More Urbanization, Fewer Bats: The Importance of Forest Conservation in Honduras

Urbanization is a phenomenon that results in fragmentation and eventual destruction of forests. Suburbanization is a subset of that same phenomenon in which fragmentation has resulted in the retention of small patches of the original forest and surviving old growth trees. Alternatively, the area surrounding the central city had been cleared for agricultural use and the suburban residents have planted many trees in parks and private property. This fragmentation will of course affect many species of bats, including species of the family Phyllostomidae. In this work, we estimate and compare the diversity of phyllostomid bats in three landscapes in Honduras: forests, suburban, and urban areas, from 2015 to 2018. Concurrently, we compared bat activity patterns based on the hour and percentage of moonlight at the time they were captured, and we compared external measurements, forearm and ear length. Urban areas are the least diverse and exhibited the lowest abundance. The forearm and ear length were significantly different only between forests and urban areas. The degree of lunar phobia also differed among those landscapes, but the time of capture did not differ. This is the first attempt to describe the activity patterns of phyllostomids in these studied areas and the effect of urbanization on Honduran bats. As expected, we found that from forests to cities, the diversity and abundance of phyllostomids decreased. However, there are many gaps in our knowledge of how totally or partially urbanized areas are affecting phyllostomid bats in Honduras.


Urbanization, suburbanization, and bats
It is generally agreed that urbanization has had a major negative effect on populations and diversity of native plants and animals. Nevertheless, humans plant a wide variety of vegetation in their urban and suburban areas, thus initiating the food chain on which most animals depend. Although diversity is clearly greatly reduced in the urban setting, at the population level the effect has been extremely heterogeneous across the animal kingdom. Bats (Chiroptera) are an extremely town as a suburban area. In Costa Rica, in a large urban center, but recording only in city parks, the number of bat passes was much greater in the larger parks than in the smaller parks [17]. Overall, those authors found considerably less activity than we consistently found in non-suburban settings [14]. The urban bats identified from the calls were all from the families Vespertilionidae and Molossidae. In another Costa Rican study the author mist-netted in parks in the city, finding a relatively small number (for the netting effort) of bats of the family Phyllostomidae, all of which were very common species that eat fruit and/or nectar [18].

Bat diversity in Honduras
Because of its location on the relatively narrow isthmus of Central America, connecting North and South America, Honduras is home to species typical of South America, others typical of temperate North America, and some that are endemic to Central America and southern Mexico. According to a recent review, [19] there are 113 species of bats currently known from Honduras, and we expect several more species will be added in the future. In Central America, only Costa Rica has more species, with 120 listed [20]. As detailed in the paper cited [19], the bats belong to seven different families, the Emballonuridae (9 species), the Phyllostomidae (59 species), the Mormoopidae (5 species), the Noctilionidae (2 species), The Thyropteridae (1 species), the Natalidae (2 species), the Molossidae (17 species), and the Vespertilionidae (18 species).

Objectives and hypothesis
We lack information demonstrating how urbanization is affecting the diversity of bats in Honduras, Central America. We hypothesize that, on a continuum from forests to cities, the diversity of New World leaf-nosed bats (Phyllostomidae) will be significantly reduced. Therefore, this works aims to estimate and compare the diversity of phyllostomid bats in three landscapes in Honduras: forests, suburban, and urban areas; to determine if the forearm and ear length has any significant effect on species composition of bat assemblages in the three landscapes; and to describe the activity patterns of those assemblages.

Concepts
When defining "urbanization" and "suburbanization", which are processes that are closely related and linked along a continuum, we follow Tammaru et al. [21]. We will consider suburbanization as the expansion of suburbs by the increase of its population from the migration of residents of the central city [21,22]. Thereby, we will refer to the Ciudad-Universitaria of the Universidad Nacional Autónoma de Honduras (CU-UNAH) as the "urban area" in this manuscript. The same authors described suburbanization is the redistribution of a population away from central cities and into suburbs. In this work, we are referring to Sabanagrande as the "suburban area". All other studied areas in this work are considered as "forests" located in the Río Plátano Biosphere Reserve. See below for the description of each studied area.

Studied areas
All the coordinates are given in Table 1 and represented in Figure 1, and each site is described below:

CU-UNAH
Surveys were carried out at the National University Campus' Botanical Garden in Tegucigalpa, capital city of Honduras, in the Department of Francisco Morazán ( Table 1). The ecosystem is a tropical dry forest [23]. The species of Fabaceae, Myrtaceae, and Asteraceae are the most common, including Muntingia calabura, Byrsonima crassifolia, Lonchocarpus sanctuarii and cultivated plants, such as Hibiscus rosa-sinensis, Russelia equisetiformis and Psidium guajava [24].

Taxonomy, mist-netting, and ethical guidelines
We followed the most recent taxonomic checklist of the bats that occur in Honduras [19]. All the bats were captured using mist-nets of 12.5 x 2.5 m with a mesh of 35 mm. We followed Kunz and Kurta [27] for positioning the mist nets according to the vegetation, landscape, bodies and water and phenophases (fruits and flowers) of the plants. Bats were identified and measured (FA = forearm length; E = ear length; BH = body height) using taxonomic keys of Timm, LaVal and Rodriguez [28] Medina-Fitoria [29], and Mora [30]. We quantified the sampling effort by multiplying the area of all of the mist-nets that were opened during each night by the number of hours that remained open [31] in which a total of 47,686.8 m 2 *h was accumulated. All the bats were handled according to the guidelines for the use of wild mammals in research and education [32].

Diversity, landscape, and activity patterns analyses
The diversity of each landscape was measured using the Alpha Diversity Index (following Jost [33] and Moreno et al. [34]), and species richness was estimated with Chao 1. These analyses were based in the sampling effort of each site and the abundances of each species using EstimateSMac 910 with 100 randomizations [35,36].
For the activity patterns analyses we used the abundances of all the species [37] and correlated them with the time and the percentage of the moon illumination in which bats were captured from each type of landscape. Moon illumination was taken for each date from the following website: https://www.moongiant.com/ [38]. We used the Shapiro-Wilk test to test for the normal distribution and Levene's test to test for the homogeneity of variances of the data. Considering that data was normally distributed, means of the forearm length, ear length, body height, time, and moon percentage were represented by the ANOVA (Analysis of Variance) analyses in Table 2. To compare means we performed posthoc Tukey tests at a confidence level of 95%. For all the analyses we considered significant differences when p ≤ 0.05. Except for time, in which we use the Spearman's correlation coefficient to determine any relation between the landscapes and the time activity.
To determine taxonomic similarities between the landscapes (urban and suburban areas and forest) we performed multiple regressions of distance matrices [39]. In addition, to represent graphically the taxonomic composition in the distinct habitat types, we performed a NMDS (non-metric multi-dimensional scaling) analysis with two dimensions and plotted the NMDS axes against landscapes [40]. All analyses were performed in R Core Team [41] version 3.4.2, using the vegan [42] and ecodist [43].

Alpha diversity in three landscapes
According to Chao 1 ( Table 2), urban areas (percentage of how many species According to Chao 1, urban areas (percentage of how many species were recorded in parentheses) are the least diverse, as expected, because only 7.5 species are expected (93.3%) followed by suburban areas (87.2%) with 19.24, and then by forests with 28.97 (82.2%). Supporting Chao 1, the Alpha diversity index was highest in forests with 5.72, followed by suburban and urban areas ( Table 2), in that order. Considering the sampling effort, urban areas are not only the least diverse but also the least abundant based on number of bats captured, followed by suburban areas and forests ( Table 2). Even though we found three different assemblages (Figure 2) we found no significant correlation between taxonomical α-diversity and the type of landscape (R 2 = 0.04; DF = 1,24; P = 0.24). However, we found species such as Artibeus jamaicensis, A. lituratus, Glossophaga soricina and Sturnira parvidens that were recorded in all three landscapes. But there were certain species that were recorded only in certain landscapes, for example, Glossophaga leachii in urban areas, Lonchorhina aurita in suburban areas, and Tonatia bakeri and Vampyressa thyone on forests (Table 3; Figure 3).

Figure 2.
Non-metric multi-dimensional scaling (NMDS) of three landscapes which represents three different bat assemblages.

Activity patterns vs. morphological traits
Considering time (Figure 4), we found no significant correlations in any type of landscape: forests and suburban areas (R 2 = 0.00; DF = 1,154; P = 0.23), forests and urban areas (R 2 = À0.01; DF = 1,139; P = 0.62), and urban and suburban areas (R 2 = À0.00; DF = 1,139; P = 0.99). In the case of body height means, only the suburban areas and forests have no significant differences (F (2,203) = 2.3, p = 0.21; Figure 5A), and were divided into two groups urban areas (a) and forests and subruban areas (b). See Table 4 to see the other p values of this and other analyses. The ear length means were only significant different in suburban and forests landscapes (F (2,165) = 4.57, p = 0.05; Figure 5B), however, all the landscapes were assigned to the same group (a). When comparing forearm length means with posthoc Tukey tests, landscapes were classified into two groups urban (a) and suburban and forests (a) and only the comparison between suburban areas and forests was not significant (F (2,431) = 21.41, p = 0.99; Figure 5C). Finally, the moon  percentage mean in which bats were captured was significant different among all the landscapes (Figure 5D), categorized into three different groups a (forest), b (urban areas), and c (suburban areas).

Discussion
As expected in Honduras, there is a consistent decrease of phyllostomid bat diversity and abundance from forests to cities. As anticipated, we found that the  diversity is less in urban areas (cities) and suburban areas in comparison to forests. However, these remnants of forest are important for bat conservation in urban areas. For example, the high abundance of Artibeus and Glossophaga in CU-UNAH are, hopefully, helping in the regeneration of trees by seed dispersal and pollination [44] in this small remnant of forest in the capital city of Honduras. On the other hand, suburban areas may also have important species. For instance, Sabanagrande may be the most important area for the conservation of Artibeus inopinatus [25,45], which is considered to be Data Deficient (DD) by the IUCN (International Union for the Conservation of Nature) [46]. That suburb also houses species such as Micronycteris microtis which is more typical of forest remnants [47]. In contrast to urban and suburban areas, forests clearly have the highest diversity and abundance of phyllostomid bats. Unfortunately, except in protected areas, these forests are being constantly fragmented and with time, they tend to become agricultural or urban landscapes. Species that cannot acclimatize or adapt to these anthropogenically modified landscapes will disappear. Those that occupy very limited geographic distributions likely will become extinct. Olivier et al. [48] point out that two major drivers of habit degradation are urbanization and agricultural intensification which decrease community stability, including bats.

How urbanization is affecting bat diversity
Urbanization is the second most detrimental anthropogenic agent of landscape change [49], since bat diversity and species abundance are comparatively lower in cities than in primary forests or rural areas [3]. This is the case not only for Honduras, but worldwide. For example, in Poland, urbanization pressure is a common phenomenon in several protected areas due to the dispersion of buildings and the expansion of summer construction [49]. Additionally, artificial lighting and sound pollution can alter commuting processes in foraging bats, especially sound which has a more deterrent effect for bats than light as some insectivorous bats feed on the insects that are attracted to streetlights [50][51][52]. Interestingly, habitat degradation affects the diversity of bat communities in more complex ways than simply population stability [48].
Bat response and sensitivity to urbanization varies among species assemblages in urbanized landscapes. In this way species with high tolerance become more abundant and dominant. However, the low diversity and abundance of urban bat fauna can be attributed, at least partially, to a shortage of roosting sites [53]. For a better understanding of bats that do not fly below canopy in urban areas, acoustic monitoring can provide data for species that are rarely captured in mist nets ( [17], and see introduction, this chapter). Unfortunately, we have little such data for Honduras as of now.
The fact that forearm length was only significantly different between forests and urban areas and the ear length between suburban areas and forests can be explained from two points of view. First, we found that Sabanagrande has a mixture of the other two assemblages, and even has species that were found only there (e.g., L. aurita). Additionally, Sabanagrande has a mixture of ecological characters, and even though it is becoming urbanized, there are still important forest remnants. Similar to Mexico [47], we found that in suburban areas (referred to as intermediate disturbance sites), the predominant species was Carollia perspicillata, (subfamily Carollinae). Our results support the hypothesis of Medellín et al. [47] in which phyllostomines (subfamily Phyllostominae) disappear from disturbed areas because of their specific requirements (e.g., foraging activity; [54]) because all the phyllostomines recorded in this study were in forest remnants.
Secondly, the functional traits varied. For example, the well-conserved forests of the RPBR have larger species that were only recorded there (e.g., T. bakeri) but in less abundance. In these species their ear length and wingspan are relatively larger, in contrast to those of the large species Artibeus lituratus, found in CU-UNAH and Sabanagrande. Ramírez-Mejía et al. [55] demonstrated that the most dominant functional group of bats in their studies were those with intermediate values of body mass and wing morphology, which represents the phyllostomids' adaptive response to landscape degradation.
Species such as Phyllostomus hastatus, which is widespread in Honduras, have a larger forearm length and therefore a larger wingspan. Additionally, these species and others of the subfamily Micronycterinae, Lonchorhininae, and Glyphonycterinae, have larger ears because they are gleaners and thus ear length is related to their dietary habits, and these gleaners are more susceptible to the gradual degradation of remnant forests and effects of urbanization (e.g., sound and light pollution) [47]. In other words, phyllostomid gleaners (e.g., phyllostomines) and related subfamilies are indicators of relatively undisturbed rainforests [47,56]. However, a few species, notably Micronycteris microtis, for example, may tolerate moderate levels of urbanization. More ecological data is needed to support this hypothesis.

From cities to forests: Activity patterns of phyllostomids in Honduras
We hypothesize that New World leaf-nosed bats in forests are more likely to be negatively affected by brightness of the moon because of safety concerns (hunting activities by visually oriented predators like owls) when the moon is brighter [15]. In contrast, urban and suburban areas have equally high light intensity every night (e.g., traffic lights, streetlights, shopping centers, etc.). Another feature that supports our hypothesis is that we found significant difference between all the areas. This is probably because the light intensity of suburban areas is increasing in the same way as in urban areas, and the bats that survive there are able to acclimatize quickly. However, the time patterns were not significantly different among the three landscapes due to the wide range and different foraging behaviors. In general, phyllostomids have an early activity peak and then declining activity through the night [57]. Habitat specialization, nutrient intake, and food procurement are features that are associated with bat success in transformed landscapes [55,58].
There are two more works describing activity patterns in Honduras. The first one, Medina-Fitoria et al. [59] studied certain areas included in the RPBR, and determined that in the Caribbean slope of Nicaragua and Honduras, mature and intact primary forests are the most important habitats to conserve. They also determined that fragmentation due to extensive cattle farming and agriculture is perhaps the major threat to these forests. And the second study, in the northwestern region of Honduras, in Cusuco National Park, by Medina-Berkum et al. [60] indicated that the presence of Chrotopterus auritus and Trachops cirrhosus only in the core zone of the park is probably because the core zone still comprises intact and closed forest habitats that may provide a higher abundance of prey for them, and that these species are generally more abundant in undisturbed areas. Primary and pristine forests are clearly among the most important remaining habitats for bat conservation in Honduras, considering that many phyllostomids depend on this type of forest. For example, the core zone and areas nearby of the RPBR are the only ones left in which Ectophylla alba may be found in Honduras due to the requirements of the species [61]. (Even in 1967 that was the only habitat in which one of us, RKL, captured Ectophylla alba).
Although we predicted that from forests to cities, the diversity of phyllostomids will decrease, this is the first attempt to describe their activity patterns in these areas in Honduras. Considering the extension of forests, Duarte et al. [62] mentioned that 48% of the Honduran territory is covered by forests. With the high rate in which they are being diminished is approximately 23,303.56 hectares per year [63], the probability of losing bat species in Honduras is all too real. On the other hand, there are some species that have been adapting very well, as is the case of A. jamaicensis and A. lituratus. These are species that were recorded in all three landscapes we studied and were the most common in urban areas. And even if they are considered as tolerant species, they have an important role in urban and suburban area as seed dispersers [64], and probably they should be considered as the phyllostomid species most tolerant to urbanization in Honduras. It appears that species of Sturnira and Glossophaga are also adapting well to this phenomenon.

Conclusion
Undoubtedly, the RPBR is one of the most important regions in Honduras, and probably in Central America, for bat conservation due to the large extensions of pristine forests and the limited occurrence and abundance of certain species (e.g., Chiroderma gorgasi) in that region. Unfortunately, even this region will be subject to the effects of encroaching clearing for agriculture and urbanization, if inadequately protected. Indeed, some species become acclimatized to urbanization, and some species, most notably A. jamaicensis and A. lituratus, now tolerate higher levels of disturbance than many other species. More studies are needed to determine and explain their tolerance to urbanization. There are other species that have very specific requirements, including intact primary forest, for their survival. (e.g., T. saurophila). This is the first attempt to study how urbanization is affecting a mammalian group in Honduras and is also the first comparison of bat diversity among three different landscapes. Yet, there are many variables that should be analyzed, compared, and described. For example, we recommend measuring light and sound intensity in urbanized areas and comparing them with those of the forests, to determine more specifically how these characteristics are affecting the diversity and abundance of phyllostomid bats in Honduras. Finally, it is still unknown which morphometrical characteristics are important in explaining the adaption of some phyllostomid bat species to urbanized areas. Another factor is the relative amount of fragmentation in the various areas we studied. There are many gaps in our knowledge of how totally or partially urbanized areas are affecting phyllostomid bats in Honduras, and even though there are some similar activity patterns, we can conclude that their diversity and abundance is decreasing in urbanized areas. Considering the increase of urbanization in Honduras plus the high rate of deforestation (approximately 63.85 hectares per day), a conservation plan for Honduran bats is fundamental.
helped us during all this research. Finally, MATC wants to personally thank Marcia