Amazonian Small Mammal Abundances in Relation to Habitat Structure and Resource Abundance

By Lambert, Thomas D; Malcolm, Jay R; Zimmerman, Barbara L

Previous studies in tropical rain forests suggest that most small mammal species reach their highest densities in disturbed habitats; however, only a few sites have been examined. Consequently, habitat and resource use for many species is poorly understood. This is especially true in the Amazon Basin, where no studies of microhabitat associations of small mammals have been undertaken. We studied relationships with habitat variables and resource abundances for 5 species of marsupials and 9 species of rodents at a site in southeastern Amazonia. Small mammals were sampled with traps placed both on the ground and in the understory. Eight habitat variables were measured to quantify habitat structure. Measures of insect biomass were collected by the use of sticky traps, and fruit abundance was quantified. Patterns of habitat use were examined using logistic regression, multiple regression, and ordinations. Many species showed increased abundances with habitat features indicative of edge-affected or disturbed habitats, showing negative relationships with understory openness, understory woody-stem density, tree density, and tree size; and positive relationships with number of vines per tree, mean log size, number of logs, and volume of downed wood. We obtained support for the hypothesis that the cause of this pattern is increased resource abundances in these areas, because both insect biomass and number of fruiting trees showed similar relationships. However, for many species, measures of resource abundance were not important once habitat features were entered into the models, indicating that the relationship to resources is an indirect one.

Key words: Amazonia, habitat use, marsupials, resource abundance, rodents, small mammals, Xing Basin

Small rodents and marsupials in the families Muridae, Echimyidae, and Didelphidae can comprise up to 66% of the nonvolant mammal diversity in neotropical forests (Voss and Emmons 1996). These mammals have been shown to play important roles in forest ecology, acting as seed predators and dispersers (Adler and Kestell 1998; Asquith et al. 1997; Forget 1991; Hoch and Adler 1997), and changes in their abundances have even been shown to affect forest regeneration and succession (Terborgh et al. 2001). To date, most studies of small mammals in the Amazon Basin have focused on their systematics and patterns of overall abundance and species richness (e.g., Lambert et al. 2005a; Malcolm 1995, 1997; Patton et al. 2000; Voss and Emmons 1996). Although several studies have examined microhabitat associations of small mammals in Central America (Lambert and Adler 2000; Tomblin and Adler 1998), northern South America (Adler et al. 2000; Lambert et al. 2003), and Cerrado habitats (Henriques and Alho 1991; Lacher and Alho 2001), little is known about the habitat relationships of most Amazonian small mammal species, and what is known is often based on anecdotal observations (Eisenberg and Redford 1999; Emmons and Feer 1997).

These habitat relationships are of considerable interest given the increasing rate of anthropogenic disturbances in neotropical forests. Neotropical small mammals often increase in abundance and species richness after disturbances such as fragmentation and logging (Adler 1996a; Adler and Levins 1994; Lambert et al. 2003, 2005b; Malcolm 1997; Ochoa 2000), and such increases many seriously affect forest regeneration (Terborgh et al. 2001). These increases in some cases have been attributed to decreases in predation pressure (Adler 1996a; Adler and Levins 1994; Lambert et al. 2003; Terborgh et al. 2001); in others, to increases in resource abundance, especially insect biomass (Malcolm 1995, 1997; Ochoa 2000). However, it is difficult to distinguish between effects due to predation and those due to resource abundance. Small fragments are often assumed to be free of major predators (e.g., Terborgh et al. 2001); however, these fragments also have greater levels of edge habitat (Leigh et al. 1993; Lovejoy et al. 1986; Malcolm 1994). Edge habitat will likely not only increase resource abundance (e.g., Malcolm 1997; Malcolm and Ray 2000), but provide dense vegetation that also may afford protection from predators (e.g., Trejo and Guthman 2003). The response of species to resources also takes various forms, further blurring the distinction between resource- and predator-related effects. For example, a species might respond directly to resources, increasing in abundance as resources increase, or its habitat use may be indirectly related to resource availability, with preferential occupancy in habitats where resources are more likely to be abundant. Detailed studies of habitat use and relations of species to resources may aid in understanding the complex responses of small mammals seen after disturbance and provide insights into the trade-off between energy returns and risk of predation (Brown and Kotler 2004). Lambert et al. (2003) suggested that although reduced predation pressure led to the overall increase in rodent abundance seen after fragmentation, differences in habitat preferences among species accounted for which species became the dominant component on an island. Such results suggest that differential habitat use among species may be important in structuring tropical small mammal communities.

Local variation in forest structure has been shown to be correlated with variation in the abundance of several small mammal species in the tropics, suggesting that these species may use specific habitats for foraging, nesting, or both (August 1983; Malcolm 1995). Partitioning of vertical space among species also is thought to be important in structuring tropical small mammal communities (Bakker and Kelt 2000; see also Malcolm 2004); however, disturbed forests often show a relative lack of separation among species with respect to their height distributions (Lambert et al. 2005a; Malcolm and Ray 2000). Although differential exploitation of food resources also contributes to niche partitioning, it has been suggested that because insectivorous mammals are able to feed on nearly any size of insect, niche partitioning of this food resource is not possible (Dickman 1988). Similar arguments could be applied with regard to frugivory and granivory. Information on natural history, including patterns of habitat use, therefore is needed to better understand the ecology of tropical mammals in general and the effects of human disturbance on these systems (Ehrlich and Wilson 1991; Mares and Ernest 1995; Wilson 1988).

This study examines patterns of habitat use among 14 species of small mammals (5 marsupials and 9 rodents) at a site in southeastern Amazonia. In addition to measures of physical habitat structure, including forest structure, abundance of downed woody debris, and ground cover, we also measured the abundances of key resources, including insects and fruit. This is the 1st study to examine correlations between habitat features and abundances for these species, and in a broader sense, is the 1st study in the Amazon region to specifically examine habitat and resource correlations for a broad spectrum of the small mammal community.

MATERIALS AND METHODS

Study site.-The study was conducted in southeastern Amazonia in the vicinity of the Pinkait Research Station, Par, Brazil (746’14””S, 5157’43”W). The Pinkait Research Station is within the Kayap Indigenous Area, 1 of 6 legally ratified indigenous territories controlled by the Kayap that together span 13 10^sup 6^ ha in the Xing River basin. These Kayap lands are to a large extent pristine and consist of approximately 70% forest and 30% Cerrado (Zimmerman et al. 2001). The region hosts an intact flora and fauna that includes numerous rare and endangered species. The small population of indigenous Kayap in the area derive their living primarily from hunting, gathering, and small-scale agriculture. The Pinkait Research Station consists of an 8,000-ha reserve within the Kayap Indigenous Area that is protected from logging and hunting through an arrangement with the Kayap village of A’ukre. In 1984, the Kayap began selling big-leafed mahogany (Swietenia macrophylla) from their lands and logging has now occurred over large areas of the Kayap Indigenous Area surrounding the reserve. Overall, harvest intensity has been low and large tracts of undisturbed forest and savanna are common; however, areas where adult mahoganies were locally concentrated were subjected to extensive logging damage.

Forest structure at the site is heterogeneous in both vertical and horizontal dimensions and includes areas of tall-stature forest in addition to areas with a low canopy and dense understory vegetation (Lambert et al. 2005b; Peres and Baider 1997). Average annual rainfall in the town of Redeno, some 220 km from Pinkait, is 1,640 mm/year (Peres and Baider 1997). Rainfall measured at the site from 1 February 2002 to 31 January 2003 was 1,522 mm. Rainfall is highly seasonal with an intense 4-month dry season (<50 mm/month) lasting from the end of May until mid-September, followed by an intense rainy season with peak rainfall occurring in January and February. The study site is crisscrossed \with numerous seasonal streams. Elevation at the site ranges from 230 to 400 m above sea level (Peres and Baider 1997).

Small mammal trapping.-To assess habitat use, we established 19 trapping grids that spanned a full range of habitat conditions in the area, including 10 grids in unlogged forest within the Pinkaiti reserve and 9 grids at 2 nearby logged sites. Details on sites, grid placement, logging impacts, and additional trapping conducted in the area can be found in Lambert et al. (2005a, 2005b). Each grid consisted of 2 parallel transects 100 m in length set 50 m apart. Each transect contained 5 trap stations, spaced at 20-m intervals, for a total of 10 trap stations per grid and a total of 190 stations across the whole study area. Each trap station had 1 Tomahawk live trap (41 13 13 cm; Tomahawk Live Trap Company, Tomahawk, Wisconsin) and 1 Sherman live trap (23 9 8 cm; H. B. Sherman Traps, Inc., Tallahassee, Florida) placed on the ground and in the understory (tied with wire or a small bungee cord to vegetation 1.5- 2 m aboveground). Each grid was trapped 3 times during the course of the study, once in the dry season (end of May 2002 until the beginning of August 2002), once in the early wet season (October 2002 until November 2002), and once during the peak of the wet season (February 2003 to March 2003). Traps on each grid were set for 10 consecutive nights, for a total of 100 station-nights (400 trap-nights) per grid per sampling session. Thus, a total of 5,700 station-nights (22,800 trap-nights) were accumulated during the course of the study. Traps were baited with a combination of freshly ground peanuts, oatmeal, and ripe banana, and checked each morning.

At the time of capture, all mammals were identified, fitted with a uniquely numbered ear tag (National Band and Tag Company, Newport, Kentucky), weighed, and sexed. Because the taxonomy of many Amazonian species is poorly understood, voucher specimens of each species were collected. Because diagnostic characters are poorly understood for Oecomys species, most captured animals of this genus were collected. Taxonomic experts were consulted to aid in the identification of difficult genera (J. Patton for Proechimys species and R. Voss for Oecomys species, Akodon sp. nov., and Oryzomys species). Specimens are deposited at the Museu de Zoologia da Universidade de So Paulo, So Paulo, Brazil; the Museu Paraense Emlio Goeldi Belem, Brazil; and the Royal Ontario Museum, Toronto, Ontario, Canada. All handling and collection of the mammals used in this study was done in a humane and ethical manner with the prior approval of the University of Toronto’s animal care and use committee and met with the guidelines set forth by that organization and those of the American Society of Mammalogists (Animal Care and Use Committee 1998).

Habitat sampling.-A series of habitat variables was collected to quantify microhabitats at each trap station. For this purpose, four 10-m-long sampling lines were laid out from each station at 45, 135, 225, and 315 relative to the transect. Understory openness was measured at 5 m from the station along each of these sampling lines by use of 2.5-m pole banded in 10-cm segments; the number of bands unobstructed by vegetation as viewed from the station was recorded for each sampling point (see Malcolm and Ray 2000). Litter depth to the nearest millimeter was measured at 5 and 10 m along each sampling line. Sapling density was recorded as the number of woody stems > 1 m tall and < 5 cm diameter at breast height within 1.25 m on either side of the sampling lines. Woody understory stems (woody stems < 1 m tall) and herbaceous stems were quantified by counting the number of plants whose leaves fell directly below the sampling line. Diameter of all downed wood was measured at the point of intersection with the line for all logs > 5 cm diameter. Diameter at breast height and vine load (number of vines coming in contact with the tree) were recorded for all trees > 5 cm diameter at breast height within a 10-m radius of the trap station.

Resource abundances.-Insect abundance was sampled at each trap station with the use of a “yellow white fly and aphid” sticky trap (Seabright Labs, Emeryville, California) placed at breast height at each trap station. Sticky traps were left out for 5 nights during each trapping cycle. Total number of captured insects was tallied by size classes of <2, 2-<5, 5-<7, 7-<9, 9-<11, 11-<13, 13-<15, 15- <17, 17-<20, and 20+ mm, and insect biomass was estimated using the length-to-weight equation of Rogers et al. (1976; see also Malcolm 1997).

Number of fruiting trees was quantified on each grid by a Kayapo field assistant. During each trapping cycle, the individual walked the length of each transect and recorded the location (to the nearest trap station) and Kayapo name of all fruiting trees. Kayapo folk names in many cases correspond with scientific names, especially for tree species that produce fruits used by game species (Pinto 2003).

Data analysis.-Analysis of habitat use was undertaken at both the trap-station level (e.g., Adler 1996b; Adler et al. 1998; Lambert et al. 2003) and grid level (e.g., Malcolm 1995). For all analyses, we included only the 1st capture of an individual to avoid problems of statistical dependence of multiple captures of the same individual. Because our interests here concerned overall species-habitat relationships, and not logging effects per se, we pooled data from the logged and unlogged sites. Results on the effects of logging have been reported elsewhere (Lambert et al. 2005b).

Habitat variables.-Mean values for understory vegetation openness, litter depth, number of saplings, numbers of woody stems, numbers of herbaceous stems, total number of vines, total number of trees, tree basal area, tree diameter at breast height, number of logs, volume of down wood, and log size were calculated for each station and grid. To better quantify the diverse forest types seen at Pinkaiti, we additionally calculated the ratio of woody understory to herbaceous understory stems and number of vines per tree. Areas such as vine forest are characterized by dense vine tangles and few trees (Balee and Campbell 1990). Such areas tended to have a low number of total vines, because fewer trees were sampled; however, the number of vines per tree was high. These variables together comprised the habitat matrix. Because many of these variables were highly intercorrelated, factor analysis was used to produce composite variables that summarized as much variation as possible using the fewest variables possible (Morris 1987; Reyment and Joreskog 1993; Seamon and Adler 1996). For this purpose, we followed the factoring protocol of Cureton and D’Agostino (1983). An initial factoring was used to determine communalities, which were then inserted into the matrix as the diagonals. A final factoring was then performed, followed by a varimax rotation of the axes. The number of factors to be retained for further analysis was determined by examining a scree plot (Cureton and D’Agostino 1983). Factors retained represented gradients in microhabitat structure (Morris 1987; Seamon and Adler 1996) and were interpreted by examining correlations between the factors and the original variables.

Insects and fruit.-Small mammals may be directly responding to variation in resource abundances, and hence may show significant relationships when their abundances are regressed against insect biomass or fruit abundance. Alternatively, small mammals may be responding to resources indirectly by occupying habitats in which certain resources are more abundant. As a result, measures of insect biomass and fruit abundance were used as both independent variables in species-specific regressions (see below) and were regressed themselves against the habitat variables and factors. In the latter analysis, insect biomass relationships with habitat were examined at both the trap-station and grid level using stepwise multiple regressions (P-value for entry into the model set at 0.1 and for retention set at 0.05). At the station level, 2 models were created, 1 containing the original habitat variables and a 2nd containing the retained factors from the factor analysis. At the grid level, 3 models were used: the original habitat variables, the retained factors, and a 3rd containing the coefficients of variation (CVs) of the habitat variables (see August 1983; Malcolm 1995). Because the number of fruiting events observed was low, fruit abundance was examined only at the grid level, and also using these 3 models.

Small mammal habitat use.-At the station level, we used multiple logistic regression to predict species presence or absence as a function of the various habitat and resource variables. For this analysis, the dependent variable was presence or absence of a species at each trap station. Four sets of models (predictor variables) were analyzed: the original habitat variables, the retained habitat factors, the original habitat variables plus insect biomass, and insect biomass alone. In each of these models, we employed stepwise selection (P to enter set to 0.1 and P for retention set at 0.05). The relative effectiveness of each of these models was evaluated using Akaike’s information criterion (AIC). To further evaluate insect biomass as a predictor variable, once these models were analyzed, additional models were created including the retained variables or factors plus insect biomass if it showed a significant (or nearly so) relationship to species presence or absence. This assured that full complements of possible models were evaluated. AIC again was calculated for each of these models.

At the grid level, we used both multiple regression and redundancy analysis to examine patterns of habitat use and relationships to resources. In the multiple regressions the dependent variable was relative abundance of a speci\es on each grid, and the independent variables were various combinations of the habitat variables, habitat factors, and resource variables. Seven multiple regression models were created for each species: all habitat variables, habitat factors, CVs for habitat variables, CVs of factors, original habitat variables and insect biomass, insect biomass, and fruit abundance. Each of these models used stepwise selection (P to enter set to 0.1 and P to remain set at 0.05). Subsequently, models were compared using AIC.

In the above analyses, a large number of regression models were created. Much has been published on the need for adjustments to Pvalues when multiple statistical tests are performed (e.g., Moran 2003; Rice 1989; Roback and Askins 2005). Without a correction such as a sequential Bonferroni test (Rice 1989), the likelihood of type I error increases. We did not perform such a correction in our analyses; hence, some of our models may be significant due to chance alone. However, because our primary purpose here is hypothesis generation, not hypothesis testing, we judged the risk of type I errors to be acceptable (see Moran 2003).

Because many neotropical small mammals have been shown to increase in density after disturbances (Adler 1996a; Adler and Levins 1994; Lambert and Adler 2000; Lambert et al. 2003, 2005b; Malcolm 1997; Ochoa 2000; Terborgh et al. 2001) and thus may be responding to the same underlying ecological gradient, redundancy analysis was used to examine overall patterns of species responses. These analyses were conducted only at the grid level, because the large number of zeros in the station-level matrix can bias results, producing extraordinary long gradient lengths. For this purpose, the species matrix consisted of the abundance of each species on a grid. Three habitat matrices were used: mean values for each of the habitat variables listed above plus insect biomass and fruit abundance, CV of each of the habitat variables, and the habitat factors. Forward selection with a Monte Carlo permutation test (9,999 iterations) was used to select 5 variables that best described variation in the species matrix. Species-habitat relations were evaluated by examining species-environment correlations and by visually examining biplots (Jongman et al. 1995).

RESULTS

We captured 684 individuals of 7 marsupial and 15 rodent species in 5,700 station-nights (22,800 trap-nights). Overall trap success was 12% (individuals per station night). Five species of marsupials and 9 species of rodents had 10 or more captures and were included in the analyses of microhabitat use (Table 1). Additional information on the species captured at the site can be found in Lambert et al. (2005a).

Factor analysis produced 3 composite habitat variables (factors 1- 3). Based on correlations of the factors with the original variables (Table 2), these were interpreted as representing 1) increasing tree density, 2) increasing amounts of downed logs, and 3) increasing forest stature.

Predicting resource abundances.-Both station- and grid-level models indicated that insect biomass was highest in disturbed or edge-affected areas. The most-parsimonious model at the station level for insect biomass showed increasing biomass with decreasing tree density (factor 1, F – 10.29, d.f. = 1, 189, P = 0.0016) and decreasing amounts of downed logs (factor 2, F = 7.31, d.f. = 1, 189, P = 0.0075); the model based on the original habitat variables (difference in AIC [ΔAIC] = 4.2 relative to the factor-based model) showed increasing insect biomass with decreasing tree density (F = 11.35, d.f. = 1, 189, P = 0.0009). Similarly, grid-level analyses showed that insect biomass was highest where trees were at relatively low densities (F = 27.85, d.f. = 1, 17, P < 0.0001) and were relatively small (F = 19.93, d.f. = 1, 17, P = 0.0028). Insect biomass also increased with the CV for understory openness (F = 12.20, d.f. = 1, 17, P = 0.0028), although this model did not perform as well as the model that used the raw habitat variables (ΔAIC = 9.5). None of the habitat factors were related significantly to insect biomass at the grid level. Number of fruiting trees on a grid increased with mean log size (F = 8.75, d.f. = 1, 17, P = 0.0088) and decreased with increasing tree density (factor 1, F = 4.98, d.f. = 1, 17, P = 0.0394), indicating that fruit production may be highest near gaps.

Small mammal habitat use, station level.-Based on AIC values, models containing the original habitat variables produced the best models for 6 of the 14 species, models containing factor scores produced the best models for 3 species, and the model with insect biomass plus factor 3 produced the best model for Emmons’ oryzomys (Oryzomys emmonsae; Table 3). None of the habitat variables, habitat factors, or insect biomass models were significant predictors of the southern opossum (Didelphis marsupialis), the long-furred woolly mouse opossum (Micoureus demerarae), the gray four-eyed opossum (Philander opossum), or the climbing mouse (Rhipidomys cf. mastacalis). Mean tree size was retained more than any of the other original habitat variables (3 retentions); followed by understory openness, ratio of woody understory stems to herbaceous stems, number of vines per tree (each with 2 retentions); and litter depth, number of woody stems, tree density, mean log size, number of logs, and volume of downed wood (each with 1 retention). Number of herbaceous stems, tree basal area, total number of vines, and number of saplings appeared unimportant to small mammals because they were not retained in any of the station-level models (Table 3). The habitat associations of many of the individual small mammal species were indicative of them occupying disturbed or edge-affected habitats (Table 3).

Small mammal habitat use, grid level.-Models containing the original habitat variables produced the best models for 6 of the 14 species, models containing CVs produced the best model for 5 species, and number of fruiting trees alone was the best model for Oecomys catherinae (Table 4). No significant models were produced for M. demerarae or R. cf. mastacalis. The most important habitat variables (based on number of retentions) were mean tree size (4 retentions); followed by tree density and volume of downed wood (each with 3 retentions); ratio of woody understory stems to herbaceous stems (2 retentions); and woody stems, number of vines per tree, basal area, and mean log size (1 retention each). The variables litter depth, understory openness, and number of herbaceous stems were not retained in any model. Insect biomass showed a significant relationship with 4 species and number of fruiting trees with 2 species. Similar to what was seen with the station-level analyses, many species were most abundant in disturbed habitats (Table 4).

Many of the variables retained by the forward selection for the redundancy analysis such as tree size, tree density, and herbaceous stems also were often retained in the station- and grid-level models, indicating their overall importance to small mammals. The forward selection for the redundancy analysis on the habitat and resource variables retained the following 5 variables: tree density (P = 0.001), mean tree size (P = 0.001), insect biomass (P = 0.098), number of understory woody stems (P = 0.142), and mean litter depth (P = 0.155). The first 2 axes from the redundancy analysis explained 25.4% and 21.7% of the variation in the data set, respectively. Species-environment correlations were 0.877 for axis 1 and 0.866 for axis 2. Total variance of the species matrix explained by these 5 variables was 54%. Forward selection on the habitat variable CVs retained herbaceous vegetation CV (P = 0.006), tree density CV (P = 0.078), mean tree size CV (P = 0.106), mean log size CV (P = 0.176), and litter depth CV (P = 0.150). Axis 1 explained 23.9% and axis 2 explained 12.7% of the variation; species-environment correlations were 0.915 and 0.631, respectively. Variance explained by these 5 variables was 47%. Each of the 3 factors was included in the redundancy analysis; P-values were 0.026, 0.310, and 0.012 for factors 1-3, respectively. Axis 1 and axis 2 explained 23.5% and 7.5% of the variation, respectively, and the species-environment correlations were 0.915 for axis 1 and 0.528 for axis 2. Together the 3 factors explained 32% of total variance.

Patterns of habitat use shown by the redundancy analyses were similar to those reveled by the regressions, with many of the species patterns running in the same directions and showing positive relationships with insect biomass, tree density CV, and herbaceous vegetation CV and negative relationship with mean litter depth, number of understory woody stems, tree density, and each of the factors (Figs. 1-3). D. marsupialis tended to show the opposite relationship. The vector for the Azara’s broad-headed oryzomys (Oryzomys megacephalus) was roughly perpendicular to the axis for most of the other species, indicating positive relationships with mean tree size (Fig. 1), mean log size CV, mean tree size CV (Fig. 2), and factors 1 and 3 (Fig. 3). Vectors for O. emmonsae, the bare- tailed woolly opossum (Caluromys philander), and M. demerarae also were perpendicular to the axes for the majority of species for both the habitat and the factor redundancy analysis (Figs. 1 and 3); M. demerarae was more separated from these other 2 species in the habitat CV redundancy analysis (Fig. 2).

DISCUSSION

The majority of the species at this Xing Basin site showed similar patterns of habitat use in that they were most abundant in disturbed, 2nd-growth or edge habitats. These areas tended to have dense understories, few small trees, thick vine tangles, and abundant downed wood. Such habitat relations are typical of species occupying secondary habitats or disturbed areas (e.g., Malcolm 1997), which is consistent with other studies o\f neotropical small mammal habitat use (Adler 1996a; Adler and Levins 1994; Lambert and Adler 2000; Lambert et al. 2003, 2005a, 2005b; Malcolm 1995, 1997; Terborgh et al. 2001). The conclusion reached by Malcolm (1997) that most neotropical small mammals respond positively to secondary habitats thus received support in this study.

The increase in abundance of many of these small mammal species in disturbed habitats may be of conservation concern should the rate of human activity increase in the area. Although recent work showed the effects of logging in the area to be minimal and site specific, it was also shown that species such as Akodon sp. nov. and the 3 Proechimys species are likely important seed predators in the area (Lambert et al. 2005b). All 4 of these species were shown in this study to be more abundant in edge-affected or disturbed habitats. Should levels of logging damage or other disturbance increase, higher abundances of these species could lead to the regeneration suppression observed after logging in tropical Africa (Struhsaker 1997).

Two species, M. demerarae and R. cf. mastacalis, showed no relationships with any of the variables in this study, and thus appear to be habitat generalists. These species may be true habitat generalists in that they occupied all habitats with nearly equal frequency with no preference for any habitat features, or they may frequently occupy certain habitats based on features not quantified in this study. It has been suggested that differential use of vertical strata might be important in structuring neotropical small mammal communities (August 1983; Charles-Dominique et al. 1981; Cunha and Vieira 2002; Malcolm 1995). Arboreal species such M. demerarae and R. cf. mastacalis may be responding to canopy habitat features that are poorly represented among the primarily terrestrially biased set of measurements used here. For instance, Malcolm (1995) suggested that the semivariance (grain) of canopy structure was important for canopy-dwelling species. Pires et al. (2002) found M. demerarae became more restricted to the interior of forest fragments after fire disturbance to the forest edge, and Pardini (2004) found M. demerarae to be more common in mature forest areas. Similarly, Grelle (2003) found Rhipidomys mastacalis to be most abundant at trap heights of 5-12 m, and at the study site R. cf. mastacalis was more often captured in the canopy than on the ground (Lambert et al. 2005a). These results indicate the importance of vertical strata to these species.

Aside from those species that showed few or no habitat relationships, O. megacephalus and O. emmonsae were the only species to show patterns of habitat use that were not consistent with the preferential use of edge or gap habitats. These species’ positive relationship with variables such as number of woody stems and mean tree size indicated that they tended to inhabit more-mature, less- disturbed areas of the forest with an open understory, and large trees. In the redundancy analyses, these species’ vectors often ran perpendicular to those of many of the other species that seem to prefer disturbed habitats, indicating that the oryzomys may be responding to a different gradient than the rest of the community. These results are supported by the results of other studies on Oryzomys species. MaIygin and Rosmiarek (2005) listed Oryzomys macconnelli (a species morphologically similar to O. emmonsae) as being a forest-interior species, as did Malcolm (1991), and O. megacephalus as being a transitional forest species. Pardini (2004) found Oryzomys laticeps to be significantly more abundant within intact forest fragments than in the surrounding disturbed forest matrix. The forests around the Pinkait Research Station lie in the transition from lowland Amazonian forest to Cerrado habitat, and are characterized by a low canopy and frequent gaps. The 2 Oryzomys species found at the site appear, with respect to the other species, to be the most dependent on areas of mature forest; however “mature forest” in the vicinity of Pinkait may show many structural similarities to more disturbed areas at other forested sites. For instance, Peters (2004) noted that Pinkait has a more even vertical profile than forests near Manuas. Other species at the site such as Caluromys lanatus, Mesomys stimulax, Oecomys paricola, and Oecomys bicolor may be even more restricted to areas of mature, tall- stature forest, which may explain their rarity at the site. These species therefore may be of special conservation concern in the area of Pinkait should the rate of disturbance increase from logging or should other human activity increase. Unfortunately, we did not record enough captures to include these species in our analyses.

Other species that were rarely captured in the study may similarly be more restricted in their use of habitat. The degree of specialization in use of habitat can affect the abundance of a species within its range (Adler and Wilson 1987; MacNally 1995; Seamon and Adler 1996; Tomblin and Adler 1998). Two species of semiaquatic rodents, the web-footed marsh rat (Holochilus brasiliensis) and the South American water rat (Nectomys squamipes), were represented at the site by 1 capture each (also see Lambert et al. 2005a). Other species found at the site such as the red-nosed armored tree rat (Makalata didelphoides) and Oligoryzomys species also may be restricted in their use of habitat and thus rare at the site, or may in fact be abundant at the site and simply trap shy (Lambert et al. 2005a).

Understanding the relationships among small mammal abundances, habitat structure, and resource abundance will provide insights into the ecological processes that control mammal communities which will ultimately aid in their conservation. Measures of resource abundance (insect biomass and number of fruiting trees), like many of the small mammal species themselves, were related to habitat variables indicative of disturbed areas, edge-affected habitats, or both. This is consistent with previous studies (e.g., Ganzhorn 1995; Malcolm 1995) and supports the possibility that small mammals may be more abundant in these habitats because of increased resource abundance. Indeed, 6 species showed significant positive relationships with insect biomass and 3 species showed positive relationships with number of fruiting trees. Only for O. emmonsae was the relationship with insect biomass negative, and including insect biomass with factor 2 produced the most-parsimonious model at the station level for this species. This result is surprising in that oryzomys are often at least partially insectivorous (Charles-Dominque et al. 1981; Emmons and Peer 1997; Guillotin 1982). It is possible that O. emmonsae is responding to some habitat feature not sampled in this study of which insect biomass is a better correlate than any of the measured variables. It also is possible that O. emmonsae is less insectivorous than other better-studied members of the genus.

With the exception of O. emmonsae, only 6 times did the inclusion of insect biomass produce models that, based on AIC scores, could be considered approximations of the best model (ΔAIC < 2-Burnham and Anderson 2002). In 4 of these 6 cases, despite the fact that inclusion of insect biomass only marginally affected the AIC values, insect biomass itself was not significant. For only the short- tailed opossum (Monodelphis species), and 1 of the arboreal oecomys, O. catherinae, were models containing measures of resource abundance the most parsimonious. This would seem to indicate that species are not responding directly to resource levels, but instead are responding directly to habitat features, a result that also is supported by redundancy analysis results. Although insect biomass was retained by the forward selection as 1 of the 5 best variables and many of the species' vectors ran in the same direction as insect biomass, the P-value for insect biomass was not significant once the other variables were in the model. However, as Malcolm (1997) points out, different types of insect traps sample different parts of the insect community, thus the measure of insect biomass used in this study may not directly represent the amount of insect resources directly available to the small mammals.

Increases in occurrences of Akodon sp. nov. and Marmosa murina with increasing insect biomass, and of arboreal oecomys, O. catherinae and Oecomys roherti, with fruit abundances are expected because these genera tend to be insectivorous and frugivorous, respectively (Eisenberg and Redford 1999; Emmons and Peer 1997). However, the correlations with resource abundance for some of the other species are more surprising. Although spiny rats (Proechimys) are generally believed to be mostly frugivores and granivores (Adler 1995; Emmons and Peer 1997), Cuvier’s spiny rat (Proechimys cuvierii) appears to eat more insects than other members of the genus (Henry 1997). Thus, its association with high insect biomass may reflect a true relationship between the species and resource. However, examination of all previous data on members of the genus Monodelphis indicates that they are almost entirely insectivorous (Emmons and Peer 1997), not frugivorous. Yet, both this and previous work (Emmons and Feer 1997; Malcolm 1991) have shown a strong association between Monodelphis and logs and downed wood. The correlation with fruit abundance for Monodelphis species may be spurious, especially given that a relationship was found between average log size and fruit abundance. Large logs were more common in logged areas (Lambert et al. 2005b) and low levels of canopy disturbance, such as those often seen from selective logging, has been shown to increase fruit production (Ganzhorn 1995).

The 2 large opossum species (D. marsupialis and P. opossum) did not appear to be selecting habitats at scales that were adequately represent\ed in either the station- or grid-level analyses. We frequently recaptured P. opossum on multiple grids within the same 10-day trapping period, with some individuals moving nearly a half a kilometer in a few nights. We did not observe similar movements with recaptures of D. marsupialis; however, this species has been previously reported to be wide ranging (Emmons and Peer 1997; Malcolm 1991; Sunquist et al. 1987). Additionally, we had only 1 instance of an individual D. marsupialis being recaptured on the same grid in different trapping sessions, an event common for many other species. Models created using the CVs were the only significant models for either of these 2 species. These species likely range over large areas of the forest, moving through or nesting in certain habitats, while foraging in others.

Six other species (M. murina, Akodon sp. nov., O. catherinae, O. roberli, O. emmonsae, Proechimys goeldii, and Proechimys roberti) showed significant relationships with 1 or more of the CVs. For 3 of these species (O. roberti, P. goeldii, and P. roberti), these models were the most parsimonious, indicating the importance of habitat heterogeneity in structuring small mammal communities, a point raised previously (August 1983; Malcolm 1995). Similar to the large opossums (D. marsupialis and P. opossum), these species may forage in one habitat type while nesting in others (e.g., Malcolm 1997). In contrast to the large opossums, however, these species are presumably selecting habitat at small scales.

In summary, many of the species in this study in the Amazon Basin showed habitat relationships indicative of disturbed habitats, which is consistent with other studies of neotropical small mammals (Adler 1996a; Adlerand Levins 1994; Lambert and Adler 2000; Lambert et al. 2003, 2005b; Malcolm 1995, 1997; Terborgh et al. 2001). In some cases, we obtained evidence of increased resource abundance in these areas, although for many species measures of resource abundance were not important once the effects of habitat were taken into account. Although this study supports the hypothesis that increased resource abundance may drive the response of small mammals to disturbance (e.g., Malcolm 1995, 1997; Ochoa 2000), it indicates that the relationship of small mammals to resources may not be a direct one. It is possible that increased protection from predators and increased resource abundance interact to produce the increased small mammal abundances often seen after disturbance. In areas where animals are well protected from predators, they may more fully exploit available resources (see Brown and Kotler 2004); thus, direct measures of resource abundance may not reflect the amount of resources available to small mammals. Among the species at this site, only O. megacephalus and O. emmonsae failed to show such a pattern and were more common in more-mature, less-disturbed areas. Variance in many of the habitat variables was important for many species at the grid level, indicating the importance of habitat heterogeneity (sensu August 1983) in structuring small mammal communities. This study was the 1st to specifically examine habitat and resource use by small mammals in the Amazon Basin and our results are consistent with other microhabitat use studies from the Neotropics (Adler et al. 2000; Henriques and Alho 1991; Lacher and Alho 2001; Lambert and Adler 2000; Lambert et al. 2003; Tomblin and Adler 1998). However, more work is needed if we are to understand the response of small mammals to disturbances across a wide range of tropical forests and habitat types.

RESUMO

Estudos anteriores nas florestas tropicais sugerem que muitas das espcies de pequenos mamferos atingem elevadas densidades em habitats perturbados; entretanto, poucos locais foram estudados. Conseqiientemente, os habitats e os recursos utilizados por muitas espcies de pequenos mamferos so pouco compreendidos. Especialmente na bacia Amaznica, onde nenhum estudo sobre a associao de habitat com pequenos mamferos foi realizado. Ns estudamos as relaes de variveis do habitat com a abundncia de recursos com 5 espcies de marsupials e 9 espcies de roedores em uma rea no sudoeste da Amaznia. Pequenos mamferos foram amostrados com armadilhas colocadas tanto no cho quanto no subbosque. Oito variveis do habitat foram mensuradas para quantificar sua estrutura. A biomassa de insetos foi obtida atravs da coleta destes com armadilha de cola, e a abundancia de frutos foi quantificada por meio de censos. Padres no uso do habitat foi obtido por meio de regresses logsticas, regresses mltiplas e ordenaes. Observou-se para muitas das espcies o aumento de sua abundncia relacionada com caractersticas do habitat que indicam o efeito de borda, ou perturbaes no habitat, apresentando relaes negativas com a abertura do sub-bosque, densidade de caules lenhosos no subbosque, tamanho e densidade das rvores, e tambm, relaes positivas com o numero de lianas por rvore, o tamanho mdio, volume e quantidade de troncos cados. Ns obtivemos suporte para a hiptese em que a causa destes padres o aumento da abundancia de recursos nestas reas, pois tanto a biomassa de insetos quanto o numero de rvores frutificando mostraram relaes similares. Entretanto, para muitas espcies, a quantificao da abundancia de recursos no foi importante, quando incluiu-se as variveis do habitat no modelo, indicando uma indireta relao com os recursos.

ACKNOWLEDGMENTS

We thank the Kayapo village of A’ukre for allowing us access to their land and the Conselho Nacional de Desenvolvimento Cientifico e Technolgico, Fundao Nacional do Indio, and Instituto Brasileiro do Meio Ambiente e dos Recursos Naturals Renovavaeis for research permits. Kubenget Kayap, Ko-ket Kayap, A. Jerozolinski, P. Gelok, D. Pinto, S. Peters, N. V. de Sales, and J. Norghauer all provided valuable assistance in the field. We thank J. Solrzano-Filho for preparing the Portuguese summary and for his assistance in the field. Help with specimens was provided by M. DeVivo and the staff of the Museu de Zoologia, M. Engstrom and the staff of the Royal Ontario Museum, R. Voss, and J. Patton. Valuable comments on the manuscript were provided by S. C. Thomas, J. C. Ray, and T. E. Lcher. Financial support was provided by National Geographic Society, Wildlife Conservation Society, American Society of Mammalogists, Donner Foundation of Canada, and the Natural Sciences and Engineering Research Council of Canada. Financial support of TDL was provided by fellowships from the Organization of American States and an Ontario Graduate Scholarship.

LITERATURE CITED

ADLER, G. H. 1995. Fruit and seed exploitation by Central American spiny rats, Proechimys semispinosus. Studies on Neotropical Fauna and Environment 30:237-244.

ADLER, G. H. 1996a. The island syndrome in isolated populations of a tropical forest rodent. Oecologia 108:694-700.

ADLER, G. H. 1996b. Habitat relations of two endemic species of highland forest rodents in Taiwan. Zoological Studies 35:105-110.

ADLER, G. H., M. T. BECERRA, F. PRADO, AND B. L. TRAVI. 2000. Ecology of spiny rats, Proechimys canicollis, in northern Colombia. Mammalia 64:145-153.

ADLER, G. H., AND D. W. KESTELL. 1998. Fates of neotropical tree seeds influenced by spiny rats (Proechimys semispinosus). Biotropica 30:677-681.

ADLER, G. H., AND R. LEVINS. 1994. The island syndrome in rodent populations. Quarterly Review of Biology 69:473-490.

ADLER, G. H., D. C. TOMBLIN, AND T. D. LAMBERT. 1998. Ecology of two species of echimyid rodents (Hoplomys gymnurus and Proechimys semispinosus) in central Panama. Journal of Tropical Ecology 14:711- 717.

ADLER, G. H., AND M. L. WILSON. 1987. Demography of a habitat generalist, the white-footed mouse, in a heterogeneous environment. Ecology 68:1785-1769.

ANIMAL CARE AND USE COMMITTEE. 1998. Guidelines for the capture, handling, and care of mammals as approved by the American Society of Mammalogists. Journal of Mammalogy 79:1416-1431.

ASQUITH, N. M., S. J. WRIGHT, AND M. J. CLAUSS. 1997. Does mammal community composition control recruitment in neotropical forests? Evidence from Panama. Ecology 78:941-946.

AUGUST, P. V. 1983. The role of habitat complexity and heterogeneity in structuring tropical mammal communities. Ecology 64: 1495-1507.

BAKKER, V. J., AND D. A. KELT. 2000. Scale-dependent patterns in body size distributions of neotropical mammals. Ecology 81: 3530- 3547.

BALEE, W., AND D. G. CAMPBELL. 1990. Evidence for the successional status of liana forest (Xingu River basin, Amazonian Brazil). Biotropica 22:36-47.

BROWN, J. S., AND B. P. KOTLER. 2004. Hazardous duty pay and the foraging cost of predation. Ecology Letters 7:999-1014.

BURNHAM, K. P., AND D. R. ANDERSON. 2002. Model selection and inference: a practical information-theoretic approach. SpringerVerlag, New York.

CHARLES-DOMINIQUE, P., ET AL. 1981. Les mammifores frugivores arboricolesnocturnes d’une foret guyanaise: Interrelations plantesanimaux. Revue d’Ecologie la Terre et Vie 35:342-435.

CUNHA, A. A., AND M. V. VIEIRA. 2002. Support diameter, incline, and vertical movements of four didelphid marsupials in the Atlantic Forest of Brazil. Journal of Zoology (London) 258:419-426.

CURETON, E. E. AND R. B. D’AGOSTINO. 1983. Factor analysis: an applied approach. Lawrence Elbaum, Hillside, New Jersey.

DICKMAN, C. R. 1988. Body size, prey size, and community structure in insectivorous mammals. Ecology 69:569-580.

EHRLICH, P. R., AND E. O. WILSON. 1991. Biodiversity studies: science and policy. Science 253:758-762.

EISENBERG, J. H., AND K. H. REDFORD. 1999. Mammals of the Neotropics, the central Neotropics: Ecuador, Peru, Bolivia, Brazil. Vol. 3. University of Chicago Press, Chicago, Illinois.

EMMONS, L. H., AND F. PEER. 1997. Neotropical rainforest mammals, a field guide. 2nd ed. University of Chicago Press, Chicago, Illinois.

FORGET, P. M. 1991. Comparative recr\uitment patterns of 2 non- pioneer canopy tree species in French-Guiana. Oecologia 85:434-439.

GANZHORN, J. U. 1995. Low-level forest disturbance effects on primary production, leaf chemistry, and lemur populations. Ecology 76:2084-2096.

GRELLE, C. E. V. 2003. Forest structure and vertical stratification of small mammals in a secondary Atlantic Forest, southeastern Brazil. Studies on Neotropical Fauna and Environment 38:81-85.

GUILLOTIN, M. 1982. Proechimys cuvieri (Rodentia, Echimyidae) in the terrestrial micro-mammals community of French Guyana forest. Mammalia 46:299-318.

HENRIQUES, R. P. B., AND C. J. R. ALHO. 1991. Microhabitat selection by 2 rodent species in the Cerrado of central Brazil. Mammalia 55: 49-56.

HENRY, O. 1997. The influence of sex and reproductive state on diet preference in four terrestrial mammals of the French Guianan rain forest. Canadian Journal of Zoology-Revue Canadienne de Zoologie 75:929-935.

HOCH, G. A., AND G. H. ADLER. 1997. Removal of black palm (Astrocaryum standleyanum) seeds by spiny rats (Proechimys semispinosus). Journal of Tropical Ecology 13:51-58.

JONGMAN, R. H. G., C. J. F. TER BRAAK, AND O. F. R. VAN TONGEREN. 1995. Data analysis in community and landscape ecology. Cambridge University Press, Cambridge, United Kingdom.

LACHER, T. E., AND C. J. R. ALHO. 2001. Terrestrial small mammal richness and habitat associations in an Amazon forest-Cerrado contact zone. Biotropica 33:171-181.

LAMBERT, T. D., AND G. H. ADLER. 2000. Microhabitat use by a tropical forest rodent, Proechinws semispinosus, in central Panama. Journal of Mammalogy 81:70-76.

LAMBERT, T. D., G. H. ADLER, C. M. RIVEROS, L. LOPEZ, R. ASCANIO, AND J. TERBORGH. 2003. Rodents on tropical land-bridge islands. Journal of Zoology (London) 260:179-187.

LAMBERT, T. D., J. R. MALCOLM, AND B. L. ZIMMERMAN. 2005a. Variation in small mammal species richness by trap height and trap type in southeastern Amazonia. Journal of Mammalogy 86: 982-990.

LAMBERT, T. D., J. R. MALCOLM, AND B. L. ZIMMERMAN. 2005b. Effects of mahogany (Swietenia macrophylla) logging on small mammal communities, habitat structure, and seed predation in the southeastern Amazon Basin. Forest Ecology and Management 206:381- 398.

LEIGH, E. G., S. J. WRIGHT, E. A. HERRE, AND F. E. PUTZ. 1993. The decline of tree diversity on newly isolated tropical islands-a test of a null hypothesis and some implications. Evolutionary Ecology 7:76-102.

LOVEJOY, T. E., ET AL. 1986. Edge and other effects of isolation on Amazon forest fragments. Pp. 257-285 in Conservation biology: the science of scarcity and diversity (M. E. Soule, ed.). Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts.

MACNALLY, R. C. 1995. Ecological versatility and community ecology. Cambridge University Press, Cambridge, United Kingdom.

MALCOLM, J. 1991. The small mammals of Amazonian forest fragments: patterns and process. Ph.D. dissertation, University of Florida, Gainesville.

MALCOLM, J. R. 1994. Edge effects in central Amazonian forest fragments. Ecology 75:2438-2445.

MALCOLM, J. R. 1995. Forest structure and the abundance and diversity of neotropical small mammals. Pp. 179-197 in Forest canopies (M. D. Lowman and N. M. Nadkarmi, eds.). Academic Press, New York.

MALCOLM, J. R. 1997. Biomass and diversity of small mammals in Amazonian forest fragments. Pp. 207-221 in Tropical forest remnants: ecology, management and conservation of fragmented communities. (W. F. Laurance and R. O. Bierregaard, eds.). University of Chicago Press, Chicago, Illinois.

MALCOLM, J. R. 2004. Ecology and conservation of canopy mammals. Pp. 297-331 in Forest canopies (M. D. Lowman, N. M. Nadkarni, and H. B. Rinker, eds.). 2nd ed. Academic Press, New York.

MALCOLM, J. R., AND J. C. RAY. 2000. Influence of timber extraction routes on central African small-mammal communities, forest structure, and tree diversity. Conservation Biology 14:1623- 1638.

MALYGIN, V. M., AND M. ROSMIAREK. 2005. Notes on the ecology of Cricetinae from Peruan Amazonia and southern Bolivia. Zoologicheskii Zhurnal 84:492-506.

MARES, M. A., AND K. A. ERNEST. 1995. Population and community ecology of small mammals in a gallery forest of central Brazil. Journal of Mammalogy 76:750-768.

MORAN, M. D. 2003. Arguments for rejecting sequential Bonferroni in ecological studies. Oikos 100:403-405.

MORRIS, D. W. 1987. Ecological scale and habitat use. Ecology 68:362-369.

OCHOA, J. 2000. Efectos de la extraccin de maderas sobre la diversidad de mamiferos pequeos en bosques tierras bajas de la Guayana Venezolana. Biotropica 32:146-164.

PARDINI, R. 2004. Effects of forest fragmentation on small mammals in an Atlantic Forest landscape. Biodiversity and Conservation 13: 2567-2586.

PATTON, J. L., M. N. F. DA SILVA, AND J. R. MALCOLM. 2000. Mammals of the Rio Jurua and the evolutionary and ecological diversification of Amazonia. Bulletin of the American Museum of Natural History 244:1-306.

PERES, C. A., AND C. BAIDER. 1997. Seed dispersal, spatial distribution and population structure of Brazilnut trees (Bertholletia excelsa) in southeastern Amazonia. Journal of Tropical Ecology 13:595-616.

PETERS, S. L. 2004. Effects of selective logging on bat communities and trophic structure in southeastern Amazonia, Brazil. M.S. thesis, University of Toronto, Toronto, Ontario, Canada.

PINTO, D. 2003. Natural regeneration of bigleaf mahogany (Swietenia macrophylla King) in unlogged and selectively logged forests in southern Para, Brazil and Kayap forest tree knowledge and taxonomy. M.S. thesis, University of Toronto, Toronto, Ontario, Canada.

PIRES, A. S., P. K. LIRA, F. A. S. FERNANDEZ, G. M. SCHITTINI, AND L. C. OLIVEIRA. 2002. Frequency of movements of small mammals among Atlantic coastal forest fragments in Brazil. Biological Conservation 108:229-237.

REYMENT, R., AND K. G. JORESKOG. 1993. Applied factor analysis in the natural sciences. Cambridge University Press, Cambridge, United Kingdom.

RICE, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

ROBACK, P. J., AND R. A. ASKINS. 2005. Judicious use of multiple hypothesis tests. Conservation Biology 19:261-267.

ROGERS, L. E., W. T. HINDS, AND R. L. BUSCHBOM. 1976. General weight vs. length relationship for insects. Annals of the Entomological Society of America 69:387-389.

SEAMON, J. O., AND G. H. ADLER. 1996. Population performance of generalist and specialist rodents along habitat gradients. Canadian Journal of Zoology-Revue Canadienne de Zoologie 74:1130-1139.

STRUHSAKER, T. T. 1997. Ecology of an African rain forest logging in Kabale and the conflict between conservation and exploitation. University of Florida Press, Gainesville.

SUNQUIST, M. E., S. N. AUSTAD, AND F. SUNQUIST. 1987. Movement patterns and home range in the common opossum (Didelphis marsupialis). Journal of Mammalogy 68:173-176.

TERBORGH, J., ET AL. 2001. Ecological meltdown in predator-free forest fragments. Science 294:1923-1926.

TOMBLIN, D. C., AND G. H. ADLER. 1998. Differences in habitat use between two morphologically similar tropical forest rodents. Journal of Mammalogy 79:953-961.

TREJO, A., AND N. GUTHMANN. 2003. Owl selection on size and sex classes of rodents: activity and microhabitat use of prey. Journal of Mammalogy 84:652-658.

VOSS, R. S., AND L. H. EMMONS. 1996. Mammalian diversity in neotropical lowland rainforests: a preliminary assessment. Bulletin of the American Museum of Natural History 230:1-115.

WILSON, E. O. 1988. The current state of biological diversity. Pp. 3-18 in Biodiversity (E. O. Wilson, ed.). National Academy Press, Washington, D.C.

ZIMMERMAN, B., C. A. PERES, J. R. MALCOLM, AND T. TURNER. 2001. Conservation and development alliances with the Kayapo of southeastern Amazonia, a tropical forest indigenous people. Environmental Conservation 28:10-22.

Submitted 31 July 2005. Accepted 20 December 2005.

Associate Editor was Gerardo Ceballos.

THOMAS D. LAMBERT,* JAY R. MALCOLM, AND BARBARA L. ZIMMERMAN

Faculty of Forestry, University of Toronto, Earth Sciences Centre, 33 Willcocks Street, Toronto, Ontario M5S 3B3, Canada (TDL, JRM)

Director Kayap Project, Brazil Program, Conservation International, 1919 M Street NW, Washington, DC 20036, USA (BLZ)

Present address of TDL: Department of Biological Sciences, University of Nevada-Las Vegas, Las Vegas, NV 89154, USA

* Correspondent: thomas.lamben@uturonto.ca

2006 American Society of Mammalogists

www.mammalogy.org

Copyright Alliance Communications Group, A Division of Allen Press, Inc. Aug 2006

(c) 2006 Journal of Mammalogy. Provided by ProQuest Information and Learning. All rights Reserved.