Check List

The Seasonally Dry Tropical Forests in Peru are well known for their bird and plant diversity and endemicity, but little is known about the diversity of small mammals. We report the diversity of small volant and non-volant mammals from the Seasonally Dry Tropical Forests of the Huallaga river basin in the San Martín Department, working on both sides of the river, making a sampling effort of 3060 traps-night for non-volant and 104 mist nets-night for volant mammals. We recorded 29 species, including five marsupials, three rodents, and 21 bats. Among the bats, phyllostomids were the most diverse group with 16 species. Short-tailed Spiny-rat, Proechimys brevicauda (Günther, 1877) and Seba’s Shorttailed Bat, Carollia perspicillata (Linnaeus, 1758), featured the highest relative abundance. In addition, we report the first records for the San Martín Department of Peropteryx macrotis (Wagner, 1843), Saccopteryx bilineata (Temminck, 1838), Lonchorhina aurita Tomes, 1863, Vampyriscus bidens (Dobson, 1878), and Myotis simus Thomas, 1901. The species richness and diversity indices indicate the study site has a high diversity value; however, fragmentation and rapid changes in land-use are the main threats faced by the biodiversity of these dry forests.


Introduction
Seasonally Dry Tropical Forests (SDTF) are characterized by pronounced seasonality, annual rainfall less than 1600 mm, receiving about 80% of the rains during the wet season, resulting in several drought months during the dry season, which can last five or six months (Pennington et al. 2000;Murphy and Lugo 1995;Maass and Burgos 2011). These forests have a smaller stature and smaller basal areas than humid tropical forests (Pennington et al. 2000;Murphy and Lugo 1995). They range from Mexico to northern Argentina and southeastern Brazil, representing 22% of the forested area of South America (Murphy and Lugo 1986;Olson et al. 2001); but having a fragmented distribution with areas of different sizes (Espinosa et al. 2012).
SDTF are floristic and structurally more complex than the rainforests (Murphy and Lugo 1986), although their species diversity is generally low (Diaz-Pulido et al. 2014). These forests are unique ecosystems, with part of the fauna associated with these forests presenting morphological, physiological, and behavioral adaptations to the availability of resources between seasons (Särkinen et al. 2011;Diaz-Pulido et al. 2014). Therefore, even though SDTF have low diversity, they have endemic species (Ceballos 1995). SDTF have been considered very fragile ecosystems, threatened by changes in land use and human population growth (Janzen 1988;Maass et al. 1995).
Peruvian SDTF occur on the western and eastern slopes of the Andes and are divided into three subunits: Equatorial (Tumbes, Piura, Lambayeque, and isolated remnants in La Libertad), inter-Andean (valleys of Huancabamba, Marañón, Apurímac, and Mantaro, and in some valleys of Cusco and Puno), and Oriental (surroundings of Tarapoto in San Martín) (Linares-Palomino 2004b). Most of these forests range from sea level to 1000 m elevation, but the Mantaro Valley and Apurímac Valley reach up to 2350 m and 2400 m, respectively (Linares-Palomino 2004b; Aguirre et al. 2006;Linares-Palomino and Pennington 2007).
The diversity of plants (e.g., Linares-Palomino 2006) and birds (e.g., Vásquez-Arevalo et al. 2018;Saldaña et al. 2020) are well studied in these forests, but little is known about the diversity of small mammals. Neighboring countries such as Colombia and Ecuador have a better record of mammal species associated with SDTF (Tirira 2001;Boada and Roman 2005;Calonge et al. 2010;Diaz-Pulido et al. 2014). In Peru, two studies have surveyed small areas of dry forests in Tumbes ) and Apurímac (Pacheco et al. 2007b). As yet, no survey has been conducted in the dry forest of San Martín department, the most isolated Peruvian dry forest (Linares-Palomino 2004b). This lack of information is of great concern because San Martín is one of the most deforested departments in Peru (MINAM 2020). Previous extensive studies in this department are scarce and had been carried out mostly in the Río Abiseo National Park (Leo and Romo 1992;Gardner and Romo 1993;Leo and Gardner 1993) and the Mayo river basin , which are located in montane and humid lowland forests, respectively.
Our study evaluates the diversity, relative abundance, and species richness of small mammals in the SDTF of the Huallaga river basin, San Martín Department. We discuss the conservation status of the species and the forests and update the species richness present in San Martín department.

Study Area
The SDTF of the Huallaga river basin are within San Martín department in the provinces of Bellavista, Picota, and San Martín. The climate is seasonal, with the wet season between October and March and the dry season between April to September; the average temperature is 26 °C, and the annual rainfall is 1164.4 mm (García-Villa corta 2009). Dispersed scrub dominates the vege ta tion as "Quinilla" (Manilkara bidentata (A.DC.) A.Chev.), "Capirona" (Calycophyllum spruceanum (Benth.) K.Schum.), and "Machinga" (Brosimum alicastrum Sw.). The soils are predominantly black, with large and small stones scattered on the ground (García-Villacorta 2009). Agricultural areas are extensive, with largescale plantings of rice and cacao, and notably fragment the dry forest into isolated remnants of various sizes. We surveyed during the dry season, between 22 August and 11 September 2015, in four private conservation areas (Fig. 1). These areas are protected patches of forest surrounded by extensive agricultural areas: El Valle del Biavo (VB): Bellavista Province, Bajo Biavo District (07°09′S, 076°36′W, 520 m elevation), located at the right bank of the Huallaga river. It has a low canopy with trees 5-10 m tall. The relief is undulating, with slopes reaching 75° in some areas. There are no permanent water sources and only small shallow, dispersed pools.
El Incaico (EI): Bellavista Province, Alto Biavo District (07°18′S, 076°22′W, 800 m elevation), located at the right bank of the Huallaga river. The canopy is about 15-20 m high. The relief has moderate or steep slopes, up to 75°. The soil is mainly silty, and there are no permanent water sources.
El Quinillal (EQ): between Bajo Biavo (Bellavista Province) and Tingo de Ponasa (Picota Province) Districts (07°02′S, 076°17′W, 420 m elevation), located on the right bank of the Huallaga river. The canopy is approximately 10-20 m high. The slopes are moderate, the soil is silty clay, and there are no permanent water sources.
Ojos de Agua (OA): Picota Province, Pucacaca District (06°49′S, 076°26′W, 500 m elevation), located on the left bank of the Huallaga river. The canopy is 15-30 m high. The relief is moderate, with slopes of up to 50°, and generally clay soil. There are permanent ponds and a small, 0.5-1 m wide creek.
OA (Table 1). We installed eight mist nets (12 m wide and 2.5 m high) in the understory and checked them every 30 minutes from 18:00 to 00:00 h. The mist nets were placed individually, or as two nets spliced together, in various places depending on the terrain. The distance between mist nets was at least 20 m. Mist nets were placed near the undergrowth and on the footpaths and four of them near small bodies of water in OA.
We took the standard biometric measurements, including total length (TL), tail length (T), hindfoot length (F), and ear length (E). We also measured the tragus length (Tr) and forearm length (FA) for bats. Sex and reproductive condition were also recorded. We euthanized collected specimens following the ethical guidelines of the American Society of Mammalogy (Sikes et al. 2016). We released juvenile, pregnant, and lactating individuals. We deposited all collected specimens at the Museo de Historia Natural of the Universidad Nacional Mayor de San Marcos (MUSM). We also added unpublished data from specimens in the Mammal Collection of the MUSM.

Data analysis.
We expressed the sampling effort as traps-night (TN) for non-volants and as mist nets-night (MN) for volants. We estimated the accumulation curve by the Clench model using the formula: where "a" is the rate of increase of new species at the start of the sampling and "b" is the parameter related to the shape of the curve (Jimenez-Valverde and Hortal 2003) using Estimates v. 9, and Statistica v. 13 for graphics. We determined the relative abundance for each species and sampling locality based on sampling effort (Caughley 1977) expressed as individuals per 100 trapsnight (ind/TN) for non-volants and as individuals per 10 mist nets-night (ind/MN) for volants. We also obtained the relative abundance based on percentage for each species by dividing the captured specimens per species by 100 between the total recorded specimens. We estimated the diversity of SDTF of the Huallaga river basin by the Shannon-Wiener index (H′), Simpson dominance index (1−D), and reverse Simpson index (1/D) using the Vegan package (Oksanen 2015) in R v. 3.4.1 (R Core Team 2017).

Results
Accumulation curve. The accumulation curve for non-volant mammals, adjusted 99.13% to the Clench model, predicted 10 species (a/b = 9.96), three more than recorded (Fig. 2a). The accumulated curve for volant mammals, adjusted to 99.7%, predicted 27 species (a/b = 26.58), eight more than the number of captured bat species (Fig. 2b). Although the curves did not reach the asymptote, our samples represented 70% of the nonvolant species estimated by the model and 70.4% of the volant species.
The most notable species richness was recorded in OA (16 species) and the least in EI (10 species). Most of the species reported, mainly bats, are present at both banks of the Huallaga river. Marmosops noctivagus, N. apicalis, P. macrotis, and A. planirostris were recorded only on the left bank of the Huallaga river. While M. constantiae, P. canus, C. villosum, L. aurita, M. megalotis, P. incarum, S. giannae, V. bidens, M. keaysi, and M. simus were recorded only on the right bank of Huallaga River (Tables 2, 3).
We present the first records to the San Martín department of five bat species as follow:
Remarks. This species is known to Mexico, Grenada, Central America, Colombia, Venezuela, Trinidad and Tobago, the Guianas, Ecuador, Peru, Brazil, Bolivia, and Paraguay (Hood and Gardner 2008). In Peru, P. macrotis occurs in the Selva Baja and Sabana de Palmera ecoregions ). This record is the first for the San Martín Department; it is 430 km north of the nearest record, at San Juan, Pasco department (Tuttle 1970;Hood and Gardner 2008). Figure 4B Material examined. PERU • Ojos de Agua private conservation area, Picota Province; 06°49′S, 076°26′W; 500 m elevation; D. Ruelas obs.; 6 individuals observed on the roof of the biological station.

Saccopteryx bilineata (Temminck, 1838)
Identification. This species differs from other bats in having two whitish longitudinal wavy stripes on the dorsum. Besides, the individuals had the underparts brownish; the muzzle was simple, pointed, and without a leaf-like excrescence; the uropatagium was thinly haired to the exertion of the tail; and the tail perforated the upper surface of the uropatagium. These morphological characters agree with those described by Hall (1981), Yancey et al. (1998), and .
Remarks. This species occurs from Colombia, Trinidad, and the Guianas to Bolivia and Brazil (Hood and Gardner 2008). In Peru, S. bilineata occurs in the Bosque Pluvial del Pacífico, Selva Baja, and Sabana de Palmera ecoregions ). The new record in the eastern Andes of Peru extends the range of S. bilineata 430 km north from the nearest presious record at San Juan, Pasco department (Tuttle 1970;Hood and Gardner 2008). Figure 4M Material examined. PERU • El Quinillal private conservation area, Bellavista Province; 07°04′15″S, 076°23′ 59″W; 316 m elevation; 01.IX.2015; D. Ruelas leg.; mist nets; 1 ♀ (MUSM 43864) and another specimen accidentally released. Identification. These morphological characters allow us to recognize this species: dorsal fur uniformly reddishbrown with the underparts slightly paler; ears broad, bluntly pointed, and as long as the head; tragus more than one-half the length of the ear and notched at the base; lance-shaped nose leaf with a prominent longitudinal ridge; lance tapers to a sharp point with the entire nose leaf nearly equal in length to the ears; five fleshy protuberances collectively hide the opening to the nostrils; upper lip covered with small, fleshy warts, which hang over the bottom lip; lower lip with an inverted triangular space flanked laterally by elongate, smooth pads; tail longer than the femur, and reaches to the posterior tip of the uropatagium; wings extend to the distal end of the tibia; foot shorter than calcar. Skull elongated with rostrum narrow; nasal arched, curved, and overhang the nasal opening; palatine with large projection; first premolar tiny; second premolar higher than molars; massive molars; broad mesopterygoid fossa; dental formula: 2/2 1/1 2/3 3/3 = 34. External and craniodental measurements are shown in Table 2. These morphological characters and measurements are within the variation range for the species following Lassieur and Wilson (1989), , Morales-Martínez and López-Arévalo (2018) Identification. We identified this species by the marked white supraorbital stripes; broad lateral lancet of the nose leaf; high and broad tip of the nose leaf; conspicuous dorsal line; wide uropatagium with a fringe of hairs; elongated skull; short rostrum; rectangular nasal opening; medium posteropalatal process; asymmetric and bilobed inner upper incisors; non-caniniform first lower premolar; a pair of lower incisors; high coronoid process; dental formula: 2/1 1/1 2/2 2/3 = 28. External and craniodental measurements are shown in Table 2. These characters and measurements are within the variation range for the species following Ruelas and Pacheco (2015) and . Identification. The following combination of morphological characters distinguishes this species: relatively short conical face that lacks skin folds or appendages; relatively small eyes, small ears; thin and elongated tail fully embedded within the uropatagium reaching the free edge and longer than the hind legs; short ears, extending forward halfway from the eye to nostril; barely evident antitragal notch; pointed tragus, slightly curving outward above and convex below, with a small triangular lobule at the outer base; naked dorsally uropatagium without a fringe of hairs along the trailing edge; few scattered hairs on the basal portion of the uropatagium; calcar with a small keel; third upper premolar crowded to lingual side; sagittal and occipital crests present. These characters agree with those described by Moratelli (2012) and   Additional to these five species, we confirmed the presence of two rare species in San Martín department. Figure  Identification. The following combination of morphological characters allows us to recognize this large mouse opossum: drab, woolly dorsal fur; not a well-defined facial mask; yellowish-buffy cheeks; light dorsal fur washed with orange on the body sides; yellowishcream to yellowish buffy ventral fur; two-thirds of the distal portion in ventral view slightly depigmented; long, all-dark tail with rhomboidal scales arranged in spiral series; laterally and dorsally projected supraorbital ridges; developed postorbital processes, well-developed in most mature adults; almost completely ossified palates with short-narrow maxillopalatine openings; moderately convergent temporal ridges shaping a sagittal crest; and small auditory bullae. External and craniodental measurements are given in Table 3. Morphological characters and measurements of our specimens are within the variation range for the species following Lima Silva et al. (2019) and Voss et al. (2019)

Remarks.
Nectomys apicalis is widely distributed in eastern Ecuador, eastern Peru, and northwestern Brazil but considered rare (Bonvicino and Weksler 2015). These authors also suggest that Nectomys apicalis may represent a species complex that needs study. In Peru, it is distributed in the Selva Baja and Sabana de Palmera ecoregions ). Additional records. Tarapoto, San Martín Province (MUSM 5062); Huicungo, Mariscal Cáceres Province (MUSM 24398).

Discussion
Diversity and abundance. We present the first report of small mammals from the SDTF of the Huallaga river basin. For five of these species, we provide the first records from San Martín department. Twenty-nine species of small mammals are now known from the SDTF of the Huallaga river basin, representing 22.1% of this San Martín department's small mammal species (131 species: Appendix Table A2). The cumulative curves suggest that more sampling effort is needed to obtain a more considerable species richness; therefore, we recommend continuing the diversity studies, increasing localities, and sampling effort.
We found that Chiroptera was the order with the greatest species richness (21 species), similar to other humid and dry forests (e.g., Tirira 2001; Boada and Román 2005;Pacheco et al. 2007b;Diaz-Pulido et al. 2014). Phyllostomidae was the family with the highest species richness. The number of nectarivorous species was low (only one species), probably related to incomplete sampling (Fig. 2). In effect, the insectivorous bats were better represented (Peropteryx macrotis, Pteronotus fuscus, Micronycteris megalotis, Lonchorhina aurita, Myotis keaysi, and Myotis simus) than bats from other Table 4. Relative abundance (ind/TN) and percentages (%) of the non-volant small mammals by locality concerning the Huallaga River from the SDTF of the Huallaga river basin, on the right bank: Valle del Biavo (VB), El Incaico (EI), El Quinillal (EQ), and on the left bank: Ojos de agua (OA). * = Incidental record.

Species
Right bank Left bank % VB EI EQ OA

Order Didelphimorphia
Didelphis marsupialis* ----- food guilds, especially in EI and EQ, where the sampling area was closer to inactive or abandoned farmland (D. Ruelas pers. obs.). Although insectivorous bats tend to fly high in and over the canopy (Fleming et al. 1972;Kalko et al. 1996;Bernard 2001), they were mist-netted in the understory. This is probably because the trees in SDTF are smaller than in the humid forests (Pennington 2000;Linares-Palomino 2004a). Frugivorous species were the most abundant, especially Carollia spp. and Artibeus spp., agreeing with other studies in forests with secondary growth, fragmented or disturbed habitats (e.g., Fleming 1991; Calonge et al. 2010;Mena 2010). These species have mainly a generalist diet, including insects, and a wide dispersion capacity (Fleming 1991;York and Billings 2009). Among the non-volant species, we found Rodentia was more abundant than Didelphimorphia, mainly due to the abundance of Proechimys brevicauda (4.41 ind/ TN, 61.82%). A similar pattern was reported along the La Novia River (Purús, Ucayali department), a locality in the Peruvian humid forest, where Proechimys species increased the relative abundance of rodents (Ruelas et al. 2016).

Mammalian composition in dry forests.
The similarity in the species composition of the SDTF in the Huallaga river basin (SDTF-H), the Pacific Equatorial Dry Forest (PEDF), and the Seasonally Inter-Andean Dry Forest of Apurimac (SIDF-A) is low (Appendix Table A3). One species, D. rotundus, is shared among SDTF-H, PEDF, and SDIF-A; another species, A. planirostris, is shared between SDTF-H and SIDF-A, and six species, Didelphis marsupialis, C. brevicauda, Carollia perspicillata, Glossophaga soricina, Micronycteris megalotis, and M. keaysi, are shared between SDTF-H and PEDF. On the other hand, we noted that the PEDF have mammals from lowland humid forests, such as C. perspicillata or M. megalotis, probably because this forest is surrounded on its northern portion by the Pacific Tropical rainforests.
We found that the small mammals of the SDTF of the Huallaga river basin also occur in nearby lowland humid forests in the Selva Baja ecoregion (see Pacheco et al. 2009), but there are no endemic species of dry ecosystems. The lack of species endemic to dry ecosystems or adapted to low humidity conditions is probably due one or a combination of the following: (1) the sampling effort was low, (2) the dry forests are highly fragmented by anthropogenic causes, and (3) the eastern dry forests are historically young, with insufficient time for mammal species to evolve or adapt. From the eastern Peruvian dry forest, only the Marañón Dry Valley (MDV) contains a dry-ecosystem endemic species, Platalina genovensium Thomas, 1928 (Ruelas andPacheco 2018), which is widely distributed on the western Andes. MDV also contains endemic species of other vertebrates (e.g., Koch 2014;Koch et al. 2015Koch et al. , 2018 and plants (e.g., Särkinen et al. 2011;Marcelo-Peña et al. 2016). Although these comparisons are preliminary, they suggest that eastern dry forests may not have a typical or endemic mammalian fauna. Instead, their biota is composed of fauna elements of nearby habitats.

Conservation.
Currently, the main threats to the diversity of small mammals in San Martín are the fragmentation and loss of habitat due to changes in land use mainly to expand the agricultural frontier (Marquardt et al. 2019). Changes in land use in other Peruvian regions with forests generate changes in diversity, abundance, forest dynamics, and trophic structure (Mena 2010). The human demographic growth and changed of land use for agriculture in the SDTF date back to the 1960s when the Peruvian government promoted the occupation of the Amazonian forests and the construction of the Marginal de la Selva Highway, which completely crosses the Peruvian territory, to reduce the human population pressure on the highlands and to reduce the migration towards Metropolitan Lima (CVR 2004). These actions may have directly impacted the biodiversity of the Huallaga river basin by fragmenting the SDTF there, with human settlements and farmlands mainly in the lower basin, leaving isolated forest relicts in areas with high slopes, which are not suitable for large-scale cultivation. This may explain the current composition of the small mammals in the surveyed localities. Few studies have assessed the impact of fragmentation and deforestation on small mammals in Peru (e.g., Mena 2010; Noblecilla 2020). Unfortunately, there are no complete studies on the diversity of San Martín to determine the rate of loss. The lack of knowledge of the mammal diversity in this region is not due to geographical distance or difficulty of access, as in other Amazonian regions, as the Marginal de la Selva highway was built more than 50 years ago. We found that the lack of inventories is mainly due to the period of violence that occurred in Peru during the 1980s, which lasted approximately 20 years, and later the region remained dangerous by the increase of illegal cultivation of coca for drug trafficking and the associated narcoterrorism (CVR 2004). This complicated situation made the region relatively inaccessible to researchers, with few biological inventories as a result.
Although our inventory was carried out in the dry season, the information about small mammals in the SDTF of the Huallaga river basin is still valuable and an essential piece for the conservation of the biodiversity in this region. However, we recommend additional surveys in both seasons to assess changes in the diversity due to the seasonality, and to evaluate more localities on both sides of the Huallaga River to determine whether populations differ genetically. We also recommend additional studies in other nearby conservation areas with dry forests to establish conservation corridors to promote gene flow between forest patches.
Finally, the great diversity of small mammals of the SDTF of the Huallaga river basin, the new records, the biogeographic importance of dry ecosystems, and the Huallaga River as a barrier for dispersion of small, mainly terrestrial, mammal species invite further research in this area. Additional research may help to better understand the natural history, population dynamics, and dispersal patterns of small mammals in a forest characterized by high seasonality. Conservation efforts should address the effect of deforestation and fragmentation and propose a conservation plan and urgent decisions to maintain wellpreserved relicts or patches of forests.

Acknowledgements
We thank the Instituto de Investigaciones de la Amazonía Peruana (IIAP), which through Giussepe Gagliardi allowed one of us (DR) to be part of the expedition to the dry forests of San Martín department. We also thank the members of the Asociación de Conservación y Protección Ecológica "El Valle del Biavo", Asociación de Protección Ambiental del Bosque "El Quinillal", Asociación de Conservación y Protección Ecológica "El Bosque del Futuro Ojos de Agua", and Asociación de Conservación y Protección Ecológica "El Incaico" for the use of facilities during fieldwork for this study. We also thank Maribel Taco, Mercedes Molina, Judith Carrasco, and Rocío Bardales for helping in the identification of marsupials and rodents, and Miluska Ramos for help us with cataloging the samples in the MUSM collection. We thank the Ministerio de Agricultura y Riego (MINAGRI) for issuing the research permit outside protected areas with Resolution No. 0068-2015-SERFOR-DGGSPFFS.