Introduction

The golden-headed lion tamarin (Leontopithecus chrysomelas) inhabits wet coastal and inland semi-deciduous forests in the northern Atlantic forest, extending through southern Bahia and, in the past, northwest Minas Gerais (Pinto and Rylands 1997; Raboy et al. 2010). It is classified as Endangered on the IUCN Red List due to habitat loss and fragmentation resulting from conversion of forest and shade-cocoa agroforest to cattle pasture or other agricultural crops. It was believed that L. chrysomelas inhabited altitudes mostly below 300 m above sea level (Coimbra-Filho 1969; Hershkovitz 1977; Rylands et al. 1993). Pinto and Rylands (1997) found L. chrysomelas as high as 400 m but supposed it improbable that L. chrysomelas would use elevations higher than 500–550 m because of changes in climate, floral communities and forest physiognomy. Areas in the L. chrysomelas geographic distribution include elevations up to 1,100 m (Fig. 1). The question thus remains, to what extent (altitudinal limit, frequency and type of use) do L. chrysomelas use the higher elevation habitats? The golden lion tamarin (L. rosalia), another coastal, but more southerly species, has now been found at elevations of up to 550 m (Kierulff and Rylands 2003). The black lion tamarin (L. chrysopygus), occurring on the inland plateau of the state of São Paulo, has been recorded at elevations of 700 m (Coimbra-Filho 1970) and 900 m (C. Knogge pers. comm.).

Here we report on the occurrence of L. chrysomelas in areas above 500 m on four occasions in different areas of their distribution. We also present a frequency histogram of elevation use from two study groups followed at higher elevations. We discuss reasons why L. chrysomelas might be seen at higher elevations and the conceptual implications of higher-elevation use for the development of habitat and landscape models implemented to assist in conservation planning for this species.

Methods

We compiled results from two different studies conducted by the authors in southern Bahia: the “GHLT Connection” and the “Cabruca Project.” Researchers in the GHLT Connection conducted a survey of L. chrysomelas throughout the species' known historic distribution. The area, shown as the polygon outlined in black in Figure 1, included forests between the Rio de Contas and the Rio Jequitinhonha, from the coast westward toward the region of the rios Gongoji, Acará, Catolé Grande and Ribeirão do Salto. Researchers in the Cabruca Project studied the behavior and ecology of L. chrysomelas groups in shade-cocoa (“cabruca”) agroforest. The Cabruca Project was carried out in two phases. The first was a survey of the shade-cocoa region in the east of the range of L. chrysomelas to select study sites, and the second involved the study of radio-collared groups of L. chrysomelas in the study locations chosen. L. chrysomelas groups were followed on multiple days in the municipalities of Camacã, Una, Ilhéus, Jussari and Arataca.

The two projects implemented varying overall experimental designs. The GHLT Connection surveyed forest patches selected by stratified random sampling between November 2005 and November 2007 using playback methods outlined in Raboy et al. (2010). When L. chrysomelas were sighted, a GPS point was taken. For the most part, elevations higher than 400 m were not sampled, presuming L. chrysomelas would not be found in these areas, but occasionally points along transects reached these elevations and higher. The Cabruca Project first surveyed for possible long-term monitoring sites between June and August of 2006 and 2007 based on results from the GHLT Connection, word of mouth regarding possible locations of L. chrysomelas, and additional playback work. Following that, selected L. chrysomelas groups were monitored with radio-telemetry between April 2008 and September 2009. Two of the seven study groups ranged in areas with elevations above 500 m. At 20-min intervals, a group's geographic position and altitude were recorded using a GPS device. UTM coordinates were collected using Corrégo Alegre datum (UTM Zone 24L) for both projects. The altitude of observations was determined by measuring elevation at the location of observation with the GPS altimeter and by cross referencing UTM coordinates (re-projected from the Corrégo Alegre datum to South American 69) with a Shuttle Radar Topography Mission (SRTM) elevation map of the study region (South American 69; data courtesy of NASA/ NGA/USGS at  http://www2.jpl.nasa.gov/srtm/).

For the first part of our investigation, we noted all observations from each of the two studies documenting L. chrysomelas above 500 m. We subsequently used the results of the Cabruca Project to determine histograms of elevation use for the two high-elevation study groups of L. chrysomelas. For this we determined the number of 20-minute observations that occurred in each 100 m altitude class to ≥700 m.

Results

We registered L. chrysomelas above 500 m at four different localities in three different municipalities of Bahia (Fig. 1):

1) Floresta Azul (from GHLT Connection). Two individuals were seen at 633 m (cross referenced at 600–700 m), responding to playback calls at the border between shade-cocoa and secondary forest. The group later moved to even higher altitude, although it was not possible to register a GPS point. 426166 E 8345747 N at maximum altitude measurable.

2) Arataca (from Cabruca project - site selection phase). A group of four individuals was recorded at 515 m (cross referenced at 400–500 m but <80 m Euclidean distance from the 500 m contour) in primary forest. 463817 E 8319996 N.

3) Arataca (from Cabruca Project - monitoring phase). A group of eight individuals including two infants (approximately one month old) were observed using a maximum altitude of 551 m (cross referenced at 500-600 m) in primary forest. 455530 E 8323006 N at maximum altitude observed.

4) Camacã (from Cabruca Project - monitoring phase). Two males were observed at a maximum altitude of 650 m (cross referenced at 600–700 m) in primary forest. 439756 E 8302606 N at maximum altitude observed.

The majority of the 20-minute observations for the Cabruca Project for one reproductive group (“Bem Te Vi”) were in the 300–400 m elevation category (Fig. 2). On two of 14 days of observation, the group used altitudes above 500 m. For another group (two males; “São José”), the majority of observations were also in the 300–400 m elevation category, although the two males used five elevation classes (from 200 m to 600 m; Fig. 2). On three of seven days of observation they ranged to altitudes above 500 m. On one of those days, individuals in the São José group spent the entire day above 400 m in cabruca and primary forest, using a sleeping hole also above 400 m. Both groups of L. chrysomelas used slopes spanning at least four elevation classes or 400 m differential from highest to lowest observations.

Discussion

We documented four L. chrysomelas groups in different geographic regions using forests at 500–700 m altitude, the highest altitudes yet published for the species. Our findings imply several possibilities in relation to the previous suggestion that lion tamarins use only lower elevations. First, the hypothesis that resources are inadequate at higher elevations may be incorrect for the levels at which L. chrysomelas were found. In a study focusing on the avifauna of southern Bahia, Silveira et al. (2005) indicated that the vegetation became markedly stunted at approximately 800 m across their study sites in the Serra dos Lontras and Javi ranges. Up to this point, the forest still comprised tall trees and bromeliads (Silveira et al. 2005) and, at least in physiognomic terms, might be favorable to lion tamarins from what we know of their needs. In a botanical inventory of three montane areas in southern Bahia, Amorim et al. (2009) recorded 1,129 plants species at altitudes 300 to 1080 m above sea level. Seventeen species on this list were also present and classified as “extremely valuable” on a list of key resource species for L. chrysomelas by Oliveira et al. (2010) from a lower-lying forest (<100 m altitude). An additional 25 species were listed as “key” (useful but to a lesser degree than the “extremely valuable”; Oliveira et al. 2010). Amorim et al. (2009) also recorded high bromeliad diversity above 400 m. Golden-headed lion tamarins forage for animal prey most commonly in bromeliads (Oliveira et al. 2011; Raboy and Dietz 2004). We documented L. chrysomelas above 500 m at two locations of the Amorim study. The results from Silveira et al. (2005), Amorim et al. (2009), and the Cabruca Project documenting golden-headed lion tamarin groups eating and sleeping above 400 m are suggestive that L. chrysomelas might have adequate resources at these and higher elevations.

Figure 1.

Elevation map of southern Bahia. A light grey line delineates the boundary of Bahia state. The black polygon represents the former distribution of L. chrysomelas and the area sampled for the GHLT Connection Project (Raboy et al. 2010). The four L. chrysomelas sightings above 500 m are shown by the points outlined in squares.

Figure 2.

Elevational distribution of observations from two groups of L. chrysomelas in the Cabruca Project.

Second, a shift may have occurred in the species elevation-use patterns in response to anthropogenic change in the region, and L. chrysomelas could be using higher elevations despite their poorer resources. Kierulff and Rylands (2003) suggested that the presence of L. chrysomelas above 500 m in the Serra do Mar in the state of Rio de Janeiro was the result of deforestation at lower levels that had pushed populations into more mountainous areas. Groups were seen at these higher elevations but near to houses where they had access to cultivated fruits such as bananas. Silveira et al. (2005) believed that extensive deforestation at lower elevations also explained the presence of some bird species found in Bahia at altitudes higher than documented in other locations. Historical evidence of the absence of lion tamarins in higher elevations currently used by them would indicate a shift towards higher elevation, but this information does not exist. Comparative studies of foraging and reproductive success of L. chrysomelas ranging exclusively at higher and at lower elevations along with corresponding phenological studies to estimate food availability would help elucidate the patterns in elevation use seen in this species.

With the exception of the studies we have mentioned, little information exists indicating the effect of altitudinal gradient on potential L. chrysomelas resources in southern Bahia. Other callitrichid genera have been documented using higher elevations in the Atlantic forest. Callithrix geoffroyi, also thought to be a lowland species (500 m; Passamani and Rylands 2000; Rylands and Faria 1993) with a similar diet to lion tamarins (except, principally, its exploitation of plant exudates when fruits are scarce), has been found in the Estação Biológica de Santa Lúcia, a reserve ranging from 550–950 m in Espirito Santo (Passamani et al. 2000) and at 1274 m in the Serra do Cipó National Park, Minas Gerais (Oliveira et al. 2003). Pinto et al. (2009) found that elevation was one of the five most important predictors of species density for three (Brachyteles, Cebus and Callithrix) of five focal species in a study of primates (including Alouatta, and Callicebus) throughout four Brazilian states: São Paulo, Rio de Janeiro, Minas Gerais and Espírito Santo. For Brachyteles, Cebus and Callithrix, the relationship between species density and elevation was negative, but elevation was less consistent as a predictor of density for all five species as compared to precipitation and temperature (Pinto et al. 2009).

Sightings of L. chrysomelas groups in higher altitudes do not necessarily mean that they use resources or reproduce in these areas. A third explanation is that they continue to be low-elevation species and only use higher-altitude forests for travel or dispersal, traversing slopes and peaks with limited or absent resources for the same reasons they cross open fields (B. Raboy pers. obs.). In the GHLT Connection, we observed that higher elevation forests tended to be better preserved than many of those in lower elevations. Summits at higher elevation in the west often stood as forest islands surrounded by cattle pasture. The steeper terrain and rise in elevation decreases its accessibility or suitability for certain forms of agriculture. Moreover, Brazilian legislation (Forestry Code/Federal law 4771/65 and CONAMA resolution 303/02) considered areas of steep terrain slopes (>45 degrees), hill and mountain tops (above 2/3 height in relation to the base) and high altitude (>1800 m) as Areas of Permanent Protection (APP; CONAMA 2002). APPs must preserve the original native vegetation and may not be used for production (Sparovek et al. 2010). Although L. chrysomelas resource quality has not been quantified in these areas, forested hilltops certainly provide L. chrysomelas with cover and protection from predators.

Increasing the known upper elevational limit to which L. chrysomelas finds resources, reproduces or travels within has implications for conservation planning in that it increases the available habitat. Assuming the higher-elevation habitat is suitable for finding resources and breeding, it provides refugia from the degradation and fragmentation of the lowland forests. If high elevation forests serve only as a conduit—a corridor—for dispersal and gene flow between lower-lying populations, this still has strong conservation implications, increasing the potential functional connectivity of existing fragments. Increased connectivity facilitates gene flow in the metapopulation, which is at present extremely fragmented (Raboy et al. 2010; Zeigler et al. 2010).

Prior conservation modeling predicting future L. chrysomelas abundance excluded the possibility that L. chrysomelas used forest above 400 m. Holst et al. (2006) conducted a Population and Habitat Viability Analysis (PHVA) for populations of L. chrysomelas in two areas containing high elevations—the Serra do Baixão and Serra das Lontras. While the estimated overall areas were 32,089 ha for Serra do Baixão and 8,015 ha for Serra das Lontras, the areas deemed suitable for L. chrysomelas were only 13,782 ha and 1,668 ha respectively, principally due to the large amount of forest above 400 m elevation in these locations. If L. chrysomelas uses elevations greater than the upper limit of 400 m for maintaining territories and breeding, the predicted outcomes for population size and probability of maintaining genetic diversity in those locations could be considerably underestimated. Recent work with howler monkeys (Alouatta pigra) in cloud forest of Guatemala indicated use of much higher altitude than previously thought. The authors suggest these regions will become important for species conservation and must be included in updated estimates of the species range (Baumgarten and Williamson 2007).

Figure 3.

Map showing the remaining forest in the L. chrysomelas distribution by four elevation classes. Forest cover was determined by Sara Zeigler, based on interpretation of 2007 Landsat images (see Zeigler et al., 2010 for more details).

Superimposing a reclassified elevation map (0–300 m, 300–500 m, 500–700 m and >700 m) on a forest cover map of the L. chrysomelas range elaborated by Zeigler et al. (2010), we determined that forested areas under 500 m in the L. chrysomelas range (880,179 ha) represented 91.2% of the total forest cover including all elevations (965,861 ha; Fig. 3). Forested areas between 500 m and 700 m were 6.2% and those at elevations above 700 m were 2.6% of the total forest cover. Thus, the additional area gained, considering all forest up to 700 m as suitable for L. chrysomelas (rather than the <500 m model), increases potential forest for the species by 6.8%.

Functional connectivity also increased slightly, particularly north of the Rio Pardo, when forest at elevations above 500 m could serve as corridors for dispersing L. chrysomelas (S. Zeigler unpubl.). While high elevation areas make up only a small portion of the L. chrysomelas range, the location of these areas is significant. Many of them are located centrally in the L. chrysomelas distribution (Fig. 3) in areas that currently harbor L. chrysomelas and other threatened biodiversity (SAVE Brasil et al. 2009), such as those analyzed in the PHVA. These regions are regarded as having significant conservation potential as part of a network of reserves recently created or proposed (SAVE Brasil et al. 2009). Floresta Azul (Sighting 1) is within the Serra dos Barbados range that rises to approximately 800 m, with cabruca forest occurring up to 700 m. The region of Arataca where L. chrysomelas was found (Sightings 2 and 3), is part a chain of mountains that includes the Serra das Lontras, the Serra dos Quatis and the Serra Javi where the maximum altitude is nearly 1,000 m. Cabruca is prevalent up to altitudes of 600 m. The region of Camacã where L. chrysomelas was found (Sighting 4) is located in the Serra do Baixão chain (west of Lontras) that rises up to 900 m and is composed of a mosaic of vegetation ranging from open pasture to mature forest. Cabruca is a habitat known to be used by L. chrysomelas and provides suitable resources throughout the species' range (Oliveira et al. 2010, 2011; Raboy et al. 2010).

It is evident that L. chrysomelas can be found in higher elevations than previously thought, though exactly how they use these areas is still unclear. While observations of L. chrysomelas above 500 m are still seemingly rare, it is important to note that the two projects assessed in this paper for the most part avoided sampling for L. chrysomelas in higher altitude areas presuming the species would not be present. We predict that future systematic sampling for L. chrysomelas in elevations of 500 m to 700 m will yield a greater number of sightings than was documented from our ad libitum visitation of this elevation. Forested slopes and ridges serving as corridors (at least up to about 600–700 m) may greatly contribute to increasing the connectivity of the L. chrysomelas metapopulation and should be investigated further. Specifically, future studies are necessary to evaluate what limits the use of higher elevations by L. chrysomelas, what is truly “too high” for lion tamarins, and how ecological parameters (home range size, habitat use, sleeping site use and dispersal patterns) and social and demographic characteristics (group size, composition and biomass) of L. chrysomelas vary by elevation.