with mistnets a difficult task (^{Nogueira et al. 1999}, ²⁰⁰⁸). This sample bias can lead to information gaps, for example, six species of insectivorous bats on our list (Cynomops abrasus, Eptesicus diminutus, Eptesicus taddei, Rhogeessa io, Histiotus velatus and Myotis ruber) of the Molossidae and Vespertilionidae families, are categorized as "data deficient" by at least one of the lists that assessed the level of extinction threat towards species (Table 1). Therefore, it is imperative to carry additional bat sampling efforts in the Londrina region using alternative and complementary methods (^{Bergallo et al. 2003}), because these will surely increase the number of species recorded for the region, considering that the region was not considered sufficiently sampled. Methods include searching for roosting sites during the day (^{Simmons & Voss 1998}), the construction of artificial roost to attract bats (bat houses) (^{Tuttle & Donna 1993}), the monitoring of eco-localization callings to allow the identification of species by ultrasound (^{Arias-Aguilar et al. 2018}), and the installation of canopy mist-nets (^{Carvalho & Fabián 2011}).

Fig. 4 A Eumops perotis individual received alive by the Laboratório de Ecologia de Mamíferos, Universidade Estadual de Londrina, on August 03, 2004. The bat was identified and released. Photo: I. P. de Lima. 

Our results are based on reviewing the scientific literature and careful compilation of several small monitoring projects in Londrina since 1982, which allowed us to determine the bat species that occur in the area and the sampling gaps. The last compilation of bats for the entire Atlantic Forest (Atlantic Bats - (^{Muylaert 2017})) brings the occurrence of 27 species to the region of Londrina (distributed only in the two Conservation Units - PEMG and PMAT). Our database adds 14 species to the region of Londrina, which highlights the importance of reviewing data in primary sources (such as unpublished dissertations and theses, for example). However, this is not always possible during the construction of larger scale databases, highlighting the relevance of local reviews and compilations of species in published databases, which serve as sources of reliable and more accessible information. Besides being a reliable dataset, our review also provides the basis for more complex ecological studies, as well as guidance to where future bat surveys should be conducted in the Londrina region. Finally, we show the relevance of the conservation of different forest fragments of different types in the landscape (e.g., urban forest areas, Legal Reserves and Conservation Units) which should be highlighted to maintain the diversity of bats in the region of Londrina.