Human activities are changing landscape structure and function globally, affecting wildlife space use, and ultimately increasing human-wildlife conflicts and zoonotic disease spread. Capybaras (Hydrochoerus hydrochaeris) are linked to conflicts in human-modified landscapes (e.g. crop damage, vehicle collision), as well as the spread and amplification of Brazilian spotted fever (BSF), the most human-lethal tick-borne disease in the world. Even though it is essential to understand the link between capybaras, ticks and BSF, many knowledge gaps still exist regarding the effects of human disturbance in capybara space use. Here, we analyzed diurnal and nocturnal habitat selection strategies of capybaras across natural and human-modified landscapes using resource selection functions (RSF). Selection for forested habitats was higher across human-modified landscapes, mainly during day- periods, when compared to natural landscapes. Across natural landscapes, capybaras avoided forests during both day- and night periods. Water was consistently selected across both landscapes, during day- and nighttime. Distance to water was also the most important variable in predicting capybara habitat selection across natural landscapes. Capybaras showed slightly higher preferences for areas near grasses/shrubs across natural landscapes, and distance to grasses/shrubs was the most important variable in predicting capybara habitat selection across human-modified landscapes. Our results demonstrate human-driven variation in habitat selection strategies by capybaras. This behavioral adjustment across human-modified landscapes may be related to increases in A. sculptum density, ultimately affecting BSF.
Citation: Dias TC, Stabach JA, Huang Q, Labruna MB, Leimgruber P, Ferraz KMPMB, et al. (2020) Habitat selection in natural and human-modified landscapes by capybaras (Hydrochoerus hydrochaeris), an important host for Amblyomma sculptum ticks. PLoS ONE 15(8): e0229277. https://doi.org/10.1371/journal.pone.0229277
Editor: Daniel Becker, Montana State University, UNITED STATES
Received: February 1, 2020; Accepted: August 6, 2020; Published: August 20, 2020
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This study received financial support from the Fundação de Amparo a Pesquisa do Estado de São Paulo (www.fapesp.br; FAPESP grants 2015/04795-3 and 2013/18046-7) and the Fundação Parque Zoológico de São Paulo (www.zoologico.com.br). KMPMBF is funded by Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico (CNPq) research grant (www.cnpq.br; processes 308503/2014-7 and 308632/2018-4). Funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
An increasing number of wild species are being forced to adapt to human-modified landscapes and to live within close proximity to humans [1–3]. Across these landscapes, human disturbance has been altering wildlife distribution , behavior , activity , movement , and habitat selection . Mammals, for example, tend to move less and to be more nocturnal in human-modified landscapes [3, 6]. Human influence is also linked to the emergence of almost all zoonosis [8, 9], including tick-borne diseases such as Lyme in the United States , Encephalitis in Europe , and Brazilian spotted fever (BSF) in Brazil . In that context, obtaining accurate data of wild species in human-modified landscapes, mainly those related at some level to human-wildlife conflict and zoonosis epidemiology, is a challenging and crucial goal to wildlife managers and public health institutions.
Capybaras (Hydrochoerus hydrochaeris), the largest living rodents on the planet , are distributed across all South American countries, except for Chile . These semi-aquatic grazing mammals are usually found in habitats with arrangements of water sources, forest patches and open areas dominated by grasses [12, 13]. Water is a key resource to capybaras, used for thermoregulation, mate and predator avoidance [12, 14]. Forests provide shelter from the day heat, and a resting place at night . Low herbaceous plants are the main components of capybaras diet , and the species has been recorded grazing in open areas , where these plants are abundant. Capybaras also show daily variation in habitat use , feeding mainly during the day in the Brazilian Pantanal  and during the night across human-modified landscapes .
Benefited by the great abundance of high-quality food resources from agricultural crops and reduced presence of large predators, capybara populations have recently experienced rapid growth in human-modified landscapes over the last few decades [12, 18, 19]. Over some regions, large populations of capybaras are linked to increased crop damage , increased vehicle collisions , and the spread of Brazilian spotted fever (BSF)—the most human-lethal spotted fever rickettsiosis in the world . Capybaras are responsible for maintaining and carrying large numbers of Amblyomma sculptum ticks, the natural reservoir and main vector of the bacterium Rickettsia rickettsii, the etiological agent of BSF . Capybaras can also act as amplifying hosts of R. rickettsii among A. sculptum populations [10, 22].
The role of vertebrate-amplifying hosts in sustaining R. rickettsii populations has been well-discussed, with results showing that A. sculptum is unable to sustain the bacterium by itself over consecutive generations [10, 23]. In the Brazilian Cerrado, previous research showed that this tick species is more abundant in forested habitats (cerradão and gallery forests) than in open fields or seasonally flooded habitats [24, 25]. In this context, understanding how capybaras select their habitats across landscapes with different levels of anthropogenic disturbance and vegetation cover (open field versus forests) may have important implications for the ecological relationships between capybaras, ticks, and consequently, BSF.
In this study, we investigated and quantified the variation in diurnal and nocturnal habitat selection strategies by GPS-tracked capybaras across natural and human-modified landscapes. We tested the prediction that capybaras show daily variation in habitat selection preferences across landscapes with different levels of human disturbance, increasing their selection for forests and water sources during daytime periods in human-modified landscapes, to avoid humans.
Capybara field capture was authorized by the Brazilian Ministry of the Environment (permit SISBIO No. 43259–6), by the São Paulo Forestry Institute (Cotec permit 260108–000.409/2015), and approved by the Institutional Animal Care and Use Committee of the Faculty of Veterinary Medicine of the University of São Paulo (protocol 5948070314).
Capybaras were tracked in natural landscapes of Mato Grosso and Mato Grosso do Sul states and across human-modified landscapes of São Paulo state (Fig 1). To assess the level of human disturbance at our study sites, we incorporated the Human Footprint Index (HFI) developed by Venter et al. . This index provides a global map of human pressure in the environment, being useful to assess locations under high levels of human disturbance or areas more likely to be in a natural state . HFI ranges from 0 (natural landscapes) to 50 (high-density built landscapes) and the spatial resolution of the global dataset is 1-km.
We tracked capybaras from four groups in the Brazilian Pantanal (natural landscapes; green color), located in the states of Mato Grosso and Mato Grosso do Sul, and from seven groups in human-modified landscapes of São Paulo state (red color), in the municipalities of Americana, Araras, Piracicaba, Pirassununga, Ribeirão Preto and São Paulo. Land cover layer was downloaded from Project MapBiomas . Brazilian states shapefile was downloaded from IBGE . South America shapefile was downloaded from Orogénesis Soluciones Geográficas . Geographic Coordinate System: WGS 84 / EPSG 4326.
Study areas in natural landscapes (São José, Ingá, Ipanema and Poconé) were located in the Pantanal biome. The Pantanal is the largest wetland in the world, characterized by a mosaic of upland vegetation and seasonally flooded areas [14, 30]. This biome consists of large areas of natural vegetation and well-structured/stable ecological communities. The Pantanal supports an extraordinary concentration and abundance of wildlife , including a large assemblage of medium and large carnivores [32, 33]. Within the sampled areas of Pantanal, capybaras had no access to crops or exotic grasses.
Unlike natural landscapes, human-modified landscapes in São Paulo state underwent significant land use and cover changes during the second half of the 19th and early 20th century, transforming natural vegetation (Atlantic rainforest and Cerrado biomes) into a mosaic comprised of small forest fragments surrounded by an agro-pastoral matrix . These forest fragments likely experience large edge effects and reduced biodiversity , which affects the abundance of medium and large carnivores across the region. Jaguar (Panthera onca), puma (Puma concolor), anacondas (Eunectes spp.), and caimans (Caiman spp.) face threats in the state according to the “São Paulo State Redbook of Fauna Threatened by Extinction” .
Across human-modified landscapes, we tracked capybaras in six municipalities: Americana, Araras, Piracicaba, Pirassununga, Ribeirão Preto and São Paulo (Fig 1). With the exception of the municipality of São Paulo, all areas were located in agricultural landscapes. Sugar cane, corn, cultivated pasturelands, and small forest fragments were the dominant landscape components in the study sites. In Ribeirão Preto, the area used by capybaras was surrounded by a fence that prevented animals from accessing agricultural crops, but they did have access to exotic grasses, as it was also the case in the other human-modified landscapes. In São Paulo municipality, capybaras were monitored in Alberto Löfgren State Park, a protected area within a forest/urban matrix with plenty of cultivated grasses.
It is important to emphasize that no case of BSF has been reported in Mato Grosso and Mato Grosso do Sul states, and serological analyses of capybaras from these natural landscapes have shown no evidence of R. rickettsii exposure . In contrast, at least three study areas of human-modified landscapes in São Paulo state were classified as BSF-endemic (municipalities of Americana, Araras and Piracicaba), with recent occurrence of human cases and serological evidence of R. rickettsii infection in capybaras .
Capybara capture and collaring
From 2015 to 2018, we tracked 20 capybaras from 11 groups in Brazil (S1 Table) with Lotek Iridium Track M 2D GPS collars (Lotek Wireless, Haymarket, Ontario, CN). Among these, four capybaras were tracked from four groups in natural landscapes, and 16 capybaras from seven groups in human-modified landscapes (for more details on tracked individuals see S1 Table). In São José, Ingá and Ipanema ranches (natural landscapes of Brazilian Pantanal in municipality of Corumbá, state of Mato Grosso do Sul), individuals were tranquilized and captured with the aid of a pneumatic rifle (Dan-Inject model JM Standard, Denmark). We used a mixture of ketamine (10 mg/kg) and xylazine (0.2 mg/kg) to anesthetize captured animals . As capybaras use water , we targeted animals at a large distance (>20m) from this resource to reduce risk of drowning during tranquilization and capture. Across all other study areas, we captured capybaras through corral-type traps, following the methodology in Pereira & Eston .
To better understand movement of capybara populations and minimize the mortality risk of tracked animals, we focused GPS collaring entirely on females. Females show lower agonistic interaction rates when compared to males  and therefore, have a decreased chance of mortality. Most female capybara are found in social groups [17, 41] and are thought to be philopatric . We targeted the largest females within each group for GPS collaring because there is a significant correlation between weight and hierarchical position . Hence, we assumed that dominant female movement provided the best representation of group movement.
To avoid incorporating geolocations with large spatial errors , we removed GPS positions with a Dilution of Precision (DOP) > 9, following recommendations in Lotek’s GPS collaring manual (Lotek Wireless, Haymarket, Ontario, CN.). The day of capture was removed from analyses to reduce bias in space use related to capture-induced stress . Individuals with < 100 data points were also removed. Original GPS-data were collected every 1 or 2-hours during the first 30–40 days, and collars were reprogrammed to collect data every 4-hours and 17 minutes thereafter. GPS-data were rarified until they reached minimum time intervals of 4-hours. Data were categorized into diurnal and nocturnal according to sunrise and sunset time using the ‘maptools’ package  in the R statistical environment . If a given GPS location was collected between sunrise and sunset, it was classified as diurnal. If collected between sunset and sunrise, the GPS location was classified as nocturnal.
To assess the level of human disturbance at each study site, and consequently justify the partition of areas into natural and human-modified landscapes, we calculated the mean GPS-data coordinate of each tracked individual and created a buffer around it with radius equal to the mean dispersal distance of capybaras from their groups (3.4-km) . We then merged buffers of individuals tracked from the same group and calculated the mean HFI within them. These operations were conducted using QGIS 2.18.9 . Across natural landscapes, mean HFI ranged from 2.4 to 6.8 (; n = 4), and in human-modified landscapes mean HFI ranged from 17.4 to 37.7 (; n = 7).
To generate covariate data for our habitat selection analysis, we performed a supervised land cover classification using Random Forests, an ensemble learning method common for classifying satellite imagery . We used multispectral high-resolution imagery (2-m resolution) acquired by the WorldView-2 satellite (DigitalGlobe, Inc.) and ancillary data derived from each satellite scene for classification (Table A in S1 Appendix). We established four habitat classes across natural landscapes (forest, water, grasses/shrubs, bare soil) and five in human-modified landscapes (we added a settlements/roads class). The land cover classification was performed using the ‘RStoolbox’ package  in the R statistical environment .
We digitized 1531 training polygons in QGIS 2.18.9  based on visual interpretation of Worldview-2 satellite scenes. Polygons were divided into calibration (70%; used as input for the land cover classification) and validation (30%; used to evaluate the classification). Overall accuracy ranged from 0.95 to 1 in natural landscapes (; n = 3) and from 0.84 to 0.99 in human-modified landscapes (; n = 6). We also applied a post-classification filter to reduce ‘salt-and-pepper’ noise generated by per-pixel classifiers . More details on the land cover classification can be found in S1 Appendix.
For each study area, we calculated the Normalized Difference Vegetation Index (NDVI) , and created a binary classification of three habitat layers with ecological relevance to capybaras: forest, water and grasses/shrubs. Forest layers included all the types of forested vegetation, primary or secondary, native or not. Water layers included lakes, ponds, and rivers. Grasses/shrubs layers included native and exotic underbrush and shrubby vegetation, including pasturelands, and agricultural crops.
Using binary habitat classifications, we generated distance layers and calculated the shortest distance between each capybara tracking location and habitat classes. For forest distance calculations, we excluded 50-m from the forest edge to assess selection for areas into the forest interior and edges as well. Large double-lane highways found at some of our study sites (varying from 32 to 44 m width: Rodovia Ernesto Paterniani, Rodovia Luis de Queiroz and Rodovia Anhanguera) likely present barriers to capybara’s movement. Because tracked animals did not cross highways during our study, habitats located beyond these highways were not included in our models. Distance to forest interior, distance to water, distance to grasses/shrubs, and NDVI were used as input parameters for resource selection models.
Resource selection functions
We evaluated habitat selection by comparing the use and availability of habitats through a fine-scale third/fourth-order  resource selection function (RSF) analysis . Day and nighttime periods were analyzed separately, due to recognition that capybara habitat use varies throughout the circadian cycle . Habitat availability was determined using a set of random points generated within a buffer around each “use” point (GPS-data) [7, 55]. Buffers were generated with radius sizes equal to the maximum step length displaced by each animal over a time interval equal to our GPS-data resolution (approximately 4-hours). Therefore, each capybara had a unique set of buffers created using its maximum step length in which random points were generated to calculate habitat availability.
To determine the appropriate number of random points per ‘use’ point (GPS-data), we performed a sensitivity analysis following details described by previous works [7, 55]. We randomly selected one individual from each study area and fit multiple logistic regression models across several possibilities (1, 2, 3, 5, 10, 20, 30 and 50) of random points. We repeated the process 100 times and calculated the expectation of the coefficient estimates and the 95% simulation envelopes. We determined that a sample of 30 availability points per ‘use’ point provided stable coefficient estimates (Fig A in S2 Appendix). The analysis was performed in R .
We included habitat variables in our RSF after determining that they were not highly correlated (Pearson’s r > 0.65). To facilitate comparisons across landscapes and across time periods, we scaled and centered all data layers . We included quadratic terms for all habitat variables to test for non-linear relationships. Habitat selection was modeled applying a generalized linear mixed-effects logistic regression, following the equation: (1)
Where ω(xi) is the RSF, βn is the coefficient for the nth predictor habitat variable xn, and γ is the random intercept for the animal i. We incorporated random effects into the model structure to better account for differences between individuals, while also accounting for unbalanced sampling designs . We used nested random effects (“individual” inside “study area” inside “landscape”) to evaluate landscape-level coefficients. A hierarchical approach was used to account for non-independence between individual movements . Habitat selection was modelled using the ‘lme4’ package  in R .
We created four candidate models (forest, water, open areas and full) for each landscape and time-period (Table 1) and used Akaike’s Information Criterion (AIC) to rank them . Models were created to evaluate the importance of different resources on capybara habitat selection: (1) forest—providing shelter from daytime heat and a resting place during the night ; (2) water—used by capybaras for thermoregulation, mating and as a refuge from predator attacks ; and (3) open areas—used for grazing to meet energy demands . A fourth model, inclusive of all variables, was tested to evaluate if a combination of factors most influenced capybara habitat selection.
We compared all models to a null model using chi-squared tests in R . Coefficients of top-ranked models with confidence intervals that overlap zero were considered statistically insignificant. Top-ranking models were evaluated following the technique in , applying Spearman rank correlations between area adjusted frequencies, using presence-only validation predictions and RSF bins (S3 Appendix).
Capybara capture and collaring
A total of 20 capybaras were captured and fitted with GPS collars. Capybaras were monitored for 33 to 918 days (), with a similar number of positions collected across study areas (S1 Table). Average fix success was high for both landscapes, ranging from 87% to 99% in natural landscapes () and from 94% to 99% in human-modified landscapes (). Maximum distance displaced by individuals in 4-hour time interval ranged from 442-m to 1437-m across natural landscapes () and 268-m to 2703-m in human-modified landscapes ().
Natural landscapes’ models
The full model was top-ranked across day- and nighttime periods in natural landscapes, indicating that all habitat variables were important in predicting capybara habitat selection (Table 2). Cross-validation highlighted a strong fit to our data (Table A in S3 Appendix), with stronger results for daytime periods (day average rs = 0.83; night average rs = 0.69). In natural landscapes, distance to water was the most important variable predicting capybara habitat selection (Table 3), with higher coefficient during nighttime periods (day: β = −1.52±0.03; night: β = −1.91±0.03; Table 3). NDVI was a weak variable in predicting capybara habitat selection during day periods and was not significant during nighttime periods (day: β = 0.21±0.02; night: β = 0±0.02; Table 3).
Capybaras selected areas further from forest interiors in natural landscapes (Fig 2), with highest probabilities of selection found in areas >250-m from the forest centroid (day- and nighttime periods). Capybaras displayed strong preferences for areas near water. This trend was consistent across day- and nighttime periods (Fig 2), with the probability of selection declining with increasing distance. Preferences for areas near open areas, dominated by grasses/shrubs, were also recorded, with probability of selection decreasing sharply at short distances (Fig 3). Probability of selection by capybaras increased with increasing NDVI during day- and nighttime periods, although the relatively probability of selection plateaued at a NDVI value of approximately 0.5 during nighttime periods.
The y axis represents the relative probability of selection, ranging from 0 to 1. The x axis represents distance to the habitat. Negative values of forest graphs are related to areas into the forest interior (-50m represents areas 50m inside forest patches).
Human-modified landscapes’ models
Across human-modified landscapes, the full model was also top-ranked for both day- and nighttime periods (Table 4). Models strongly fit the data in these landscapes (day average rs = 0.89; night average rs = 0.72), with weaker results found in São Paulo municipality during nighttime, where capybaras were tracked in a non-agricultural state park (Table A in S3 Appendix). The most important variable in predicting capybara habitat selection for day- and nighttime periods was distance to grasses/shrubs (day: β = 1.03±0.03; night: β = 0.57±0.03; Table 3). Distance to water (day: β = −0.84±0.02; night: β = −0.46±0.02; Table 3) and distance to forest interior (day: β = −0.83±0.04; night: β = −0.08±0.03; Table 3) were also significant in predicting capybara habitat selection, with stronger coefficients found for daytime periods. NDVI was a weaker variable in predicting capybara habitat selection during daytime periods, when compared to other habitat variables, and was not significant during nighttime periods (day: β = 0.32±0.02; night: β = 0±0.02; Table 3).
Contrasting to natural landscapes, capybaras across human-modified landscapes were observed with higher preferences for forest interior areas and areas close to forests, with probability of selection declining with increasing distance to forested habitats (Fig 2). Capybaras also showed preferences for areas near water sources, with higher selection during the day (Fig 2). Lower preferences for areas close to grasses/shrubs were found for human-modified landscapes when compared to natural landscapes, with selection increasing at mid distances (125-m) and declining at larger distances (250-m; Fig 3). Similar to natural landscapes, the relative probability of selection increased with increasing NDVI values during daytime periods (maximum coefficients at NDVI values close to 0.7). For nighttime periods, the relative probability of selection peaked at a NDVI value close to 0.5.
This is the first study using GPS tracking, high-resolution imagery and resource selection functions (RSF) to analyze and quantify capybara habitat selection strategies across natural and human-modified landscapes. Capybaras strongly selected forested habitats across human-modified landscapes during daytime periods, whereas selection for forests was weak across both day- and nighttime in natural landscapes. This pattern of forest selection in human-modified landscapes may be a direct response to human activities (e.g. agricultural machinery, people and vehicle traffic), which are more intense in open areas of our study sites during daytime periods. As wildlife respond to human disturbance following the same principles used by prey encountering predators , capybaras may increase their selection for forests during daytime to avoid contact with humans. Indeed, other studies have suggested that forest cover may provide protection for capybaras from hunting , and capybara groups were observed seeking shelter in forests when humans approached (personal observation). Also, distance to the nearest riparian forest patch had a great influence in capybara habitat selection across human-modified landscapes of the Colombian Llanos .
The high selection for forests by capybaras across human-modified landscapes may put these amplifying hosts in closer contact with A. sculptum ticks, the main vector for the BSF agent, R. rickettsii , since degraded forests are the preferred habitats of A. sculptum ticks [24, 25]. A parallel study that evaluated same capybara groups of the present study reported an overall mean abundance of A. sculptum ticks on capybaras significantly higher across human-modified landscapes than in natural areas . In addition, the environmental density of all host-questing stages of A. sculptum (larvae, nymphs and adults) was also significantly higher across human-modified landscapes than in natural landscapes . Therefore, capybaras may be highly efficient hosts across human-modified landscapes, increasing their already described capacity in maintaining and carrying large numbers of A. sculptum , due to shared preferences for forested habitats with this tick [24, 25].
The ecological relationships between capybaras and A. sculptum are a key point in BSF epidemiology, since A. sculptum populations are not able to sustain R. rickettsii for successive generations without vertebrate-amplifying hosts [62, 63], among which capybaras stands out . Capybaras are linked to the amplification of rickettsial infection among A. sculptum populations, creating new cohorts of infected ticks during bacteremia periods (days or weeks), when they maintain R. rickettsii in their bloodstream . Consequently, minimizing the exposure of capybaras to A. sculptum reduce the populations of this tick, since capybaras are major hosts for A. sculptum . Actions resulting in a drastic reduction of A. sculptum populations across our study areas are likely to limit R. rickettsii infection from tick populations, preventing new BSF cases .
Preferences for areas nearby water sources across natural and human-modified landscapes were not surprising. Capybaras are semi-aquatic mammals and their dependence on water sources has already been well-documented, with some authors reporting these rodents hardly moving more than 500-m from water [61, 64, 65]. However, our models highlighted that capybaras were less dependent on water sources in human-modified landscapes, which may be related to human-driven variation in one or more behaviors linked to water use: reproduction, thermoregulation, or predator avoidance .
Quality and quantity of food resources from highly nutritious agricultural and pasture fields seems to have a strong influence on habitat selection by capybaras, since grasses/shrubs was the strongest variable in our human-modified landscapes’ models. Because we wanted to compare selection for similar habitats across natural and human-modified landscapes, we did not separate crops and pastures into individual habitat classes. However, in the future, more detailed habitat selection studies for capybaras might consider fine-scale spatiotemporal dynamics of agriculture and pasture fields in human-modified landscapes. Understanding selection for these resources, mainly sugar cane, which is linked to BSF spread , may be essential to develop conflict mitigation strategies for the species.
Lastly, improving NDVI temporal resolution could potentially increase the link between this vegetation index and capybaras, since this variable was weak in predicting capybara habitat selection. Higher temporal resolution of NDVI may allow for further investigations on the interaction between vegetation quality and capybara habitat use.
Despite the small number of studied animals in the Brazilian Pantanal, capybaras in this area were tracked for relatively long periods with high numbers of GPS-locations, which increases data reliability. Our results showed clear distinctions between habitat selection of capybaras in natural and human-modified landscapes, providing a background for further investigation into the potential indirect effects of human disturbance in capybara space use. The development of knowledge regarding these effects may assist future management actions aimed at reducing conflicts linked to the species, and the exposure of capybaras to A. sculptum ticks.
Through the use of GPS tracking and resource selection functions it was possible to demonstrate variation in habitat selection strategies of capybaras across natural and human-modified landscapes. Areas close to forested habitats were more selected with higher levels of probability across human-modified landscapes than across natural landscapes. In addition, capybaras consistently selected areas near water in both landscapes, but this resource was more important in predicting capybara habitat selection in natural landscapes. In contrast, grasses/shrubs (which includes crops and pasture fields) was a stronger predictor of capybara habitat selection across human-modified landscapes. Our results show the influence of anthropic disturbance in capybara space use patterns and indicate that an increased understanding of capybara habitat use in natural and human-modified landscapes may support improved human-wildlife conflict management.
S1 Appendix. Land cover classification of capybara habitats.
Methods on how habitats of studied capybaras were classified using high-resolution satellite imagery and random forest algorithm.
S2 Appendix. Sensitivity analysis performed for study areas across natural and human-modified landscapes.
We performed sensitivity analysis to set the number of random points per ‘use’ point to our habitat selection models.
S3 Appendix. Top-ranked models’ evaluation.
We used presence-only data to evaluate our top-ranked models’ performance through Spearman rank correlations between area-adjusted frequencies and resource selection functions spatial bins.
S4 Appendix. GPS-locations of tracked capybara groups in natural landscapes of the Brazilian Pantanal and human-modified landscapes of São Paulo state, Brazil.
We plotted diurnal and nocturnal capybara locations at each study area over WorldView-2 satellite imagery.
S1 Table. Summary table for GPS-tracked capybaras across natural and human-modified landscapes.
We are grateful to SESC Pantanal, UFMT, Embrapa Pantanal and Alegria and São José ranches (Corumbá) for logistical support during field work in the Brazilian Pantanal and to the “Departamento de Água e Esgoto de Americana (DAE)” for allowing us to work at the “Estação de Tratamento de Esgoto (ETE) de Carioba” in Americana municipality. Forestry Institute and Coordination of Urban Parks of the São Paulo State Secretariat for the Environment offered logistical support in captures and monitoring (SMA Process 000.409/2015). We are also very grateful to GIS Lab at Smithsonian Institution, VA, USA, for the technical training in GIS and spatial analysis.
- 1. Kokko H, López-Sepulcre A. From individual dispersal to species ranges: perspectives for a changing world. Science. 2006; 313: 789–791. pmid:16902127
- 2. Tablado Z, Jenni L. Determinants of uncertainty in wildlife responses to human disturbance. Biol Rev. 2015; 92: 216–233. pmid:26467755
- 3. Gaynor KM, Hojnowski CE, Carter NH, Brashares JS. The influence of human disturbance on wildlife nocturnality. Science. 2018; 360: 1232–1235. pmid:29903973
- 4. Paudel PK, Kindlmann P. Human disturbance is a major determinant of wildlife distribution in Himalayan midhill landscapes of Nepal. Anim Conserv. 2012; 15: 283–293.
- 5. Fortin D, Andruskiw M. Behavioral response of free-ranging bison to human disturbance. Wildlife Soc B. 2003; 31: 804–813.
- 6. Tucker MA, Böhning-Gaese K, Fagan WF, Fryxell JM, Van-Moorter B, Alberts SC, et al. Moving in the Anthropocene: Global reductions in terrestrial mammalian movements. Science. 2018; 359: 466–469. pmid:29371471
- 7. Stabach JA, Wittemyer G, Boone RB, Reid RS, Worden JS. Variation in habitat selection by white‐bearded wildebeest across different degrees of human disturbance. Ecosphere. 2016; 7: 1–17.
- 8. Kays R, Crofoot MC, Jetz W, Wikelski M. Terrestrial animal tracking as an eye on life and planet. Science. 2015; 348: 1222–1231.
- 9. McMahon BJ, Morand S, Gray JS. Ecosystem change and zoonoses in the Anthropocene. Zoonoses Public Hlth. 2018; 65: 755–765.
- 10. Labruna MB. Brazilian spotted fever: the role of capybaras. In: Moreira JR, Ferraz KMPMB, Herrera EA, Macdonald DW, editors. Capybara: Biology, Use and Conservation of an Exceptional Neotropical Species. New York: Springer; 2013. pp. 371–383.
- 11. Nowak RM, Walker EP. Walker's Mammals of the World (Vol. 1). 6th ed. Baltimore: Johns Hopkins University Press; 1999.
- 12. Moreira JR, Alvarez MR, Tarifa T, Pacheco V, Taber A, Tirira DG, et al. Taxonomy, natural history and distribution of the capybara. In: Moreira JR, Ferraz KMPMB Herrera EA, Macdonald DW, editors. Capybara: Biology, Use and Conservation of an Exceptional Neotropical Species. New York: Springer; 2013. pp. 3–37.
- 13. Corriale MJ, Herrera EA. Patterns of habitat use and selection by the capybara (Hydrochoerus hydrochaeris): a landscape-scale analysis. Ecol Res. 2014; 29: 191–201.
- 14. Herrera EA, Macdonald DW. Resource utilization and territoriality in group-living capybaras (Hydrochoerus hydrochaeris). J Anim Ecol. 1989; 58: 667–679.
- 15. Alho CJ, Rondon NL. Habitats, population densities, and social structure of capybaras (Hydrochaeris hydrochaeris, Rodentia) in the Pantanal, Brazil. Rev Bras Zool. 1987; 4: 139–149.
- 16. Barreto GR, Quintana RD. Foraging strategies and feeding habits of capybaras. In: Moreira JR, Ferraz KMPMB, Herrera EA, Macdonald DW, editors. Capybara: Biology, Use and Conservation of an Exceptional Neotropical Species. New York: Springer; 2013. pp. 83–96.
- 17. Ferraz KMPMB, Manly B, Verdade LM. The influence of environmental variables on capybara (Hydrochoerus hydrochaeris: Rodentia, Hydrochoeridae) detectability in anthropogenic environments of southeastern Brazil. Popul Ecol. 2009; 52: 263–270.
- 18. Vargas FC, Vargas SC, Moro MEG, Silva V, Carrer CRO. Monitoramento populacional de capivaras (Hydrochoerus hydrochaeris Linnaeus, 1766) em Pirassununga, SP, Brasil. Ciênc Rural. 2007; 37: 1104–1108.
- 19. Ferraz KMPMB, Ferraz SFB, Moreira JR , Couto HTZ, Verdade LM. Capybara (Hydrochoerus hydrochaeris) distribution in agroecosystems: a cross‐scale habitat analysis. J Biogeogr. 2007; 34: 223–230.
- 20. Moreira JR, Pinheiro MS. Capybara production in Brazil: Captive breeding or sustainable management? In: Moreira JR, Ferraz KMPMB, Herrera EA, Macdonald DW, editors. Capybara: Biology, Use and Conservation of an Exceptional Neotropical Species. New York: Springer; 2013. pp. 333–344.
- 21. Abra FD, Granziera BM, Huijser MP, Ferraz KMPMB, Haddad CM, Paolino RM. Pay or prevent? Human safety, costs to society and legal perspectives on animal-vehicle collisions in São Paulo state, Brazil. Plos One. 2019. pmid:30973920
- 22. Souza CE, Moraes-Filho J, Ogrzewalska M, Uchoa FC, Horta MC, Souza SS, et al. Experimental infection of capybaras Hydrochoerus hydrochaeris by Rickettsia rickettsii and evaluation of the transmission of the infection to ticks Amblyomma cajennense. Vet Parasitol. 2009; 161: 116–121. pmid:19147293
- 23. Costa FB, Gerardi M, Binder LDC, Benatti HR, Serpa MCDA, Lopes B, et al. Rickettsia rickettsii (Rickettsiales: Rickettsiaceae) infecting Amblyomma sculptum (Acari: Ixodidae) ticks and capybaras in a Brazilian spotted fever-endemic area of Brazil. J Med Entomol. 2020: 57; 308–311. pmid:31504641
- 24. Barbieri ARM, Szabó MPJ, Costa FB, Martins TM, Soares HS, Pascoli B, et al. Species richness and seasonal dynamics of ticks with notes on rickettsial infection in a Natural Park of the Cerrado biome in Brazil. Ticks Tick-borne Dis. 2019; 10: 442–453. pmid:30611725
- 25. Szabó MP, Castro MB, Ramos HG, Garcia MV, Castagnolli KC, Pinter A, et al. Species diversity and seasonality of free-living ticks (Acari: Ixodidae) in the natural habitat of wild Marsh deer (Blastocerus dichotomus) in Southeastern Brazil. Vet Parasitol. 2007; 143: 147–154. pmid:16962241
- 26. Venter O, Sanderson EW, Magrach A, Allan JR, Beher J, Jones KR, et al. Last of the Wild Project, Version 3 (LWP-3): 2009 Human Footprint, 2018 Release. Palisades, New York: NASA Socioeconomic Data and Applications Center (SEDAC). 2018. https://doi.org/10.7927/H46T0JQ4
- 27. Project MapBiomas. Collection 4.1 of Brazilian Land Cover & Use Map Series. https://mapbiomas.org/en/colecoes-mapbiomas-1?cama_set_language=n [accessed on 2020-07-27].
- 28. Brazilian Institute of Geography and Statistics (IBGE). Brazilian territory organization shapefile (Unidades da Federação–Brasil). https://s0001777a9ca986dd.jimcontent.com/download/version/1431462982/module/5497298759/name/South_America.rar [accessed on 2020-06-15].
- 29. Carlos Efraín Porto Tapiquén. Orogenesis Geographical Solutions. South America shapefile. Porlamar, Venezuela. 2015. http://tapiquen-sig.jimdo.com [accessed on 2020-06-15].
- 30. Silva MPD, Mauro R, Mourao G, Coutinho M. Distribuição e quantificação de classes de vegetação do Pantanal através de levantamento aéreo. Rev Bras Bot. 2000; 23: 143–152.
- 31. Swartz FA. The Pantanal in the 21st century—for the planet’s largest wetland, an uncertain future. In: Swarts FA, editor. The Pantanal of Brazil, Paraguay and Bolivia. Pennsylvania: Hudson MacArthur Publishers. 2000. pp. 1–24.
- 32. Harris MB, Tomas W, Mourao G, Silva CJ, Guimaraes E, Sonoda F, et al. Safeguarding the Pantanal wetlands: threats and conservation initiatives. Conserv Biol. 2005; 19: 714–720.
- 33. Tortato FR, Layme VMG, Crawshaw PG, Izzo TJ. The impact of herd composition and foraging area on livestock predation by big cats in the Pantanal of Brazil. Anim Conserv. 2015; 18: 539–547.
- 34. Metzger JP. Landscape structure changes and species richness in forest fragments of south-east Brazil. C R Acad Sci III. 1998; 4: 319–333.
- 35. Costa-Neto JDB, Maretti CC, Lino CF, Albuquerque JLR. A Reserva da Biosfera da Mata Atlântica no Estado de São Paulo. 2nd ed. São Paulo: Conselho Nacional da Reserva da Biosfera da MataAtlântica; 1996.
- 36. Bressan PM, Kierulff MCM, Sugieda AM. Fauna Ameaçada de Extinção no Estado de São Paulo: Vertebrados. 1st ed. São Paulo: Fundação Parque Zoológico de São Paulo, Secretaria do Meio Ambiente; 2009.
- 37. Luz HR, Costa FB, Benatti HR, Ramos VN, Serpa MCA, Martins TF, et al. Epidemiology of capybara-associated Brazilian spotted fever. Plos Neglect Trop D. 2019. pmid:31490924
- 38. Nishiyama SM, Pompermayer LG, de Lavor MSL, Mata LBSC. Associação cetamina-xilazina, tiletamina-zolazepam e tiletaminazolazepam-levomepromazina na anestesia de capivara (Hydrochoerus hydrochaeris). Ceres. 2006; 53: 406–412.
- 39. Pereira HDFA Eston MR. Biologia e manejo de capivaras (Hydrochoerus hydrochaeris) no parque estadual Alberto Löfgren, São Paulo, Brasil. Rev Inst Florest. 2007; 19: 55–64.
- 40. Herrera EA, Macdonald DW. Aggression, dominance, and mating success among capybara males (Hydrochaeris hydrochaeris). Behav Ecol. 1993; 4: 114–119.
- 41. Garcias FM, Bager A. Estrutura populacional de capivaras na Estação Ecológica do Taim, Brasil, RS. Cienc Rural. 2009; 39: 1–7.
- 42. Herrera EA, Salas V, Congdon ER, Corriale MJ, Tang-Martínez Z. Capybara social structure and dispersal patterns: variations on a theme. J Mammal. 2011; 92: 12–20.
- 43. Lewis JS, Rachlow JL, Garton EO, Vierling LA. Effects of habitat on GPS collar performance: using data screening to reduce location error. J Appl Ecol. 2007; 44: 663–671.
- 44. Morellet N, Verheyden H, Angibault JM, Cargnelutti B, Lourtet B, Hewison MA. The effect of capture on ranging behaviour and activity of the European roe deer Capreolus capreolus. Wildlife Biol. 2009; 15: 278–287.
- 45. Bivand R, Lewin-Koh N. maptools: Tools for reading and handling spatial objects. R package version 0.8; 2013.
- 46. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing; 2013.
- 47. Herrera EA. Growth and dispersal of capybaras (Hydrochoerus hydrochaeris) in the Llanos of Venezuela. J Zool. 1992; 228: 307–316.
- 48. QGIS. An Open Source Geographic Information System. QGIS. 2017.
- 49. Gislason PO, Benediktsson JA, Sveinsson JR. Random forests for land cover classification. Pattern Recogn Lett. 2006; 27: 294–300.
- 50. Leutner B, Horning N. RStoolbox: Tools for Remote Sensing Data Analysis. R Package version 0.1; 2016.
- 51. Lu D, Weng Q. A survey of image classification methods and techniques for improving classification performance. Int J Remote Sens. 2007; 28: 823–870.
- 52. Tucker CJ, Sellers PJ. Satellite remote sensing of primary production. Int J Remote Sens. 1986; 7: 1395–1416.
- 53. Johnson DH. The comparison of usage and availability measurements for evaluating resource preference. Ecology. 1980; 61: 65–71.
- 54. Manly BFJ, McDonald LL, Thomas DL, McDonald TL, Erickson WP. Resource selection by animals: statistical analysis and design for field studies. 2nd ed. Massachusetts: Springer; 2002.
- 55. Northrup JM, Hooten MB, Anderson CR, Wittemyer G. Practical guidance on characterizing availability in resource selection functions under a use–availability design. Ecology. 2013; 94: 1456–1463. pmid:23951705
- 56. Gillies CS, Hebblewhite M, Nielsen SE, Krawchuk MA, Aldridge CL, Frair JL, et al. Application of random effects to the study of resource selection by animals. J Anim Ecol. 2006: 75: 887–898. pmid:17009752
- 57. Bates D, Maechler M, Bolker B, Walker S. lme4: Linear mixed-effects models using Eigen and S4. R package version 0.1; 2014.
- 58. Burnham KP, Anderson DR. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. New York: Springer; 2002.
- 59. Boyce MS, Vernier PR, Nielsen SE, Schmiegelow FK. Evaluating resource selection functions. Ecol Model. 2002; 157: 281–300.
- 60. Frid A, Dill L. Human-caused disturbance stimuli as a form of predation risk. Conserv Ecol. 2002; 6: 1–16.
- 61. Pardo LE, Campbell MJ, Cove MV, Edwards W, Clements GR, Laurance WF. Land management strategies can increase oil palm plantation use by some terrestrial mammals in Colombia. Sci Rep-UK. 2019; 9: 1–12.
- 62. Polo G, Acosta CM, Labruna MB, Ferreira F. Transmission dynamics and control of Rickettsia rickettsii in populations of Hydrochoerus hydrochaeris and Amblyomma sculptum. PLoS Neglect Trop D. 2017.
- 63. Polo G, Acosta CM, Labruna MB, Ferreira F, Brockmann D. Hosts mobility and spatial spread of Rickettsia rickettsii. Plos Comput Biol. 2018. pmid:30586381
- 64. Campos-Krauer JM, Wisely SM, Benitez IK, Robles V, Golightly RT. Rango de Hogar y uso de Hábitat de Carpinchos en Pastizales recién invadido en el Chaco Seco de Paraguay. Therya. 2014; 5: 61–79.
- 65. Rocha VJ, Sekiama ML, Gonçalves DD, Sampieri BR, Barbosa GP, Dias TC, et al. Capybaras (Hydrochoerus hydrochaeris) and the presence of ticks (Amblyomma sculptum) at the UFSCar Araras campus, São Paulo. Cienc Anim Bras. 2017; 18: 1–15.
- 66. Polo G, Labruna MB, Ferreira F. Satellite Hyperspectral Imagery to Support Tick-Borne Infectious Diseases Surveillance. Plos One. 2015; 10: 1–12.