The conversion of forests for agriculture is a major cause of tropical forest fragmentation and degradation (Gibbs et al., 2010). The resulting landscape of smaller forest fragments embedded within a generally inhospitable matrix represents particular challenges to animals, especially rendering matrix-intolerant species more vulnerable to extinction (Benchimol and Peres, 2014). Matrix permeability critically determines dispersal between fragments, which in turn is essential for maintaining a sufficient level of gene flow and genetic diversity among small fragmented populations (Donald and Evans, 2006).

Forest-dwelling primates are particularly sensitive to the effects of anthropogenic forest fragmentation, with many species experiencing population decline or extinction (Benchimol and Peres, 2014). Loss of habitat due to agricultural expansion is the principal threat to primate populations worldwide, with livestock farming and ranching constituting the second most important threat in the Neotropics (Estrada et al., 2017). Yet high human population growth and poverty rates in primate-range countries, as well as global market economics, increase the likelihood of continued land-cover changes, forest loss and degradation (Estrada, 2013; Estrada et al., 2017). Documenting the ability of primates to use the matrix between fragments and identifying agricultural land-use practices compatible with primate survival, reproduction and/or dispersal across the landscape is imperative for developing guidelines for landscape management that reconcile primate conservation with subsistence of local communities (Estrada, 2013; Flesher, 2015).

Golden-headed lion tamarins (Leontopithecus chrysomelas) are frugivorous group-living primates of the Atlantic Forest of south Bahia (Brazil) that are classified as endangered and under threat from habitat loss, fragmentation and degradation (Kierulff et al., 2008). The landscape in south Bahia (Brazil) is a mosaic of Atlantic Forest remnants embedded within an agricultural matrix dominated by shaded cacao (Theobroma cacao) agroforests, locally known as cabruca (shaded cacao plantations in which the middle and understory trees of intact forests are removed and replaced with cacao trees; Alger and Caldas, 1994). Other economically important crops are present within the matrix, namely bananas (Musa paradisiaca) and rubber (Hevea brasiliensis), in addition to lesser quantities of coconut palm trees (Cocos nucifera) and manioc (Manihot esculenta) (IBGE, 2014). Golden-headed lion tamarin home ranges can encompass mature and degraded forests, in addition to cabruca (Raboy and Dietz, 2004). To date, no studies have reported the species using any other type of vegetation within the matrix, although the use of non-monoculture rubber plantations is suspected by former researchers in the area (B. E. Raboy, personal communication, 2014). Given that the long-term survival of this species may depend on its ability to use the matrix between fragments across its geographic range, this study aimed to investigate this potential use of rubber plantations.

We report on the presence of a group of golden-headed lion tamarins in a rubber plantation on a farm in south Bahia, thus constituting the first systematic observation of the species using this form of land use within the matrix; and we provide preliminary information on the group's behaviour and use of space in a fragmented landscape dominated by agricultural land use.

The fieldwork was conducted on a 29.4 ha privately owned farm, Fazenda Santo Antônio (-15.283796∘; -39.134477∘) in the municipality of Una (Bahia), in south Bahia. Approximately 35 % of the farm comprises patchy evergreen forest, while the remaining 65 % is planted with perennial rubber, cupuaçu (Theobroma grandiflorum), soursop (Annona muricata), pepper (Piper nigrum), bananas and ephemeral crops such as manioc and garden vegetables. The farm also includes a 30-year-old rubber plantation of approximately 2.5 ha surrounded by secondary forest patches, agricultural plantations and open fields. Many of the older rubber trees bear epiphytic bromeliads of various sizes. Additional trees, such as cacao, banana, jackfruit (Artocarpus heterophyllus) and guava (Psidium guajava), are interspersed with the rubber trees (Fig. 1).

Between August 2013 and January 2014, we spent 103.5 h in the field, 69.28 h of which were in the presence of the group (meaning we either had visual contact or were within a few metres of the spot where we knew them to be hiding). We saw the group in the rubber plantation on 16 of 22 observation days (73 %) during which we were able to establish visual contact (Table 1; Fig. 2). Additional vegetation types used were initial, medium and advanced secondary forest, and a regenerating field.

On 17 occasions (77 % of observation days), we were able to determine the group's sleeping site in the morning or evening. In 16 of these 17 occasions (94 %), the group used one particular individual palm tree locally known as “catulé” (Syagrus sp., Family Arecaceae) at the northern edge of the rubber plantation, bordering a passion fruit plantation (Table 1; Fig. 2).

We saw Wied's marmosets (Callithrix kuhlii) together with the group in the rubber plantation on 10 of 16 sightings (63 %), occurring either in the rubber plantation or in the small patch of degraded forest located to the north of the rubber plantation (Table 1; Fig. 2). We observed the group of golden-headed lion tamarins using both this northern edge of the rubber plantation, where the palm tree sleeping site is located, and the area along a central line running north to south through the rubber plantation. We also observed the group in a patch of advanced secondary forest to the southeast of the rubber plantation encountering another group of golden-headed lion tamarins just across the border on the neighbouring property. We recorded the group eating in a jackfruit tree in the centre of the plantation. We additionally observed allogrooming, feeding and scent-marking behaviour during waking hours in the rubber plantation, in addition to hiding and travelling.

This study is the first to provide evidence of the use of a rubber plantation in south Bahia by golden-headed lion tamarins. We observed rubber plantation use on the majority of study days. On several occasions, we saw the group in association with Wied's marmosets and observed behaviours such as eating, sleeping and grooming, thus suggesting that the group may have been using the plantation as part of its home range and not merely as a corridor between fragments. We further observed selective site use, in the sense that the group used some areas of the rubber plantation more than others; however, given the preliminary nature of our observations and the limited number of observation days (due to being early in the habituation process), further investigation is needed to confirm this pattern. We are currently following this and two more groups to study how they use rubber plantations spatially (entire area vs. edges; specific pathways) and to obtain detailed information on their resource use and behaviour.

Golden-headed lion tamarins have long been thought to depend on mature forests for survival, due to their requirement of resources such as epiphytic bromeliads for foraging and tree holes for sleeping (Rylands, 1996). More recent studies have reported the species being able to survive and reproduce in secondary forests, including cabruca agroforests (Catenacci et al., 2016; Oliveira et al., 2011). Being predominantly a frugivore, the species' diet varies between areas, reflecting differences in floristic composition and plant density; this indicates that the species adapts easily to changes in fruit availability (Catenacci et al., 2016). The extent to which the availability of sufficient sleeping sites and/or foraging sites limits its capacity for exploring altered habitats is as of yet undocumented. Therefore, maintaining a forest mosaic that includes patches of forest providing these resources is considered important (Catenacci et al., 2016). The rubber plantation in this study is relatively old (approx. 30 years) and infrequently managed through annual removal of undergrowth. With respect to potential resources for golden-headed lion tamarins, several of the older rubber trees bear epiphytic bromeliads, providing potential microhabitats for insect foraging (Raboy and Dietz, 2004). Bromeliads are listed as principal foraging sites at all study areas where golden-headed lion tamarins have been observed (Raboy and Dietz, 2004; Guidorizzi, 2008; Oliveira et al., 2011; Catenacci et al., 2016). Although we did not record our study group foraging in bromeliads in the rubber plantation during the observation period described here, we observed this behaviour at a later time. We suspect the lack of bromeliad foraging during the study period to be due to the habituation process, which frequently caused the group to hide, usually in bromeliads, and made the visualisation of feeding and foraging behaviour more difficult.

In addition to bromeliads, the rubber plantation contains some native fruit trees from plants included in the species' diet (e.g. Tapirira guianensis; Oliveira et al., 2010) and jackfruit trees. Jackfruit, an exotic species introduced in Brazil in the 18th century, is widely distributed along the Brazilian coast (Correia, 1975). Many animal species consume its fruits, including Leontopithecus chrysomelas; the consumption of jackfruit has been observed at all sites in the ombrophyllous forest where this primate has been studied (Raboy and Dietz, 2004; Oliveira et al., 2011; Catenacci et al., 2016).

All of these resources may attract golden-headed lion tamarins to these plantations and could therefore be used as a parameter to evaluate the suitability of rubber plantations for use by golden-headed lion tamarins and possibly Wied's marmosets. We advocate further investigation of this hypothesis, since the gained knowledge may help in defining management practices that aim at increasing the suitability of rubber plantations for both species.