Ecology , endemism , and conservation status of birds that collide with glass windows in Monteverde , Costa Rica

Worldwide, billions of birds die annually due to window collisions. Nevertheless, few accounts document bird-window collisions in the Neotropics. In this study, we document species that collided with windows in Monteverde, Costa Rica, and describe their ecological and conservation status. We gathered information from different sources, including data from museum records and accounts by Monteverde residents who participated as “citizen scientists” between May 2014 and December 2017. We conducted carcass searches between March 2015 and February 2016. We classified window-strike species by migratory, forest dependence, trophic guild, weight, abundance, conservation, and endemism status. We registered 103 species striking windows in Monteverde, which includes 98 of 267 species known to occur in three life zones in Monteverde and five not registered in the area. Window strike casualties’ frequencies differed by species, trophic guild and migratory status. Most window victims were residents, small, insectivorous, considered common or fairly common, with declining population trends. The families with the most species represented were Parulidae (14 spp.), Trochilidae (13 spp.), Turdidae (10 spp.), and Tyrannidae (9 spp.). Most species were passerines (Order Passeriformes) (71 spp.). No hawks or vultures were found colliding with buildings. The three species most commonly killed by windows were frugivores: Swainson’s Thrush (Catharus ustulatus), Northern Emerald-Toucanet (Aulacorhynchus prasinus), and Black-faced Solitaire (Myadestes melanops). Among windowkills were five species whose status on the IUCN Red List are Near Threatened and one Vulnerable, including the Resplendent Quetzal (Pharomachrus mocinno) and the Three-wattled Bellbird (Procnias tricarunculatus). Six species are listed as in danger of extinction and four are listed as species with reduced populations by the National System of Conservation Areas for Costa Rica (SINAC). 12 endemic species are strike casualties. The premontane wet forest is the life zone where more species were found (n=64 spp.), followed by the premontane moist forest (n = 49 spp.) and the lower montane wet forest (n = 31 spp.). These findings demonstrate the urgent need for conservation measures to mitigate bird mortality due to window collisions. Promoting use of methods to protect birds from windows should be an important goal for this IBA and the rest of Costa Rica. We also recommend collecting data in order to increase understanding about bird window collisions.


Introduction
Billions of birds are killed annually worldwide by window collisions (Klem, 2015). In the United States, the annual estimated death toll is 365 to 988 million ( Monteverde-Arenal is one of 21 Important Bird Areas (IBA) designated for Costa Rica (Sánchez, Criado, Sánchez, & Sandoval, 2009). Its creation was based on three global criteria: the presence of threatened species, species with restricted range, and restricted biome (Bird Life International, 2017). Monteverde is a well-known tourist destination (Davis, 2009) where urban development has favored the use of panels of glass in the infrastructure which is often within or adjacent to forest areas (Menacho-Odio, 2018). Although Young &McDonald (2014) stated that hundreds of birds die every year due to window collisions in Monteverde, no studies describe the species affected. In this study, we documented species and quantified species-specific characteristics of window casualties in the highly diverse rural Monteverde area, Costa Rica (Young &McDonald, 2014). We include characteristics such as the life zones where they were found, migratory habits, forest dependence, trophic guild, weight, endemism, and conservation status of each species. Our findings provide a baseline for future studies to understand and quantify the full extent of this problem in Monteverde as well as elsewhere in Costa Rica and the Neotropics.

Material and methods
Study area: We conducted this study from August 2014 to December 2017 in Monteverde, located on the Pacific slope of the Tilarán mountain range (10°18' N, 84°49W; elevation 700-1 400 m a.s.l.), Puntarenas, Costa Rica. The category "Monteverde" refers to the region that includes the community area of Monteverde, the Monteverde Cloud Forest Preserve, the Children´s Eternal Rainforest, and areas on both sides of the Continental Divide down to about 700 m elevation (Nadkarni & Wheelwright, 2014) (Fig. 1). Monteverde encompasses seven life zones of the twelve found in Costa Rica (Holdridge, 1967). Our study area includes three life zones (Fogden, 2014): premontane moist forest, premontane wet forest, and lower montane wet forest (Fig. 1). The moisture gradient on the Pacific slope causes a more abrupt variation than on the Caribbean slope, resulting in high species diversity both for plants and birds in each life zone (Kappelle, 2016 (Sandoval, 2014).

Data collection:
We documented window-strike species by using museum records and citizen observations, and by monitoring designated buildings (Klem, 1989). Specifically, in 2014 and 2017, we obtained from the National Museum of Costa Rica and the Museum of Zoology of the University of Costa Rica a list of specimens with tags specifying that they had collided with windows in Monteverde. We also obtained records from residents of Monteverde who submitted specimens of window-killed birds.
Carcass surveys: Buildings designated for carcass surveys were those of organizations dedicated to education, conservation, and research, as well as private homes of residents, because they were willing to collect and share data. These included the Monteverde Institute, Monteverde Friends School, and a private home located in premontane wet forest. In lower montane wet forest we visited the Monteverde Cloud Forest Reserve, the Monteverde Biological Station, and one private building. The private residences were chosen because the residents of one of them reported high window-strike mortality and the other one because it had large panes of glass. All the buildings are one or two stories, 2.5 m to 20 m height, and surrounded by secondary forest and gardens. Another criterion was that the buildings were close enough together for the main researcher (RMMO) to visit during a day in order to do a carcass search. In all cases, we obtained permission from property owners (Fig. 1). From March 2015 to February 2016, RMMO conducted surveys of carcasses eight days per month, with the exception of four days in April and seven days in September and February, for a total of 90 sampling days per buildings. She conducted surveys around 27 buildings including the Monteverde Institute (n = 5 buildings), Monteverde Biological Station (n = 7), Monteverde Cloud Forest Reserve (n = 8), Friends School (n = 5), and two houses belonging to area residents. The protocol consisted in looking for carcasses around each building on a transect of two meters width from the building´s wall (Hager & Cosentino, 2014). She collected all the carcasses, taking pictures of the bird and the window where she judged it had hit. From October 2014 to March 2016, we asked inhabitants of Monteverde and the staff of the collaborating organizations to collect bodies of birds killed by window collisions. We made this request through electronic media, such as the newsletters of the Monteverde Conservation League, the Monteverde Institute, and the Friends School, and through signs posted in the area, and personal visits to community meetings. The carcasses were stored in freezers at the Monteverde Institute, the Monteverde Biological Station, and the Monteverde Cloud Forest Reserve laboratory. Following this study, we delivered most of the bodies to the Natural History Department of the National Museum of Costa Rica. These specimens are registered with the following catalogue numbers: MNCR-O28910, MNCR-O28912, MNCR-O28913, MNCR-O28915, MNCR-O28916, MNCR-O28918, MNCR-O28920, MNCR-O28921, MNCR-O28923, and MNCR-O28927 (G. Alvarado, pers. comm., February 20, 2018).
Photo reports: Through social media, we also received from Monteverde residents photo reports of birds, dead or alive, that collided with windows. Photographs from bird guides were used to identify the species (Kummer, Bayne, & Machtans, 2016). From March 2015 to December 2017, we received and took photographs of birds that had collided with windows at the University of Georgia Costa Rica campus in San Luis Valley, located in premontane moist forest (Fig. 1).
Naturalists: From August 2014 to December 2017, we received specimens of birds that were collected and frozen by local biologists and residents. Each specimen was identified, a photo was taken, and the date and place where it was found were registered.
Expert interviews: From August to November 2014, we interviewed five biologists who have lived in Monteverde for more than ten years. We asked them to identify the species they were aware of striking windows most frequently.
Community survey:  February 2016, with the collaboration of two volunteers, we interviewed 58 Monteverde residents aged 15 and older. One survey was conducted per household. The Monteverde community area was divided into seven conglomerates with an average of 20 houses, and RMMO and two volunteers visited each conglomerate during half of a day, aiming to visit as many houses as possible. The survey questionnaire asked: Has a bird ever died when it hit a window in this house? The people who answered affirmatively were asked: What species have hit the house? A list of species was obtained and the frequency of strike records per species was collected. Additionally, residents were shown eight pictures with names: Green Hermit (Phaethornis guy), Swainson's Thrush (Catharus ustulatus), Three-wattled Bellbird, Resplendent Quetzal (Pharomachrus mocinno), Long-tailed Manakin (Chiroxiphia linearis), White-tipped Dove (Leptotila verreauxi), Lesson's Motmot (Momotus lessonii), and Northern Emerald-Toucanet (Aulacorhynchus prasinus). We asked each person: Which of these birds died colliding with windows? We categorized the answers as "yes", "no", "I don´t know" or "no answer" for each picture. Carcasses collected were compared by age, sex, and collision frequency per month. To determine the species that collided most frequently, we used records from interviews with biologists, specimens collected, and the area surveys.

Number of species:
We recorded 103 species, classified in 25 families and eight orders, that collided with windows in Monteverde (Table 1). 69 % (71 species) belonged to the order Passeriformes. We recorded 98 species (36.7 %) of the 267 species documented and five species not registered in the three life zones by Fogden (2014). The families with the most species represented were Parulidae (14), Trochilidae (13), Turdidae (10), and Tyrannidae (9).

Forest dependence:
We found different levels of forest dependence among the species that collided with windows. Most species (35 %) occur in areas with at least 50 % of forest cover and just five species were classified as highly dependent on mature forest and secondary forest (1 and 1.5) ( Table 3). In lower wet montane forest we found thirteen of 31 (42 %) species that collided as highly dependent on mature and secondary forest (1 and 1.5),13 (54 %) were also found to collide in the premontane wet forest; and only five (21 %) in the premontane moist forest (Table  1). Feeding guilds: Collisions were mostly by insectivore species (57spp) ( Table 4). Only 36 % of the insectivore species documented by Fogden (2014) were found to collide. On the other hand, 64 % of nectarivore and 63 % of frugivore were documented to have collided with glass panes. Insectivores not documented in our study as window-kills include swallows, woodpeckers, and 28 species (76 %) of the Tyrannidae family (Fogden, 2014). Alternatively, the Olive-striped Flycatcher hit frequently in lower montane wet forest. No carnivores, scavengers, omnivores, nor birds that feed on aquatic macroinvertebrates were recorded as window casualties (Table 4). Weight: Most collision species weighed less than 80 grams (84 spp.; 81 %). The smallest species were Canivet's Emerald (Chlorostilbon canivetii) 2.6 g and the Ruby-throated Hummingbird 2.8 g. The heaviest were Gray-headed Chachalaca (Ortalis cinereiceps) 500 g.; Keel-billed Toucan (Ramphastos sulfuratus) 500 g.; Montezuma Oropendola (Psarocolius montezuma) 520 g.; and the Black Guan (Chamaepetes unicolor) 950 g. Most of the birds are small or medium sized. Neither Accipitridae family species documented in the area (n =15, Mdn = 480 g) were found to collide with windows, nor Cathartidae family species (n = 2, Mdn = 1600 g).

Abundance per life zone:
More species collided with windows in premontane wet forest than in premontane moist forest or in lower montane wet forest (Table 5). Nine species were found in the three life zones, 21 only in premontane moist forest, 32 only in premontane wet forest, and 16 only in lower montane wet forest (Table 1). Most victims in all life zones were classified as Common and Fairly Common, but among the affected were also species previously classified as Uncommon, Rare, Vagrant, and others not reported by Fogden (2014) in the life zones (see Table 5). Naturalists. Eighty-one frozen specimens (31 species) collected prior to this research and stored in freezers were shared by naturalists (biologists and people interested in nature) of Monteverde. Date of collection was not always available, but the oldest specimens with information dated from 2008. Twenty-eight specimens were collected in lower montane wet forest, 45 in premontane wet forest and eight did not have information about the place of collection. The most abundant species was Swainson´s Thrush (n=16), followed by Black-faced Solitaire (n=10), Northern Emerald Toucanet (n= 7), and Long tailed Manakin (n=7). For all individuals of all species combined, 59 were males, 33 females, and 60 unknown. For Swainson's Thrush we identified seven females, nine males, and 18 unknown. For Northern Emerald-Toucanets, both sexes were equally represented with six males, six females, and two unknown. For the Black-faced Solitaire we counted seven males, two females, and five unknown. In relation to migratory status, 62 (41 %) were altitudinal migrants, 51 latitudinal migrants (34 %), and 37 residents (25 %). We found an increase in the number of long distance migrants colliding in April 2015 (n = 17) and October 2015 (n = 10) (Fig. 2). Fifteen (88 %) of the long-distance migrants in April were Swainson's Thrush, the other two were Wood Thrush and a Ruby-throated Hummingbird. In October, eight of the migratory species were Swainson's Thrush and the other two were Black-and-white Warbler and Yellow-throated Vireo (Vireo flavifrons). The number of resident species colliding with windows increased in June and July. The number of immature residents collected increased only slightly in June (nine immature, seven mature), but more so in July (11 immature, five mature) and August (seven immature, 0 mature) (Fig. 3.) Despite this, citizen science was crucial for gathering data and also probably increased awareness and knowledge about this subject (Chandler et al, 2017;Cooper et al., 2007). c) A further factor that could have affected bird collisions detectability was the removal of the bird bodies by scavengers, such as coatis (Nasua narica) and domestic cats (Felis domesticus). Also, carcasses of birds that fell into vegetation, or injured birds that moved before dying, would also go undetected (Bracey et al., 2016). We found that most species affected by window collisions in the study area were residents. In this case, incentives, like the Costa Rican government´s Certification for Tourism Sustainability, that promote planting vegetation to attract resident and migratory birds around hotels should be re-evaluated. Simultaneously, methods preventing window collisions should be promoted (Instituto Costarricense de Turismo, 2013). Hummingbird behaviors such as high-speed flight, male territoriality and associated aggression, may also increase the risk of collisions (Graham, 1997). Artificial feeders are common in Monteverde and in many other touristic areas of Costa Rica. In Costa Rica it is illegal to feed wildlife; however, this regulation is not enforced so it is important to communicate to residents and managers of lodges that feeders should be located less than one meter from any glass surface (Klem, 2014 Thus the increasing number of buildings and the preference for large panes of glass in residences and tourist structures (Menacho-Odio, 2018), should be monitored and regulated to reduce avian mortality. It is also necessary to focus attention on species such as the Orange-bellied Trogon, Black Guan, Three-wattled Bellbird, the Golden-browed Chlorophonia and other Neotropical endemic species whose populations are decreasing.
Twenty-one altitudinal migrant species were recorded as casualties in this study. Window collision risk is predicted to increase as birds move to lower more urbanized areas during their migratory passage. The Three-wattled Bellbird, the Black-faced Solitaire, the Resplendent Quetzal, and the Golden-browed Chlorophonia are known window casualties that are altitudinal migrants. Our results document window strikes by two juvenile Three-wattled Bellbirds; one (survived) in Los Llanos and another (died) in Monteverde (premontane wet forest), during June and July 2015 (Fig.  1). An average of 90 Three-wattled Bellbirds were counted in the last six years in Monteverde (Sandoval, 2014), making every individual killed in a window a significant loss to the local population (Debra Hamilton, pers. comm., 2015.). The expected increase in the number of buildings in Monteverde is anticipated to intensify the risk of bird-window collisions, as occurred in the premontane wet forest in our study area. Currently, Costa Rica Wildlife Conservation Law (SINAC, 2017) does not include any regulation for preventing bird-window related deaths. Moreover, there are no regulations in Monteverde to prevent an increase in urbanization. Consequently, window-kills are expected to become an increasing threat to the local avifauna. Education and research for collision prevention will be required for conservation, economic, legal, and ethical planning to avoid the unwanted losses of the birds that more and visitors come to Monteverde to see (Klem, 2014). For example, there is a need to promote the use of effective methods to prevent bird-window collisions and to discourage the use of non-effective large panes of glass on new buildings. Future studies should address the estimated or, ideally, actual abundance of specific species in and around human dwellings, and the use of more rigorous protocols by volunteers, as suggested by Loss, Will, & Marra (2015), to obtain a more comprehensive understanding. It is essential to include the monitoring of hotels and other commercial buildings in future investigations, given risks these specific structures, with their large areas of sheet glass, pose to birds. Bird-window collisions affect a high number of species of birds in Monteverde. Most windowkilled species are residents, some endemics, threatened with declining populations. Seeking a solution to protect birds from windows should be an important goal for this IBA and the rest of Costa Rica.
We suggest that additional research should include a greater proportion of the population of this area in order to increase concern and knowledge about this important conservation issue for birds and people, especially among commercial interests such as hotel owners and others employed in the birding-related industry. Additionally, more effort should be given to areas where high diversities of birds coexist with buildings boasting large surface area in windows. This study reinforces the importance of public participation (citizen science) for achieving a comprehensive understanding of the problem.
Ethical statement: authors declare that they all agree with this publication and made significant contributions; that there is no conflict of interest of any kind; and that we followed all pertinent ethical and legal procedures and requirements. All financial sources are fully and clearly stated in the acknowledgements section. A signed document has been filed in the journal archives.