Sex matters: Modeling male and female habitat differences for jaguar conservation Dalia A. Conde a,b,c,*, Fernando Colchero a,c, Heliot Zarza d, Norman L. Christensen Jr. b, Joseph O. Sexton e, Carlos Manterola f, Cuauht��moc Ch��vez d, Antonio Rivera c, Danae Azuara g, Gerardo Ceballos d a Max Planck Institute for Demographic Research, Rostock 18057, Germany b Nicholas School of the Environment, Duke University, Durham, NC 27708, USA c Jaguar Conservancy, A.C. El Carmen 31, M��xico D.F., Mexico d Instituto de Ecolog��a, Universidad Nacional Aut��noma de M��xico, M��xico D.F., Mexico e NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA f Comisi��n Nacional de ��reas Naturales Protegidas, Tlalpan, M��xico D.F., Mexico g Anima Efferus, A.C. El Carmen 31, M��xico D.F., Mexico a r t i c l e i n f o Article history: Received 1 February 2010 Received in revised form 19 April 2010 Accepted 26 April 2010 Available online 31 May 2010 Keywords: Generalized linear model Use-availability habitat model Land cover Mayan Forest Large carnivore conservation Gender Roads Tropical forest a b s t r a c t Competition for mates, cub rearing, and other behaviors differ between males and females in large car- nivores. Although these differences can be reflected in patterns of habitat use, gender has rarely been incorporated into habitat models. We evaluated differences in habitat use between male and female jag- uars in the Mayan Forest of the Yucatan Peninsula by modeling occupancy as a function of land cover type, distance to roads, and sex. Nested models were fitted to high-spatiotemporal resolution satellite (GPS) telemetry, controlled for temporal autocorrelation, and eliminated selection bias of pseudo- absences using a semi-non-parametric bootstrap. Although both male and female jaguars prefer tall for- est, short forest was also preferred by females but avoided by males. Whereas females significantly avoided roads, males didn���t and ventured into low-intensity cattle ranching and agriculture. Females��� preference for intact forests and against roads led to their habitat being fragmented to a greater degree than that of males. Models that ignored sexual differences failed to capture the effect of roads and agri- culture on jaguar habitat use, blurred the distinction of use between short and tall forest, and underes- timated fragmentation of female jaguar habitat but incorporating these differences increased precision of habitat maps and allowed the identification of potential jaguar-human conflict areas associated with male���s use of cattle and agricultural lands. Specifying sex differences increases the power of habitat mod- els to understand landscape occupancy by large carnivores, and so greater attention should be paid to these differences in their modeling and conservation. �� 2010 Elsevier Ltd. All rights reserved. 1. Introduction Human encroachment on the last remaining wildernesses has challenged the long-term conservation of large carnivores (Woodroffe, 2000). These species are in global decline due to habitat loss, fragmentation, and hunting for fur trade and livestock protection (Weber and Rabinowitz, 1996 Ogada et al., 2003). How- ever, despite dramatic decreases in large carnivore populations worldwide, still little is known about many species��� environmental requirements and their associated habitat losses. Large carnivores frequently exhibit sexually distinct social behaviors and spacing patterns, especially those associated with gestation and parental care (Wilson, 1975 Crook et al., 1976 Eisenberg, 1981). Female cheetahs (Acinonyx jubatus), for example, tend to show stronger preference than males for denser vegetation types, which provide greater protective cover, higher densities of prey, and lower densities of competitors and cub predators (Broomhall et al., 2004). In addition, some species show clear sex- ual differences in their response to roads (Mace et al., 1996 Maehr, 1997 Kerley et al., 2002 Gaines et al., 2005). For example, roads are avoided and thereby delineate territories of female Florida pan- ther (Puma concolor coryi), but males cross roads more frequently and therefore suffer higher mortality rates from car collisions (Maehr, 1997). Likewise, studies on livestock depredation show that higher mobility of males and possible differences in diet pref- erences lead to males being the primary culprits for most attacks, resulting in a higher number of males killed by local people (Linnell et al., 1999). Although females have greater energy requirements during gestation and lactation periods, males have larger home ranges, possibly as a result of males��� farther dispersion and mate-seeking behaviors (Lindstedt et al., 1986). 0006-3207/$ - see front matter �� 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2010.04.049 * Corresponding author at: Max Planck Institute for Demographic Research, Rostock 18057, Germany. Tel.: +49 381 2081 267 fax: +49 381 2081 202. E-mail address: Conde@demog.mpdr.de (D.A. Conde). Biological Conservation 143 (2010) 1980���1988 Contents lists available at ScienceDirect Biological Conservation journal homepage: www.elsevier.com/locate/biocon

In recent decades, habitat models have become a key tool for the design of conservation policies worldwide (Pearce and Ferrier, 2000 Gavashelishvili and Lukarevskiy, 2008). However, despite known sex-related differences in large carnivores, these models usually fail to incorporate gender as a variable. Although a few habitat models of large carnivores do focus on modeling females alone (e.g., Mace et al., 1996 Gaines et al., 2005), the majority of habitat models typically ignore sex as an explanatory variable (e.g., Palma et al., 1999 Schadt et al., 2002 Linkie et al., 2006 Mat- thew et al., 2006 Klar et al., 2008). When behavioral differences do exist between the sexes, disre- garding them in models can blur behavioral preferences between the sexes and bias their estimation toward that of the more fre- quently observed sex. For those species for which there are signif- icant intersexual differences specifically in habitat use, lack of discrimination could result in underestimating important environ- mental factors for one sex while overestimating them for the other. To evaluate gender differences in habitat use, we developed a model that discriminates male and female preferences in jaguar (Panthera onca) habitat occupancy. We tested the hypothesis that males are more tolerant to human activity than females, as re- flected in the probability of their occurrence as a function of land-use type and proximity to roads. In particular, we predict that: (1) both will use undisturbed forests as their preferred cover type but (2) males will occur with higher probability in human- dominated land-use types (i.e., low-intensity agriculture and rangeland) and in closer proximity to roads than females. This analysis quantifies the distribution and fragmentation of available habitat for male and female jaguars in the Mayan Forest region and identifies likely human-jaguar conflict zones for the development of conservation policies. This study therefore de- scribes the habitat use of one of the least studied large carnivores in the world, and provides a fundamental tool for wildlife conser- vation and management in a region where pressure from infra- structure projects is rapidly increasing. 2. Methods 2.1. Study area and species This study was conducted in the Mayan Forest, which holds the largest jaguar population and the largest patch of tropical forest in Central America. The area is one of the priority jaguar conservation units (JUC) defined by the Wildlife Conservation Society Jaguar Conservation Program (Sanderson et al., 2002). Located in the core of the Mayan Forest, the study area includes central and northern Belize, a large portion of the Mexican states of Campeche and Quintana Roo, and the majority of the Pet��n of Gua- temala. The region is covered predominantly by tropical semi- deciduous forest, with an average annual rainfall of 1350 mm, a pronounced dry season between February and June, and relatively flat topography ranging between 0 and 700 m above mean sea le- vel (Holdridge et al., 1971 Pennington and Sarukhan, 2005). The area also contains subtropical moist forest (Holdridge et al., 1971) composed of two subtypes (1) tall forests, found in areas of greater relief and characterized by a high, closed tree canopy and (2) short forests, which are mostly seasonally inundated dry- deciduous (i.e., ������raingreen���) lowland alluvial forests with short, relatively open canopies and dense understories (Novack et al., 2005). The region also includes secondary vegetation which grow- ing on abandoned lands previously cleared for slash and burn agri- culture (Ucan et al., 1999) and swamps, which are classified as Mesoamerican palustrine vegetation (Garc��a and Secaira, 2006). Agriculture and cattle ranching are practiced at varying intensi- ties in the region. Low-intensity agriculture and cattle grazing are dominant, especially in the communal forests and lands (ejidos) and bordering legally protected areas. Broad-scale, mechanized agriculture and cattle grazing are rare and extremely localized. The jaguar is the largest Neotropical felid and the third largest cat worldwide, but it is the least studied of the big cats. The species is considered near threatened (IUCN, 2009), and although it has a wide distribution from Mexico to Argentina, the species maintains only 54% of its historical range (Kinnaird et al., 2003). Jaguars are solitary, generalist predators, with foraging strategies for selected prey species (Weckel et al., 2006). Conflict between jaguars and humans is increasing. Male jag- uars are responsible for approximately 70% of cattle predation in the region. This results in greater male mortality due to hunting by humans (Rabinowitz, 1986). 2.2. Jaguar data We captured jaguars in the borders of Calakmul Biosphere Re- serve and the Caoba forestry Ejido (Mexico) during the dry seasons of 2001���2007 and in the borders of the Mayan Biosphere Reserve (Guatemala) during the 2006���2007 dry seasons. The captured jag- uars were chemically anaesthetized using a projectile dart (Cebal- los et al., 2002), examined for general body condition, measured, weighed, and fitted with GPS collars (Televilt, Lindesberg, Sweden, see URL http://www.televilt.com). The GPS collars were pro- grammed to record animal positions at varying time intervals over 2���12 months, depending on collar battery life. Most of the individ- uals were recaptured the year following their initial capture to re- cover data and replace collars. We used a total of 5246 high-resolution GPS points from three females captured in Ejido Caoba and three males captured at the limits of the Calakmul and Mayan Biosphere Reserves (Fig. 1). All the individuals captured were adults, with ages ranging from 4 to 10 years old. None of the jaguars were captured close to broad-scale mechanized agriculture and broad-scale ranching activities. Temporal autocorrelation among locations can lead to biased estimates of environmental effects and inflated parameter vari- ances, making comparisons between alternative models difficult (Lennon, 1999). We avoided this by calculating the cross-correla- tions of latitude and longitude within serial jaguar locations at lags from 1 h to 7 days (Legendre and Legendre, 1998) and filtered points based on an empirical cross-correlation threshold of 0.3 (i.e., 72 h). This resulted in a sample size of 444 independent obser- vations of jaguar presence 218 points for females and 226 points for males. A random sample of 10,000 pseudo-absences were se- lected from the minimum convex polygon enclosing each individ- ual���s full (unfiltered) GPS trail, buffered by 10 km using the Hawth���s Analysis Tools extension in ArcMap 9.1 (Beyer, 2004). 2.3. Landscape data We obtained geospatial environmental data from the Selva Maya Zoque y Olmeca database (Garc��a and Secaira, 2006), includ- ing vegetation, population centers, and roads for the year 2000. We included in our analysis only selected paved and unpaved roads with high traffic, since small trails are known not to affect the behavior of large carnivores (Mladenoff and Sickley, 1998 Zarza et al., 2007). Based on previous studies (Chavez Tovar, 2006 Zarza et al., 2007 Conde, 2008), we used three variables that could potentially influence jaguar occurrence in the region: land cover (L), distance to roads (R), and sex (S). Preliminary analysis indi- cated that road type (i.e. paved versus non-paved), vegetation, sea- sonality, and elevation did not significantly affect jaguar habitat use (Conde, 2008). Although elevation has been a significant vari- able to model jaguar and other large carnivore habitat (Gaines et al., 2005 Mace et al., 1996 Hatten et al., 2005), we did not include D.A. Conde et al. / Biological Conservation 143 (2010) 1980���1988 1981