Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/280925346

Density,occupancy,andactivitypatternoftwosympatricdeer(Mazama)intheAtlanticForest,Brazil

ARTICLEinJOURNALOFMAMMALOGY·AUGUST2015

ImpactFactor:2.23·DOI:10.1093/jmammal/gyv132

DOWNLOADS

24

VIEWS

38

3AUTHORS:

ÁtillaColomboFerreguetti

RiodeJaneiroStateUniversity

3PUBLICATIONS2CITATIONS

SEEPROFILE

WalfridoMoraesTomas

BrazilianAgriculturalResearchCorporation(…

116PUBLICATIONS384CITATIONS

SEEPROFILE

HelenaBergallo

RiodeJaneiroStateUniversity

98PUBLICATIONS772CITATIONS

SEEPROFILE

Availablefrom:ÁtillaColomboFerreguetti

Retrievedon:18August2015

1

Density, occupancy, and activity pattern of two sympatric deer (Mazama) in the Atlantic Forest, Brazil

Átilla C. Ferreguetti,* Walfrido M. Tomás, and Helena G. Bergallo

Department of Ecology, Rio de Janeiro State University, Graduate Program in Ecology and Evolution, Rua São Francisco Xavier, nº 524, Pavilhão Haroldo Lisboa da Cunha, 2º andar, sala 224. Bairro Maracanã, Rio de Janeiro, RJ CEP 20550-019, Brazil (ACF, HGB)Wildlife Laboratory, Embrapa Pantanal, Rua 21 de Setembro, n° 1.880, Bairro Nossa Senhora de Fátima, Corumbá, MS CEP 79320–900, Brazil (WMT)

* Correspondent: atilla.ferreguetti@gmail.com

The red brocket (Mazama americana) and gray brocket deer (Mazama gouazoubira) are sympatric in the Atlantic Forest and present a number of ecological similarities in their diet and habitat use, although interspecific competition in these species is poorly understood. This study aimed to compare abundance and produce evidence of niche partitioning between these species. We estimated population density, activity patterns, habitat occupancy, and detection probabilities for the 2 species in a large remnant of the Atlantic Forest of Espírito Santo, Brazil. Abundance was estimated using the distance sampling technique, and occupancy, detectability, and activity patterns were assessed using camera-trap monitoring at 39 sample sites over a 1-year period. During surveys, we obtained 44 sightings of M. americana and 74 of M. gouazoubira, with 199 records in which the species was not discriminated. We applied a correction for the unidentified sightings to avoid underestimating density. The corrected population size estimates were 3,668 individuals (confidence interval [CI] 95%: 2,989–4,601) for M. americana and 6,701 (CI 95%: 5,857–7,667) for M. gouazoubira. Occupancy probability for M. americana was best described by the large trees density (diameter at breast height > 50 cm) and the edge forest distance, while M. gouazoubira did not present any habitat preferences based on the covariates used for modeling. M. americana was mostly nocturnal, whereas M. gouazoubira was active mainly during the day. We conclude that, despite the abundance of these sympatric species, and their occupation of the same habitats, they may avoid competing for habitat and feeding resources through differences in their activity patterns.

Os veados mateiro (Mazama americana) e catingueiro (Mazama gouazoubira) são espécies simpátricas na Mata Atlântica, e apresentam várias semelhanças ecológicas tais como o uso do habitat e dieta. A hipótese de competição interespecífica entre essas espécies ainda não foi testada. Este estudo teve como objetivo comparar a abundância e verificar evidências de separação de nicho entre essas duas espécies simpátricas. Estimou-se a densidade, o padrão de atividade, e a probabilidade de ocupação das duas espécies em um fragmento de Mata Atlântica no estado do Espírito Santo, Brasil. A abundância foi estimada utilizando a técnica de amostragem distância (esforço total: 908 km), e o padrão de atividade e a probabilidade de ocupação foram avaliadas usando armadilhas fotográficas em 39 locais de amostragem durante o período de um ano. Na transecção linear obtivemos 44 observações de M. americana e 74 observações de M. gouazoubira, com 199 registros em que as espécies não foram discriminadas. Nós aplicamos uma correção para a falta de discriminação de espécies para evitar subestimar a densidade. As estimativas corrigidas de tamanho de população foram 3.668 indivíduos (95% CI: 2.989–4.601) e 6.701 indivíduos (95% CI: 5.857–7.667) para M. americana e M. gouazoubira, respectivamente. A probabilidade de ocupação de M. americana foi melhor descrita pela densidade de árvores com diâmetro altura do peito > 50 cm e a distância da borda da floresta, enquanto M. gouazoubira não mostrou preferência de habitat com base nas co-variáveis utilizadas para a modelagem. M. americana foi principalmente noturna, enquanto M. gouazoubira mostra um padrão de atividade diurno. Podemos concluir que, apesar destas duas espécies serem simpátricas, abundantes, e ocupando os mesmos habitats eles não compartilham o mesmo período de atividade. Esta separação no tempo poderia ser uma adaptação para evitar a competição direta, uma vez que as espécies do gênero Mazama apresentam um nicho muito semelhante quando o habitat e a dieta são considerados.

Key words: abundance, Atlantic Forest, border effect, Cervidae, competition, habitat selection, poaching

© 2015 American Society of Mammalogists, www.mammalogy.org

Journal of Mammalogy, xx(x):1–10, 2015DOI:10.1093/jmammal/gyv132

Journal of Mammalogy Advance Access published August 11, 2015

2 JOURNAL OF MAMMALOGY

The coexistence of species belonging to the same genus (con-generics) may depend on the evolution of adaptations that reduce interspecific competition (MacArthur and Levins 1967; Bulmer 1974; Leibold 1998). Generally, this competition relates primarily to the exploitation of resources such as food and space, and specific adaptations in morphology or behavior patterns may be required to allow the effective coexistence of the 2 species. Niche partitioning may include the use of distinct habitats and dietary resources, as well as different activity pat-terns (Abrams 1983).

The deer species of the genus Mazama are adapted morpho-logically for life in forest habitats and are found in a variety of vegetation types throughout their ample Neotropical distri-bution, extending from southern Mexico to central Argentina, and including the whole of Brazil (Duarte 1996; Eisenberg and Redford 1999; Weber and Gonzalez 2003). However, there appear to be some differences in habitat preferences. Whereas the red brocket deer, Mazama americana, is gener-ally thought to be dependent on mature forest, the gray brocket deer, M. gouazoubira, has been found in sparser vegetation, including grasslands, and even disturbed habitats such as agri-cultural plots (Schaller 1983; Eisenberg and Redford 1999; Leeuwenberg et al. 1999), while maintaining a degree of depen-dency on forested habitats (Desbiez et al. 2009, 2010). These 2 species are sympatric in many parts of their geographical dis-tribution, including the northern portion of the Brazilian state of Espírito Santo (Eisenberg and Redford 1999; Weber and Gonzalez 2003; Rivero et al. 2005; Chiaravalloti et al. 2010).

Data on habitat selection by individuals or populations are fundamentally important information for management and con-servation measures (Caughley 1994, Otis 1997). The quality of the habitats used by 1 species within the home range may reflect both their needs as ecological tolerances in places where they live (Jones 2001). Thus, understanding which factors determine the presence of a species in a biome as threatened as the Atlantic Forest contributes to the development of con-servation measures to ensure their survival in such areas (Jones 2001). However, virtually nothing is known about habitat pref-erences and use in brocket deer, or the factors that may permit them to occur in sympatry.

With the aim of contributing to the understanding of ecologi-cal patterns in Neotropical ungulates, as well as the conserva-tion of their populations, we evaluated the density, abundance, occupancy, detectability, and activity patterns of the red and gray brocket deer in a remnant of the Atlantic Forest of Espírito Santo, Brazil. In addition to determining the spatial and tem-poral distribution of the 2 species, this study analyzed their densities in different habitats. In particular, we expected that the 2 species have similar population densities in the reserve, since both species are hunted and share several characteristics. We tested the hypothesis that, when they occur in sympatry, the 2 species should present distinct patterns in space (habi-tat use) and/or time (activity period). To check if there exists a difference in habitat selection between species, we predicted the nature (direction) of the response of 7 covariates. A priori, we expected that the covariates measured should represent key

habitat features for the species (i.e., trees density, understory cover, lianas density, and water resource distance) or elements possibly avoided by 1 or both species (i.e., main road distance, forest edge distance). We also expected to find a potential effect of poaching in the occupancy probability and detectability of the brocket deer species.

Material and MethodsStudy area.—The present study was conducted in the Vale

Natural Reserve (VNR), a protected area of 23,500 ha belong-ing to the Vale do Rio Doce Company. The reserve is located in the neighboring municipalities of Linhares and Jaguaré (19°06′–19°18′S and 39°45′–40°19′W), in northeastern Espírito Santo, Brazil (Fig. 1).

The VNR was established through a gradual process of land acquisition, which started in 1955, when Vale bought its 1st properties in the region, and reached its current boundaries in 1973. The reserve is composed of 1 main block of forest (approximately 98.1% of the total area), and a much smaller fragment, known as Biribas Reserve, to the southwest of the main block.

According to the Vegetation Map of Brazil (IBGE 1993), the VNR is covered by dense rainforest, composed of a mosaic of habitats with 4 main vegetation types (adapted from Jesus 1987; Peixoto and Gentry 1990): coastal plain forest (“tabu-leiro” forest), riparian forest, “mussununga” forest, and natural grassland. The evergreen tabuleiro forest has 2 or more upper strata and high densities of lianas and epiphytes, and covers approximately 68% of the total area of the reserve. The riparian forest, which covers some 4% of the reserve, is a mixed type of vegetation associated with streams, characterized by widely spaced trees and a predominance of palms. The Mussununga (covering approximately 8% of the VNR) is a type of woody vegetation on sandy soils, physiognomically similar to the tab-uleiro forest at an early or intermediate stage of regeneration. The natural grasslands are isolated patches that occur through-out the forest at the sites of former ponds, in the geological past, and cover around 6% of the area of the reserve. In addition to these formations, approximately 8% of the reserve is covered by wetlands (swamps) and streams (Fig. 2).

Line-transect surveys—To estimate the density and abun-dance of the 2 brocket deer species, we established four 5-km-long line transects, following the RAPELD protocol (i.e., rapid assessments, such as the RAP surveys carried out by Conservation International). To understand ecological pro-cesses within sites, we need long-term ecological research (LTER) sites. The Brazilian acronym for LTER is PELD; hence the name for the method, RAPELD (Magnusson et al. 2005). Transects were separated from one other by a minimum of 4 km.

During a 13-month period (April 2013 to May 2014), the transects were surveyed using distance sampling techniques (Buckland et al. 2001; Fig. 2). On a given field day, a tran-sect was surveyed twice by a single observer starting at sun-rise (between 0530 and 0630 h). The observer waited 3 h before

FERREGUETTI ET AL.—POPULATIONS OF TWO MAZAMA SPECIES 3

starting the afternoon survey, between 1300 and 1400 h. The observer walked the transects at a speed of approximately 1 km/h. Each transect was surveyed twice a month, with the order in which the transects were surveyed being alternated each month.

For each sighting, we recorded the species and perpendicu-lar distance of the animal from the transect (with a measur-ing tape), the length of the transect walked to that point, the date, the timing of the record, and the transect number. Species identification was possible only when the animals were sta-tionary. We used external morphological characteristics to distinguish the species, including body size (M. americana is larger than M. gouazoubira) and color (M. americana is more reddish than M. gouazoubira), and the size of the ears, which are larger in M. gouazoubira.

The violation of one of the assumptions of distance sam-pling—the correct identification of species (Buckland et al. 2001)—had a potentially deleterious effect on the estimates of population density, with considerable underestimates resulting from the non-identification of sightings. The morphological similarities of the 2 species often hampers the identification of individuals sighted during transect surveys, even by expe-rienced observers. Facing the same problem in the Brazilian Pantanal wetlands, Schaller (1983) simply divided the uniden-tified sightings equally between the 2 species, while estimating density from a simple strip transect (not considering detectabil-ity of < 1).

Whenever the identification was uncertain, due to the physi-cal similarities of the species, the sighting was recorded as “unidentified.” To correct for this error of identification, we randomized these records using the proportion of each species found in the total number of transect sightings and camera-trap records. In support of this, the 2 species were recorded in similar proportions in the 2 sampling procedures, i.e., 38% M. americana versus 62% M. gouazoubira in the transect sur-veys and 40% versus 60%, respectively, in the camera-trap

data. Using the proportion of observations that have been iden-tified with certainty, we did 1,000 randomizations by setting a looping function in R software using the “Distance” package (R Development Core Team 2008; Miller 2012). We fixed the estimated proportions in the looping function to designate the observations each time in the randomization process. At the end of the looping function, we obtained an average of corrected estimates. This correction increased the total number of records available for each species, allowing for more accurate estimates of population density and size.

Camera trapping.—We selected 39 sampling sites using a systematic random design stratified by vegetation type to ensure that all 4 of the principal vegetation types found in the VNR were represented (i.e., tabuleiro forest, riparian forest, mussununga, and natural grassland). This scheme was designed to model the occupancy probability of each species, as well as documenting their respective activity patterns (Fig. 2). We placed a random grid over a digital map of the reserve, and identified the sampling points by selecting grid cells.

Camera traps were installed in the center of each selected grid cells, where animal tracks were observed. We did not use bait to attract the deer. This approach resulted in a relatively even distribution of points within the VNR, while maintain-ing independence among points, which were all separated from one another by a distance of more than 1 km. At each site, we installed 1 passive infrared Bushnell camera trap (Bushnell Outdoor Products) in picture function, approxi-mately 40–50 cm above the ground. All stations were checked every 20–25 days to change batteries, when necessary. The traps were programmed to operate for 24 h/day. We collected samples between May 2013 and June 2014, where sampling intervals were 5 days with a 5-day gap in between in which the target species were either detected or not by the camera traps.

Covariates.—We used the same 7 covariates to model the occupancy probability for the 2 deer species. These were the edge forest distance (dist_edge), trees with a diameter at breast

Fig. 1.—Location of the Vale Natural Reserve in the state of Espírito Santo, Brazil.

4 JOURNAL OF MAMMALOGY

height (DBH) of more than 50 cm density (dens_trees), lianas density (dens_lian), nearest stream distance(dist_water), under-story cover (und_cov), and main road distance (dist_main_road). The potential effect of poaching on detectability was also analyzed, using records obtained by the reserve administration over a 10-year period (Vale Natural Reserve, Linhares, Brazil, pers. comm.).

The density of poaching within the study area was calcu-lated using the georeferenced database of 10 years of poaching records collected by the reserve’s security guards (Vale Natural Reserve, Linhares, Brazil, pers. comm.). These records were used to calculate a poaching density (records per km2) for each grid square in which a camera trap had been installed.

At each sampling point, we established 4 plots (30 × 50 m) arranged by the cardinal compass points (north, south, east, and west). In each plot, we measured the DBH of each tree and counted the number of large trees (DBH > 50 cm). Understory

cover and the density of lianas were measured along the cen-tral (longitudinal) line of each plot, considering 5 m on each side of the line (transect width of 10 m). The density of lianas was based on the number of stems rooted within each 10 × 50 m transect. Understory cover was measured every 10 m using a 2 × 0.5 m sighting frame (each 0.5 m2 portion representing 25% visibility). Understory cover within each transect was estimated by the mean visibility of the sighting frame at the 5 points sampled within the transect. The spatial covariates were quantified for each of the 39 sample sites using ArcGIS soft-ware (ArcMap 10.1—ESRI).

Data analysis.—Population density was estimated based on the total number of survey records in which the deer species was identified, as well as the corrected data (identified + unidenti-fied records attributed to the species). We used DISTANCE 6.2 software to estimate these densities (Buckland et al. 2001). As no sightings were recorded in other habitat types, only forested

Fig. 2.—Habitat mosaics inside the Vale Natural Reserve, Espírito Santo, Brazil. Black triangles represent the location of each camera trap, and the red lines represent the survey transects.

FERREGUETTI ET AL.—POPULATIONS OF TWO MAZAMA SPECIES 5

areas were considered in this analysis. We considered the popu-lation as not being closed for the purposes of this analysis.

Based on the approach of MacKenzie et al. (2006), we esti-mated site occupancy (Ψ) and the detection probability (P) for each deer species, with 3 possible outcomes: 1) the site was occupied and the species was detected (Ψ × P); 2) the species was present but not detected (Ψ × [1 – P]), and 3) the species was not present and therefore was not detected (1 – Ψ).

We estimated detection probabilities by sampling each site on multiple occasions. The probability was the parameter pro-jected by a maximum likelihood estimation of the proportion of sites occupied (Ψ) during the sample period. We verified that occupancy was closed (i.e., did not change) for both species using the single-season model. This exercise indicated that the occupancy status for each species was constant throughout the study, allowing us to use closed occupancy models (MacKenzie et al. 2006).

In our occupancy analysis, we considered each species sepa-rately and assessed the covariates that might affect occupancy and detectability, in an attempt to identify divergences in habi-tat preferences. We constructed a set of 11 candidate models selected by a priori hypotheses based on 3 different approaches: 1) considering occupancy probability and detectability as constant across all sites for both species, 2) considering the variation in occupancy as a function of habitat covariates, and 3) considering the variation in detectability as a function of some covariates (i.e., poaching density, understory cover, and trees density). This allowed us to evaluate whether species-specific differences in habitat preferences were determined by a single covariate or a set of covariates, which would contribute to an improvement in the model’s performance.

The occupancy modeling was run in the PRESENCE 9.3 software (Mackenzie and Royle 2005) with 2,000 bootstraps to assess the adjustment fit (p) and the over-dispersion param-eter (ĉ). In our assessment of occupancy closure and the factors that influenced occupancy and detection, we ranked all mod-els according to Akaike’s information criterion (AIC—Akaike 1973). All models with a ΔAIC value of less than 2 were con-sidered to be equivalent. We choose the best model from the entire set of models by using the Akaike weight (w). The Akaike weight provides a straightforward interpretation of the probabil-ities of each model’s being the best model in an AIC sense (i.e., by calculating a weighted average of the beta of each covariate and using the AIC weight as the weighting variable).

The circadian cycle of each species was defined based on the timing of the photographs obtained from the camera traps. To avoid pseudoreplication, records were only considered to be independent when separated by an interval of 24 h. Circular summaries (Lund and Agostinelli 2007) were used to deter-mine the mean overall timing of the activity of each species as recorded by the camera traps. A resampling approach was used to derive unbiased estimates of the 95% confidence intervals (95% CIs), whereby the circular mean from a random sample of 100 data points for each species was calculated 10,000 times with replacement to generate reliable estimates. To test if the activity period of the both brocket deers was affected by the

poaching density, we used the Mardia–Watson–Wheeler test (Batschelet 1981). Analyses were conducted in the “Circular” package in R (R Development Core Team 2008) with associ-ated analytic packages.

ResultsDensity and abundance.—The 4 transects were surveyed

196 times (49 times each). During these surveys, M. ameri-cana was sighted 44 times (38% of identified sightings) and M. gouazoubira, 74 times (62%). We obtained a further 199 records in which the species could not be identified.

The corrected data on M. americana fitted best to a half nor-mal curve with cosine adjustment. Effective strip width (ESW) was 3.95 ± 0.27 m, with sightings being obtained at distances of between 0 and 12 m from the transect. The density of the red brocket deer in the study area was estimated at 0.15 ± 0.015 individuals/ha, with a total population of 3,569 ± 354 individu-als. The coefficient of variation for both parameters was 9.92% and the detection probability and encounter rate were respon-sible for 49% and 51% of the total variation, respectively. The corrected density estimate was 2.14 times higher than the uncorrected estimate (0.07 + 0.01 individuals/ha).

The corrected data on M. gouazoubira fitted best to a hazard-rate curve with cosine adjustment. The ESW was 5.07 ± 0.44 m, with sightings between 0 and 15 m from the transect. The cor-rected density estimated for this species was 0.21 ± 0.02 deer/ha, with a total population size of 5,816 ± 545 individuals. The coefficient of variation for both parameters was 9.37%, and the detection probability and encounter rate were responsible for 88.6% and 11.4% of the total variance, respectively. The cor-rected estimate of density was 1.75 times higher than the uncor-rected estimate (0.12 + 0.01).

Occupancy models

Mazama americana.—The red brocket deer was observed at 29 of the 39 sampling sites, resulting in a naïve occupancy probability of 0.74. Based on the 11 different occupancy mod-els produced (Table 1), the probability of occupancy was best described by 2 habitat covariates—tree density (dens_trees) and the distance from the forest edge (dist_edge) and poaching. The probability of occupancy was higher (Ψ > 0.8) at sites where tree density (DBH > 50 cm) was higher than 800 individuals per hectare (Fig. 3B). Occupancy was also higher (Ψ > 0.9) at sites located more than 500 m from the edge of the forest (Fig. 3A). The occupancy was negatively affected by the poaching den-sity, with occupancy probabilities ranging from Ψ = 0.99 to 0.16 (Fig. 4). Detection probability also had a negative rela-tionship with poaching density (Fig. 4). Detection probabilities ranged from P = 0.44 to 0.31. Detection probability was also affected by the understory cover (und_cov), showing a negative relationship. The higher the percentage of the understory cover (25%, 50%, 75%, and 100%), the lower the detection of the species in the camera traps. The density of lianas (dens_liana) had a negative effect on the detectability, ranged from P = 0.58 to 0.01 (Fig. 5).

6 JOURNAL OF MAMMALOGY

Mazama gouazoubira.—The gray brocket deer was observed at all 39 sites, resulting in a naïve estimate of the occupancy probability equal to 1. As M. gouazoubira was observed at all sites without any pattern in the detection history between occasions, it was impossible to discriminate the effects of the covariates on the occupancy and detection probabilities for this species (Table 1).

Activity patterns. We obtained 1,560 photographs of M. americana and 2,523 of M. gouazoubira. M. americana was mostly active during the night (peaking between 0100 and 0400 h), whereas M. gouazoubira presented the opposite pat-tern, being active primarily during the day, with 2 peaks, one be-tween 0600 and 0900 h, and the other between 1300 and 1700 h (Fig. 6). We did not find a relationship between the activity pe-riod with the poaching density (χ2 = 38.23, d.f. = 38, P = 0.53).

DiscussionThe correction strategy used in the present study allowed us to include all sightings, and thus maximize sample size. While using this correction appeared to compensate for the under-estimates of population density, it is interesting to note that it had little effect on the relative abundance of the 2 species, with a ratio of 0.7 red brocket to gray brocket before correc-tion, and 0.6 red:gray after correction. This would be consistent with the difference in body mass (adult body weight 34.9 kg in M. americana and 16.3 kg in M. gouazoubira—Groves and

Grubb 2011) and related ecological parameters, i.e., carrying capacity (Robinson and Redford 1986).

The red brocket deer (M. americana) presented a positive relationship with both the distance from the forest edge and the density of large trees (DBH > 50 cm). This may reflect the preference of this species for denser and better preserved for-ests (Eisenberg and Redford 1999), and the possible avoid-ance of edge effects, primarily the increase in the understory cover of non-arboreal plants more resistant to direct sunlight, such as bamboo, lianas, and vines (Tabarelli et al. 1999). A number of previous studies have shown a strong associa-tion of M. americana with mature forest habitats (Rivero et al. 2005; Tiepolo and Tómas 2006; Chiaravalloti et al. 2010). As expected, the detectability was lower in places with higher percentage of understory cover and density of liana, what could be explained by the fact that 80% of the VNR is composed by forests (Jesus and Rolim 2005) with dense understory, which difficult the detection of the specie in the camera traps.

By contrast, the occupancy probability for M. gouazoubira was not affected by any of the covariates, and these deer were detected in all vegetation types in the VNR. This is because M. gouazoubira presents a higher plasticity in habitat use, being frequently found in savanna habitats and ecotones in South America (Eisenberg and Redford 1999; Rivero et al. 2005; Chiaravalloti et al. 2010). A number of studies have shown that M. gouazoubira can occur in both open and forested

Table 1.—Occupancy models for Mazama americana and M. gouazoubira in the Vale Natural Reserve, Brazil, estimated by camera trapping between May 2013 and June 2014, grouped in sampling intervals of 5 consecutive days. Covariates: forest edge distance (dist_edge); trees with diameter breast height > 50 cm density (dens_tree); poaching density (poaching); understory coverage (und_cov); main road distance (dist_main_road); lianas density (dens_liana); and water resources distance (dist_water). Ψ = occupancy, p = detectability, AIC

w = Akaike weight, ĉ = disper-

sion parameter.

Model AIC ΔAIC AICw

Number of parameters

ĉ

M. americana Ψ(dist_edge;dens_trees;poaching); p(poaching;und_cov;dens_liana) 1,026.32 0 0.77 8 1.06 Ψ(dist_edge;dens_tree;poaching); p(poaching;dens_tree) 1,029.34 3.01 0.13 7 1.09 Ψ(dist_edge;dens_tree;poaching); p(poaching) 1,031.23 4.91 0.04 6 1.13 Ψ(dist_edge;dens_tree); p(poaching) 1,034.56 8.24 0.03 5 1.18 Ψ(poaching); p(dens_liana) 1,091.1 64.78 0.006 4 1.22 Ψ(dist_edge;dens_tree); p(und_cov) 1,091.5 65.18 0.005 5 1.21 Ψ(dens_lian); p(.) 1,092.6 66.28 0.003 3 1.22 Ψ(dens_lian;dist_water;poaching); p(poaching;und_cov) 1,092.6 66.28 < 0.001 7 1.21 Ψ(.); p(.) 1,095.3 68.98 < 0.001 2 1.32 Ψ(dist_main_road); p(.) 1,095.8 69.48 < 0.001 3 1.25 Ψ(dist_water); p(.) 1,096.7 70.38 < 0.001 3 1.31M. gouazoubira Ψ(.); p(.) 1,458 0 0.83 2 1.23 Ψ(dist_water;dist_edge); p(poaching) 1,464.5 6.5 0.05 5 1.29 Ψ(dens_tree;dist_edge); p(und_cov) 1,465.3 7.3 0.04 5 1.32 Ψ(dist_edge;dens_tree;poaching); p(poaching) 1,467.6 9.6 < 0.001 6 1.33 Ψ(dens_liana); p(poaching;und_cov) 1,468.9 10.9 < 0.001 5 1.34 Ψ(dist_edge;poaching); p(dens_liana;poaching;und_cov) 1,470.3 12.3 < 0.001 7 1.33 Ψ(dist_main_road); p(.) 1,472.7 14.7 < 0.001 3 1.32 Ψ(und_cov); p(.) 1,476.34 18.33 < 0.001 3 1.33 Ψ(dist_edge); p(.) 1,476.45 18.45 < 0.001 3 1.33 Ψ(dens_liana); p(.) 1,477.45 19.45 < 0.001 3 1.31 Ψ(und_cov); p(.) 1,478.56 20.56 < 0.001 3 1.32

FERREGUETTI ET AL.—POPULATIONS OF TWO MAZAMA SPECIES 7

environments, which may result in habitat-related differences in density between the two species (Eisenberg and Redford 1999; Tiepolo and Tomas 2006).

The occupancy probability and detectability of M. ameri-cana were affected by poaching pressure, reflecting the fact that this species is one of the most prized game animals in the Neotropics (Bodmer 1995; Wemmer 1998; Fa et al. 2002). Chiarello (2000) has already emphasized the need for the implementation of effective measures to reduce poaching in the VNR as a necessary measure for the conservation of the spe-cies. However, poaching is still a major concern in the VNR, which is further supported by the results of our study. This pres-sure could lead to local extinction (Cullen et al. 2000; Peres 2001) and eventually turn this important reserve into an “empty

forest” (Redford 1992). Di Bitetti et al. (2008) also identified M. americana as being highly sensitive to poaching pressure in Amazonian forests.

Although the 2 deer species overlapped spatially, they pre-sented quite distinct temporal niches in the VNR. Whereas M. gouazoubira is predominantly diurnal, with activity peaking in the hottest hours of the day, M. americana is mostly noctur-nal. The overlap in the spatial dimension may, thus, be offset by the reduced overlap in the temporal one (Schoener 1974), contributing to the niche partitioning necessary for the long-term coexistence of these closely-related species (Pianka 1988; Begon et al. 2006). In Bolivia, Rivero et al. (2005) also reported

Fig. 3.—A) Relationship between the occupancy of Atlantic Forest by Mazama americana and the forest edge distance; B) Relationship between the occupancy of Atlantic Forest by M. americana and the large trees density (diameter at breast height > 50 cm), estimated by camera trapping at the Vale Natural Reserve between May 2013 and June 2014, grouped in sampling intervals of 5 consecutive days.

Fig. 4.—Relationship between detectability (black circles) and occu-pancy (gray triangles) of Mazama americana with the poaching den-sity in the Vale Natural Reserve between May 2013 and June 2014, grouped in sampling intervals of 5 consecutive days.

Fig. 5.—Relationship between detectability of Mazama americana and the lianas density in the Vale Natural Reserve between May 2013 and June 2014, grouped in sampling intervals of 5 consecutive days.

8 JOURNAL OF MAMMALOGY

that sympatric M. americana is most active at night, while M. gouazoubira tended to be more active around dawn. As the 2 species share very similar diets, this behavioral asynchrony may contribute to a reduction in interspecific competition (Brannan et al. 1985; Gayot et al. 2004). In southeastern Peru, Tobler et al. (2009) also found that the red brocket was more nocturnal and the gray brocket more diurnal. In Argentina, how-ever, Ferrari (2005) reports that M. americana and M. nana are active during the same time phase, whether sympatric or not.

Although we found no direct evidence that poaching could be affecting the activity period, the activity pattern recorded in this study for M. americana could also be at least partly due to the strong poaching pressure in the VNR. This species does in fact appear to be able to adjust its activity pattern in response to poaching, becoming more nocturnal in areas where pressure is intense (Di Bitetti et al. 2008).

By contrast, diurnal behavior appears to be more typical of M. gouazoubira. Barrientes and Maffei (1999) recorded 2 peaks of activity (0600–1000 and 1600–1800 h) in 2 radiocol-lared animals, while Maffei et al. (2002) also recorded diurnal behavior in an area where M. americana was almost completely absent.

Overall, M. gouazoubira is more flexible and uses a wider variety of habitats than M. americana in the VNR. This pro-vides some evidence of spatial niche partitioning, although the primary dimension appears to be the difference in activity pat-terns. Gilbert et al. (2006) noted that, due to the enormous mor-phological similarities found among Latin American cervids, independent adaptations might have developed over time in Mazama species. This would account for the temporal segrega-tion observed in our study.

Even so, additional studies of habitat use, diet, and activity patterns of the different brocket deer species under sympatric

and allopatric conditions will be required before more con-sistent conclusions can be drawn on their ecological patterns. From a conservation perspective, future studies in the area should monitor population trends over time to develop a better understanding of these sympatric species. This will be impor-tant for the understanding of possible changes in habitat use and species interactions, especially in the case of M. ameri-cana, which is more sensitive to poaching pressure as well as habitat loss and disturbance.

AcknowledgmentsWe thank Vale Natural Reserve for their support of this research. We thank B. LaBumbard for the English review. We also thank the anonymous reviewers for the considerations to improve this paper. AFC is also grateful to Coordenacão de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for a graduate scholar-ship. HGB thanks Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ; E-26/103.016/2011), Prociência/Universidade do Estado do Rio de Janeiro (UERJ), and National Counsel of Technological and Scientific Development (CNPq) (457458/2012-7, 307715/2009-4) for research and productivity grants. This study was financed by the Research Program in Biodiversity Mata Atlântica (PPBio MA). Data storage is sup-ported by PPBio MA (CNPq - 457458/2012-7) and FAPERJ (E-26/111.394/2012).

Literature CitedAbrams, P. 1983. The theory of limiting similarity. Annual Review of

Ecology, Evolution, and Systematics 14:76–359.Akaike, H. 1973. Information theory and an extension of the maxi-

mum likelihood principle. Pp. 267–281 in Second international

Fig. 6.—Activity patterns of Mazama americana and M. gouazoubira in the Atlantic rainforest in the Vale Natural Reserve, Espirito Santo, Brazil, estimated by camera trapping between May 2013 and June 2014.

FERREGUETTI ET AL.—POPULATIONS OF TWO MAZAMA SPECIES 9

symposium on information theory (B. N. Petrov and F. Csaki, eds.). Academiai Kiado, Budapest, Hungary.

Barrientes, J., and L. Maffei. 1999. Pp. 161–164 in Radio telemetría de la urina (Mazama gouazoubira) en el campamento Cerro Cortado, Izozog, Santa Cruz, Bolivia. Estudios de inses-tigacíon y manejo de fauna en el Izozog y Parque Nacional Kaa Iya. CABI-WCS, Santa Cruz, California.

Batschelet, E. 1981. Circular statistics in biology. Academic Press, New York.

Begon, M., C. R. Towsend, and J. L. Harper. 2006. Ecology: from individuals to ecosystems. Blackwell Publishing, Oxford, United Kingdom.

Bodmer, R. E. 1995. Managing Amazonian wildlife: biological correlates of game choice by detribalized hunters. Ecological Applications 5:872–877.

Brannan, W. V., M. C. Woekhoven, and R. L. Marchington. 1985. Food habits of brocket and white-tailed deer in Suriname. Journal of Wildlife Management 49:972–978.

Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers, and L. Thomas. 2001. Introduction to distance sampling: estimating abundance of biological population. Oxford University Press, Oxford, United Kingdom.

Bulmer, M. G. 1974. Density-dependent selection and character dis-placement. The American Naturalist 108:45–58.

Caughley, G. 1994. Directions in conservaton biology. Journal of Animal Ecology 63:215–244.

Chiaravalloti, R. M., W. M. Tomás, A. R. Camilo, and M. Aiza. 2010. Separação de nicho entre duas espécies simpátricas de vea-dos do gênero Mazama em uma paisagem complexa no pantanal. Simpósio sobre recursos naturais e socioeconômicos do Pantanal, 5. Embrapa Pantanal, Corumbá, Mato Grosso do Sul.

Chiarello, A. G. 2000. Influência da caça ilegal sobre mamíferos e aves das matas de tabuleiro do norte do estado do Espírito Santo. Boletim Museu de Biologia Mello Leitão 11:229–247.

Cullen, L., Jr., R. E. Bodmer, and C. Valladares-Pádua. 2000. Effects of hunting in habitat fragments of the Atlantic Forests, Brazil. Biological Conservation 95:49–56.

Desbiez, A. L. J., R. E. Bodmer, and W. M. Tomas. 2010. Mammalian densities in a Neotropical wetland subject to extreme climatic events. Biotropica 42:372–378.

Desbiez, A. L. J., S. A. Santos, and W. M. Tomas. 2009. Habitat partioning and biomass of four species of deer in the central region of the Brazilian Pantanal. IUCN Deer Specialist Group Newsletter 23:8–16.

Di Bitetti, M. S., A. Paviolo, C. A. Ferrari, C. Angelo, and Y. Di Bianco. 2008. Differential responses to hunting in two sympatric species of brocket deer (Mazama americana and Mazama nemo-rivaga). Biotropica 40:636–645.

Duarte, J. M. B. 1996. Guia de identificação de cervídeos brasileiros. Fundação de Apoio a Pesquisa, Ensino e Extensão (FUNEP), Jaboticabal, Brazil.

Eisenberg, J. F., and K. H. Redford. 1999. Mammals of the neo-tropics: The central neotropics. The University of Chicago Press, Chicago, Illinois.

Fa, J. E., C. A. Peres, and J. Meeuwig. 2002. Buchmeat exploitation in tropical forests: an intercontinental comparison. Conservation Biology 16:232–237.

Ferrari, C. A. 2005. Abundância, uso del hábitat y horários de actividad de la corzuela roja (Mazama americana) y la corzuela enana (Mazama nana) en la selva Paranaense. Thesis. Universidade Nacional de Mar Del Plata, Mar Del Plata, Buenos Aires.

Gayot, M., O. Henry, G. Dubost, and D. Sabatier. 2004. Comparative diet of the two forest cervids of the genus Mazama in French Guiana. Journal of Tropical Ecology 20:31–43.

Gilbert, C., A. Ropiquet, and A. Hassanin. 2006. Mitochondrial and nuclear phylogenies of Cervidae (Mammalia: Ruminantia): sys-tematics, morphology, and biogeography. Molecular Phylogenetics and Evolution 40:101–117.

Groves, C., and P. Grubb. 2011. Ungulate Taxonomy. Johns Hopkins University Press, Baltimore, Maryland.

IBGE. 1993. Mapa de Vegetação do Brasil. Fundação Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Brazil.

Jesus, R. M. 1987. Mata Atlântica de Linhares: Asectos Florestais. Pp. 35–53 in Desenvolvimento Econômico e Impacto Ambiental em Áreas de Trópico Úmido Brasileiro - A experiência da CVRD (Secretaria Estadual de Meio Ambiente, Idaho Water Resource Board, and Companhia Vale do Rio Doce, eds.). A Secretaria, Rio de Janeiro, Brazil.

Jesus, R. M., and S. G. Rolim. 2005. Fitossociologia da Mata Atlântica de Tabuleiro. Boletim Técnico da Sociedade de Investigações Florestais 19:1–149.

Jones, J. 2001. Habitat seletion studies in avian ecology: a critical review. Auk 118:557–562.

Leeuwenberg, F., I. Oliveira-Cabral, and S. Lara-Resende. 1999. Gray brocket deer (Mazama gouazoubira) in the brazilian savanne. Deer Specialist Group News. 15:1–14.

Leibold, M. A. 1998. Similarity and local coexistence of species from regional biotas. Evolutionary Ecology. 12:95–100.

Lund U., and C. Agostinelli. 2007. Circular: circular statistics. R package version 0.3-8. http://CRAN.R-project.org/package5cir-cular. Accessed 5 September 2014.

MacArthur, R., and R. Levins. 1967. The limiting similarity, convergence, and divergence of coexisting species. The American Naturalist 101:377–385.

Mackenzie, D. I., and J. A. Royle. 2005. Designing occupancy studies: general advice and allocating survey effort. Journal of Applied Ecology 42:1105–1114.

Mackenzie, D. I., J. Nichols, J. Royle, K. Pollock, L. Bailey, and J. Hines, J. 2006. Occupancy estimation and modeling: Inferring patterns and dynamics of species occurrence. Elsevier Publishing, San Diego, California.

Maffei, L., E. Cuéllar, and A. Noss. 2002. Uso de trapas cámera para la evaluación de mamíferos em el ecotono chaco-chiquita-nia. Revista Boliviana de Ecologia y Conservacion Ambiental 11:55–65.

Magnusson, W. E., et al. 2005. RAPELD: Uma modificação do método de Gentry para o levantamento da biodiversidade em áreas. Biota Neotropica 5. http://www.biotaneotropica.org.br/v5n2/pt/abstract?point-of-view+bn0100522005. Accessed 15 October 2014.

Miller, D. L. 2012. On smooth models for complex domains and distances. Ph.D. dissertation, University of Bath, Bath, United Kingdom.

Otis, D. L. 1997. Analysis of habitat selection studies with mul-tiple patches within cover types. Journal Wildlife Management 61:1016–1022.

Peixoto, A. L., and A. Gentry. 1990. Diversidade e composição florística da Mata de Tabuleiro na Reserva Florestal de Linhares (Espírito Santo, Brasil). Revista Brasileira de Botânica 13: 19–25.

Peres, C. A. 2001. Synergistic effects of subsistence hunting and hab-itat fragmentation on Amazonian forest vertebrates. Conservation Biology 15:1490–1505.

10 JOURNAL OF MAMMALOGY

Pianka, E. R. 1998. Evolutionary ecology. 4th ed. Harper and Row, New York.

R Development Core Team. 2008. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. Accessed 15 October 2014.

Redford, K. H. 1992. The empty-forest syndrome. BioScience 42:412–422.

Rivero, K. G., B. Rumiz, and A. B. Taber. 2005. Differential hab-itat use by two sympatric brocket deer species (Mazama ameri-cana and Mazama gouazoubira) in seasonal Chiquitano Forest of Bolívia. Mammalia 69:169–183.

Robinson, J. G., and K. H. Redford. 1986. Body size, diet and population density of Neotropical forest mammals. The American Naturalist 128:385–395.

Schaller, G. B. 1983. Mammals and their biomass on a Brazilian ranch. Arquivos de Zoologia 31:1–36.

Schoener, T. W. 1974. Resource partitioning. Pp. 91–126 in Community ecology: pattern and processes (J. Kikkawa and D. J. Anderson, eds.). Blackwell Scientific, Palo Alto, California.

Tabarelli, M., W. Mantovani, and C. A. Peres. 1999. Effects of habitat fragmentation on plant guild structure in the montane

Atlantic forest of southeastern Brazil. Biological Conservation 91:119–127.

Tiepolo, L. M., and W. M. Tomas. 2006. Ordem Artiodactyla. Pp. 283–300 in Mamíferos do Brasil (N. R. Reis, A. L. Peracchi, W. A. Pedro, and I. P. Lima, eds.). Edifurb, Londrina, Brazil.

Tobler, M. W., S. E. Carrillo-Percastegui, and G. Powell. 2009. Habitat use, activity patterns and use of mineral licks by five species of ungulate in south-eastern Peru. Journal of Tropical Ecology 25:261–270.

Weber, M., and S. Gonzalez. 2003. Latin American deer diversity and conservation: a review on status and distributions. Ecoscience 10:443–454.

Wemmer, C., and IUCN/SSC Deer Specialist Group. 1998. Deer status survey and conservation action plan. International Union for the Conservation of Nature, Gland, Switzerland.

Submitted 30 January 2015. Accepted 14 July 2015.

Associate Editor was Harald Beck.