BIOTROPICA 40(2): 211���217 2008 10.1111/j.1744-7429.2007.00350.x Brazilian Tapir Density in the Pantanal: A Comparison of Systematic Camera-Trapping and Line-Transect Surveys Mogens Trolle1 Mammal Department, Zoological Museum, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmark Andrew J. Noss Wildlife Conservation Society-Bolivia, Casilla 6272, Santa Cruz de la Sierra, Bolivia Jose�� Luis Passos Cordeiro Laborat �� orio de Geoprocessamento, Centro de Ecologia, Universidade Federal do Rio Grande do Sul, Caixa Postal 15007, Porto Alegre, RS, Brazil and Luiz Flamarion B. Oliveira Depto. de Vertebrados, Sec �� �� ao de Mastozoologia, Museu Nacional, UFRJ, Quinta da Boa Vista, 20.940-040 Rio de Janeiro, RJ, Brazil ABSTRACT The density of Brazilian tapirs (Tapirus terrestris) was studied in the northeastern part of the Pantanal wetlands of Brazil using two simultaneous and independent methods: (1) systematic camera trapping combined with capture���recapture analysis, with camera traps spaced 1 km apart and distributed over 54 km2 and (2) line-transect sampling using an array of 12 linear transects, from 3.8 to 7.2 km long, covering the principal open and forest habitat types across the entire 1063 km2 SESC Pantanal Reserve. The two methods yielded conservative density estimates of 0.58 �� 0.11 tapirs/km2 (camera trapping) and 0.55 (95% CI 0.30���1.01) tapirs/km2 (line transects). The study suggests that certain Pantanal habitats and sites can sustain relatively high population densities of tapirs when these animals are protected from hunting. Further testing of the camera-trapping methodology as applied to tapirs is required, particularly focusing on extending the survey period. As it represents a relatively rapid method for estimating population density, in comparison to line-transect surveys, and as it generates information simultaneously on multiple species that are conservation priorities, we recommend that camera-trapping surveys be applied more widely across a variety of Pantanal habitats and land-use categories in order to confirm the value of the vast 140,000 km2 wilderness region for this vulnerable species. Abstract in Portuguese is available at http://www.blackwell-synergy.com/loi/btp. Key words: Brazil Neotropics Pantanal wetlands Tapirus terrestris. THE BRAZILIAN TAPIR TAPIRUS TERRESTRIS IS VULNERABLE TO LOCAL extinction throughout its range as a result of habitat destruction and hunting (Bodmer & Brooks 1997). The large browser is known to favor waterside habitats such as swamps and river edges (Padilla & Dowler 1994, Emmons & Feer 1997) and several studies have noted its presence in the Pantanal wetlands of southwestern Brazil (Miller 1930, Schaller 1983, Trolle 2003). The Pantanal covering approximately 140,000 km2 is one of the wildlife hotspots of South America (Mittermeier et al. 1990, Alho & Lacher 1991, WWF et al. 1994, Swarts 2000). Trolle (2003) suggested that the region is a stronghold for the Brazilian tapir, and a 2005 workshop of tapir experts (Taber et al. 2006) estimated that over 120,000 km2 of the Pantanal remains potential tapir habitat, representing a priority ���Tapir Conservation Unit��� (TCU) for the survival of the species over the next 100 yr. Given that over 95 percent of the Brazilian Pantanal is occupied by private cattle ranches, the Pantanal TCU is considered to have only a medium probability of maintaining tapir Received 14 August 2006 revision accepted 21 March 2007. 1 Corresponding author Current address: Pile All�� e 19a 2tv, 2000 Frederiksberg, Denmark e-mail: mail@mogenstrolle.dk populations over the long term, requiring appropriate conservation actions to be taken by landowners. No studies to date, however, have estimated actual tapir densities in the region, even though such information is of decisive importance in assessing the role of the Pantanal in protecting the tapir and in guiding management decisions. We studied Brazilian tapir density at a site in the northeastern Pantanal using two independent methods: camera trapping com- bined with capture���recapture analysis and line-transect sampling. The camera-trapping methodology is relatively new, but has been a revolution within the study of abundance of secretive, nocturnal, and/or rare mammal species in the Neotropics (Noss et al. 2003 Trolle & K�� ery 2003, 2005 Maffei et al. 2004, 2005 Silver et al. 2004 Sanderson & Trolle 2005 Trolle et al. 2006). Researchers working on three different tapir species have presented abundance indices based on capture frequency from camera-trap records (see Wallace et al. 2002, Holden et al. 2003, Kelly 2003, Moraes Junior et al. 2003, Silveira et al. 2003, Trolle 2003). Such abundance in- dices are problematic because they do not distinguish cases of single individuals with numerous records from cases of numerous individ- uals with single records. Maffei et al. (2002) and Noss et al. (2003), C 2008 The Author(s) Journal compilation C 2008 by The Association for Tropical Biology and Conservation 211

212 Trolle, Noss, Cordeiro, and Oliveira working in Bolivia, showed that systematic camera-trapping sur- veys are also useful for estimating population density of the Brazil- ian tapir, through identification of individuals from camera-trap photographs and the application of capture���recapture analysis. The second method, line-transect sampling, has been a popular and widely used technique for estimating abundance of biological populations in a variety of habitats and situations (Wilson et al. 1996, Buckland et al. 2001), including a number of terrestrial her- bivorous mammals (Southwell 1994, Khan et al. 1996, Rudran et al. 1996, Heydon & Bulloh 1997, Karanth & Nichols 1998). This is the first study to simultaneously apply both methods on the same species at the same site to estimate population density, and therefore provides a unique opportunity to compare the two methods. METHODS STUDY AREA.���The study took place in the private reserve Estancia �� Ecologica �� SESC Pantanal, situated between the Rio Cuiab�� a river and the tributary Rio S�� ao Lourenc ��o in the Bar�� ao de Melgac ��o region, the northeastern Pantanal, Mato Grosso, Brazil. For the camera trapping, an area of approximately 54 km2 was studied in the northeastern corner of the reserve (field stations at 16���42.655 S, 56���01.648 W and 16���41.201 S, 56���10.486 W, re- spectively). For the line-transect sampling, almost the entire reserve was studied, i.e., 1063 km2. Both studies covered habitats represen- tative of the region. The landscape is a mosaic of closed and open as well as permanently dry and seasonally inundated habitat types, in- cluding gallery forest (closed, seasonally inundated), semi-deciduous forest with the understory dominated by acuri palms (Attalea phaler- ata) (closed, permanently dry), Cerrado woodland (dominated by low, gnarled trees) (open, permanently dry), scrubland (open, per- manently dry), and grassland (open, seasonally inundated). During the rainy season (usually a 3-mo period around December through FIGURE 1. Camera-trap distribution for SESC Pantanal tapir survey. Gray dots indicate where tapirs were photographed and number of individuals photographed at each location: smallest = 1 individual, medium = 2, large = 3. February) large areas flood, although by the end of the dry season, when the camera trapping took place (October���December 2002), no permanent water occurred in the study area except for a number of artificial water holes. Typical of the Pantanal (Trolle 2003), which consists mainly of private cattle ranches, in the past the study area has been affected considerably by ranching-related activities such as dry-season fires to promote growth of introduced pasture. Since the establishment of the SESC reserve in 1998, the cattle have been removed from this area facilitating natural succession and resulting in a high cover- age of grass, dense scrub, and thick forest undergrowth. Currently, tapirs are not hunted in the area. The mammal assemblage includes seven other ungulates (Tayassu pecari, Pecari tajacu, Mazama ameri- cana, M. gouazoubira, Blastoceros dichotomus, Ozotoceros bezoarticus, and Sus scrofa), three large herbivorous rodents (Hydrochoerus hy- drochaeris, Cuniculus paca, and Dasyprocata azarae), five edentates (Dasypus novemcinctus, Euphractus sexcinctus, Priodontes maximus, Myrmecophaga tridactyla, Tamandua tetradactyla), and six carnivores (Cerdocyon thous, Chyrsocyon brachyurus, Nasua nasua, Procyon can- crivorus, Leopardus pardalis, and Puma concolor) (Trolle & K�� ery 2005, Trolle et al. 2006). SAMPLING.���The general framework of our camera-trapping survey followed Karanth and Nichols (1998) and Nichols and Karanth (2002). The study area was divided into four subareas roughly equal in extent (Fig. 1). In each subarea, 14 trapping stations were camera trapped for nine consecutive nights. We covered the study area with a grid configuration, with trapping stations spaced approximately 1 km apart and situated at sites where tapirs and other mammals were likely to pass: on dirt roads (N = 29) rarely used by vehicles, and on animal trails (N = 27). Camera-trap placement was not intended to be random but rather sought to maximize capture probability of the target species (in accordance with Karanth & Nichols 1998, Nichols & Karanth 2002, Silver et al. 2004). The 56