Nordic Society Oikos

Non Flying and Landscape Changes in the Tropical Rain Forest Region of Los Tuxtlas, Mexico Author(s): Alejandro Estrada, Rosamond Coates-Estrada, Dennis Meritt, Jr. Source: Ecography, Vol. 17, No. 3 (Sep., 1994), pp. 229-241 Published by: Blackwell Publishing on behalf of Nordic Society Oikos Stable URL: http://www.jstor.org/stable/3683048 Accessed: 16/06/2009 15:19

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=black.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Nordic Society Oikos and Blackwell Publishing are collaborating with JSTOR to digitize, preserve and extend access to Ecography.

http://www.jstor.org ECOGRAPHY17: 229-241. Copenhagen 1994

Non flying mammals and landscape changes in the tropical rain forest region of Los Tuxtlas, Mexico

Alejandro Estrada, Rosamond Coates-Estrada and Dennis Meritt, Jr.

EstradaA., and Coates-Estrada,R. and Merrit,Jr., D. 1994. Non flying mammalsand landscapechanges in the tropicalrain forest region of Los Tuxtlas,Mexico. - Eco- graphy17: 229-241.

The rapidand extensiveconversion of Neotropicalrain forests to a landscapeconsist- ing of pasturelands and otheragricultural habitats has meantthe local disappearance, fragmentationand isolation of vertebratepopulations. To avoid furtherlosses in biodiversityin Neotropicalrain forests, it is imperativeto determinehow the different that constitutethe differentvertebrate communities have respondedto the anthropogenicalterations of theirnatural habitats. To providedata in this direction,we samplednon flying mammalsin 35 forestfragments, in six disturbedforest sites, in 12 regeneratingforests, in 20 arborealagricultural habitats (cacao, coffee, mixed, citrus and allspice),in fourlive fence sites andin fourpasture sites at Los Tuxtlas,Veracruz, Mexico. Samplingindicated that area of forestfragments and isolatingdistance were importantvariables influencing species richness.Disturbed forest had fewer species and individualspresent and supporteda lower non flying mammalbiomass than undisturbedand regeneratingforests. Shadedplantations (cacao, coffee and mixed) were richerin species and in individualsthan unshaded ones (citrus,allspice). Species richnessin forestsand agricultural habitats were associated to horizontalplant diversity and verticalfoliage diversity.Pasture habitats were the poorestman-made habitats in non flying mammalspecies. Live fences supporteda higherbiomass and species of non flying mammalsthan pastures and were similar to forestand shaded man-made habitats in species assemblages.

A. Estrada and R. Coates-Estrada, Estaci6n de Biologia "Los Tuxtlas", Inst de Biologia - UNAM, Apdo postal 176, San Andres Tuxtla, Veracruz,Mexico. - D. Meritt, Jr., Lincoln Park Zoo, 2200 North Cannon Dr., Chicago, IL, USA.

Introduction ica (see Lovejoy et al. 1984, 1986) and more recently for Mexico (Estrada et al. 1993a). The rapid and extensive conversion of Neotropical rain Such information is fundamental to determine what forests to a landscape consisting of pasture lands and remains of the original biological diversity, as repre- other agricultural habitats has meant the local disappear- sented by particular taxa, and to assess the adap- ance, fragmentation and isolation of vertebrate popula- tive responses of to tropical rain forest frag- tions. To avoid further losses in mammalian biodiversity mentation and replacement by agricultural vegetation in Neotropical rain forests, it is imperative to determine (Saunders et al. 1991, Terborgh 1992). These data may be which species have survived the fragmentation useful in designing conservation scenarios for wildlife and isolation of their tropical rain forest habitat and how involving different sets of species under different types of they are responding to the environmental change. Reports land management systems. As a necessary corollary to exist only for a few localities in Central and South Amer- this, we also need basic information on the little known

Accepted8 March1994 Copyright? ECOGRAPHY1994 ISSN 0906-7590 Printedin Denmark- all rightsreserved

ECOGRAPHY 17:3 (1994) 229 fences to divide the pasture land into smaller plots, to delimit the boundaries of their properties, and, in some cases, to rotate their cattle from one pasture plot to anoth- er. Because these posts grow rapidly in height and trunk width and produce moderately foliated crowns, some single rows of these live fences resemble corridors of vegetation criss-crossing the pasture lands. This paper presents information on how the non flying members of the mammalian community are responding to (a) biogeographical aspects of forest fragmentation (e.g., size and isolating age and distance of forest patches), (b) the continued deteriorationof isolated forest patches, and (c) the conversion of forest to pastures and to other agricultural habitats at Los Tuxtlas in southern Veracruz, Mexico. To achieve this, we examined field data on relative abundance and detection rates of mammal species present in isolated rain forest fragments, in disturbed and regen- erating forest habitats, in man-made vegetation islands representing five types of agricultural vegetation, in live fences and in The data are a Fig. 1. Northeasternsection of the region of Los Tuxtlasin pastures. presented comple- southern,Veracruz, Mexico. Lowland rain forest remaining ment to information published on the flying mammals of shown as dispersedfragments. The area was dividedinto four this animal community and on the frugivore guilds of the zones following longitude and latitude (95?10'W,18?30'N). avian and mammalian communities (Estrada et al. sites (forests, habitatsand live fences)located Study agricultural 1993a, b). in zones II and IV. Map drawnfrom satelliteimages takenin April 1990. 1 = San Martin Volcano (1600 m), 2 = Lake Sonte- comapan,3 = Gulf of Mexico, 4 = Lake Catemaco. Methods patterns of species composition and distribution for dif- We conducted field work in the northeastern area of the ferent animal communities in various localities in the of Los Tuxtlas (95?00'W, 18?25'N; max. 1600 m neotropics. region In the region of Los Tuxtlas, in southern Veracruz, Mexico, lowland rain forests reach their north- tropical Table1. Non flying mammalspecies and individuals censused at ernmost distribution in the American continent and are the habitatsinvestigated. Numbers between parentheses are spe- notable for their high biological diversity (Pennington cies and individualmean (+ SD) detectionrates per site in each and Sarukhan 1968, Estrada et al. 1985, Dirzo and Garcia habitat. 1992). However, land have con- management practices Habitat Sites Species N Mean + SD verted 80-90% of these forests to pastures for cattle Biomasskg ranching. The remaining forests in the landscape consist of collections of fragments of various sizes and with Forest* 35 38 1518 3298 different histories of isolation scattered throughout the (10.8?6.8) (44.7?35.3) (89.0+215.0) region (Estrada and Coates-Estrada 1988). Within this Cacao 4 16 144 277 (5.0?0.81) (36.0?19.2) (17.0+ 26.0) context, agricultural activities occur sporadically in time Coffee 4 13 106 308 and space in a "sea" of pastureland and occupy only (4.5?1.2) (26.5?14.4) (24.0+ 36.0) c. 3% of the land. These consist of cultivation of arboreal Mixed 4 13 145 427 plants such as citrus Citrus sinensis, Rutaceae, allspice (8.0?0.81) (36.2?19.0) (33.0+ 49.0) Citrus 4 8 38 49 Pimienta dioica, and cacao Theobroma ca- Myrtaceae, (2.0?0.81) (9.5?4.5) (6.0+5.0) cao, Sterculiaceae. Farmers cultivate coffee Coffea ara- Allspice 4 7 42 59 bica, Rubiaceae as a single crop. Less common is the (2.0?1.1) (10.5? 3.8) (8.0+8.0) cultivation of plots of mixed crops that include coffee, Livefence 4 11 83 123 (8.0?1.8) (20.2? 6.8) (12.0+12.0) cacao, citrus, and banana Musa spp., Musasea. Rain for- Pasture 4 4 93 57 est trees left by farmers provide shade in the case of (2.0+0.81) (23.2+ 3.8) (14.0+12.0) cacao, coffee and mixed plantations. Total 63 39 2169 It is a common farmers and ranchers of practice among * = = Los Tuxtlas as elsewhere in the lowland wet Forest size classes: < 2000 ha N 3, < 1000 ha N 3, < 800 ha Neotropics N = 3,600 ha N = 3, <400 ha N = 3, <200 ha N = 3, < 100 ha to use live posts of Bursera simaruba (Burseraceae) and N = 3, <50 ha N = 3, <20 ha N = 3, < 10 ha N = 3, <5 ha Gliricidia sepium (Leguminosae) to hold barbed wire N=3, 1 ha N=2.

230 ECOGRAPHY 17:3 (1994) a.s.l.) in southern Veracruz, Mexico (Fig. 1). Mean an- patches < 50 ha, 600 m for patches < 100 ha, 800 m for nual temperature is 27?C and mean annual precipitation patches <200 ha, 1000 m for patches <400 ha, 1500 m is 4964 mm (SD + 862 N = 20 yr) with a seasonal for patches <800 ha, 2000 m for patches < 1000 ha, distribution (Estrada et al. 1985). Non flying mammals 2500 m for patches < 2000 ha. were censused between 1990-1992 in 35 undisturbed Census routes were chosen to minimize trail overlap forest fragments (isolated forest sites without apparent within and between days and nights and thus the prob- further human intervention), in six disturbed forest frag- ability of viewing the same individual more than once. ments (strong edge effect, hunting and selective removal The census trails were at least 10 m from the edge of the of hardwoods, firewood and other), in six old second vegetation patch under investigation. We recorded the growths (regenerating forests 20-35 yr old) and in six number of individuals of each species detected. We con- young second growths (5-15 yr). In addition, non flying ducted the censuses between 0600 h and 1200 h and mammals were censused in 20 agricultural habitats, in between 1600 h and 1800 h and between 1900 h and four live fences and in four pastures. 2400 h. We used at night, in addition to flash lights, a We selected the study sites in two zones within the night vision scope (Javelin Electronics, model 221) to study area to have some control over spatial variation in minimize disturbance to detected animals in the nocturnal distribution of the organisms under investigation (Fig. 1). censuses. On most cases 2-3 people walked together with Sites ranged in elevation from sea level to c. 800 m. The either of the two senior authors. 35 undisturbed forest fragments studied varied in isolat- To sample terrestrial species with relatively small ing distance (distance to edge of the nearest forest frag- home ranges (most <200 g) and medium size ment: 0.2-8.0 km). Area of forest fragments, obtained by species that are cryptic in their behavior and thus difficult digitizing aerial photos and by corroboration in the field, to detect by visual means, we set up at each site 50 ranged from 1 to >2000 ha (Table 1). wire-mesh cage traps (size ranged from 48 x 17 x 17 cm At each site we estimated time (years) since isolation to 66 x 24 x 24 cm, Tomahawk Live Trap Co.; 25 baited (isolation = complete separation of forest from major with sardines and 25 baited with fruits and vegetables) forested land mass) by examining vegetation maps for the and 200 Sherman traps (23 x 8 x 8 cm) baited with oats, period before 1967, aerial photographs taken in 1967 and peanut butter and vanilla along 4 parallel lines; distance 1979, satellite images taken in 1986 and 1990 and by between lines was 20-25 m. field surveys. Isolating age ranged from 5 to 35 yr. Six We placed Sherman traps in pairs at 3 m intervals and undisturbed forests fragments selected at random from single Tomahawk traps at 5 m intervals. Trap lines were our data base were compared to the six disturbed forest at least 10 m from the site's borders. At the live fence habitats and to the six young and six second growth sites, we formed three trap lines separated from each habitats in species richness and composition. other by 50 m and running parallel to the barbed wire We sampled non flying mammals in four replicates of fence. We operated traps at all sites for two consecutive each of the following arboreal and perennial agricultural nights (dusk to 2300-2400 h) and for two consecutive habitats: cacao, coffee, mixed crops, citrus, and allspice. afternoons (1600 h to dusk). We recorded the species Tall (> 15 m) rain forest trees left by the farmers provided detected, its body weight and when possible its sex and shade in the case of cacao, coffee, and mixed plantations. age before releasing each mammal. The number of traps The agricultural plots ranged in size from 2-10 ha and was constant across all sites because we wanted to maxi- ranged in age from 12-15 yr; all habitats were fruit mize the number of animals detected by this means. productive. Straight line distance to the nearest forest Preliminary probes in 10 sites with number of traps kept fragment regardless of size ranged from 200 to 6000 m; constant showed that our trapping success was distance to the nearest plantation ranged from 200 to c. 10-20%. Varying the number of traps per area of forest 1000 m. sites showed that in the smaller sites trapping success was We also censused non flying mammals in four live nil, but that increasing the number of traps resulted in fence sites c. 2 km long each located across the pasture detections of animals. Detection success was found to land and in the vicinity of the forest and agricultural reach an asymptote with the use of 40-50 traps. habitats studied; each of these sites was at least 5 km To control for seasonal variations in the presence and away from the others. We surveyed the non flying mam- activity of mammals, surveys in the man-made habitats mals in four areas of pasture, 25 ha each, totally devoid of included eight samples per habitat under investigation, the original forest vegetation and at least 5 km apart from two in each quarterof the year (8 x 28 = 224 samples). In each other, but within a 5 km radius of the forest sites we forest sites, samples took place in each site in each bi- studied. In the pastures, we located the sampling area in monthly period of the year (4 x 53 sites = 212 samples). the center of each of the 25 ha pasture plots. While the use of trail censuses and traps are comple- We conducted visual census of non flying mammals mentary methods, they have certain biases. With the along existing trails or along trails demarcated by us with traps, we only sampled one level of the habitats studied a cotton thread. These trails were walked at a slow pace c. during the day and night and species of non flying mam- 2 km and both sides of trail were scanned visually. mals may be caught unequally by traps. However, this is Length of census trails at each site was 400 m for forest the only way to assess the presence of small terrestrial

ECOGRAPHY 17:3 (1994) 231 100 - LOGABUNDANCE (%) Fitch 1992). We present means (? SD) throughout the text. 10 -' **. We did not consider in our the mammal ama study aquatic Trichechus manatus because our 1 *mi - (Trichechidae) surveys were mainly terrestrial (but see Coates-Estrada and Es- 0.1 - ...... trada 1986). Likewise, we ignored in our comparison two introduced rodents, Mus musculus and Rattus rattus, as 0.01 , , I 0 5 10 15 20 25 30 35 40 well as the coyote, Canis latrans. We never encountered SPECIES RANKIN ABUNDANCE the introduced rodents, but they are common in human settlements. The coyote has recently expanded its range Fig. 2. Dominancediversity curve for the non flying mammal into Los Tuxtlas as a result of tropical rain forest con- communitydetected. Note the asymmetryof the curve and the version to We observed this mammal at the relativelyhigh dominanceof first five species. pasturelands. forest edges, but never in the forest interior. We refer, for the sake of economy, to non flying mammals as NFM in and/or cryptic mammals. With trail censuses, one runs the the rest of the paper. risk of counting very mobile and active mammals more than once. Bearing these considerations in mind and our efforts to minimize these biases, the reported data may represent a biased portrait of the mammal community Results present in the region. Taxonomic nomenclature for mam- General aspects mals followed Hall (1981). We obtained measurements of body weights from our own field records and from the In our sample of 35 undisturbedforest fragments and the literature (Emmons 1990). agricultural habitats including pastures, we censused To census the vegetation we recorded, at each site, in 2169 NFM of 39 species. Five species accounted for 66% six 10 x 10 m plots, all trees > 10 cm in circumference at of all records (Didelphis marsupialis, Peromyscus mex- breast level and > 1.5 m in height. We located the plots icanus, Alouatta palliata, Philander opossum, Sciurus randomly within the area where we sampled the mam- aureogaster), but 87% contributed to < 5% of the records mals; the plots were at least 30-40 m from each other. (Fig. 2 and Appendix 1). Sherman traps captured all We recorded for each tree censused, the species (taxo- murid and heteromid rodents recorded (7 species, 552 nomic nomenclature for plants follows Ibarraand Sinaca individuals) and Tomahawk traps captured only three 1987), maximum height and circumference. marsupials, D. marsupialis, Didelphis virginiana and We measured vertical foliage density at four randomly P. opossum. Traps recorded 84% of the individuals cen- selected spots within each of the six plots by scoring sused of these three species (N = 626). vegetation intercepts along a vertical pole at the follow- ing intervals: 0-0.5 m, 0.51-1.0 m, 1.1-2 m, 2.1-3 m, 3.1-5 m, 5.1-8 m, 8.1-10 m, 10.1-15 m, > 15 m. In this Non flying mammal species in undisturbed later case we expressed intercepts at each height interval forest fragments as the proportion of total intercepts recorded per site/ habitat. We expressed horizontal plant diversity per hab- In these habitats we censused 1518 NFM representing 38 itat as the mean number of tree species censused per species. Of these, species such as P. mexicanus, A. palli- site/habitat and vertical foliage diversity using the Shan- ata, D. marsupialis and P. opossum accounted for 50% non-Weaver information index (H') (Ludwig and Rey- of the records. The rest of the species contributed to 1% nolds 1988). or less of detections (Appendix 1). Because the cumu- Because mammals were censused using two different lative species curve in the forest sites tended to level off methods (traps and visual census), we compared data after the accumulation of > 2000 ha, we feel our sample across vegetation types using proportions and mean de- tection rates of species and individuals site in each per 40 - habitat (site/habitat from here on). We used rarefaction as c U 35 an alternative to richness indices to 30 species compare M pS richness habitats where sizes dif- E25 species among sample L fered (James and Rathbur 1981). We used the dissimi- A 20 TI15T 115 larity index (PD) and the polar ordination procedures of I E 10 V5 Bray-Curtis (Ludwig and Reynolds 1988) to measure the E of resemblance when forests and man- 0 I I I I I I I I I I I degree comparing 0 250 500 750 1000 1250 1500 1750 2000 2250 2500 made habitats in We used 2750 3000 species assemblages. Spearman AREA SAMPLED(HA) rank correlation, partial correlation analysis and "t" tests for when dependent samples comparing variables and Fig. 3. Cumulativenumber of non flying mammalspecies de- means between groups (Ludwig and Reynolds 1988, tected and cumulativearea sampledin forest fragments.

232 ECOGRAPHY 17:3 (1994) 25 -. . Table2. Species and individualdetection rates in four types of foresthabitat. Numbers between parentheses refer to numberof S 20 sites sampledper habitatand distanceto the edge of the nearest Pp undisturbedforest site. E 15 C * 2, Habitat Mean (? SD) Mean (? SD) I10 . numberof numberof E mammals S ? ? * species per per 5 _ m . site/habitat site/habitat 0 1 1 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 Undisturbed(N = 6) 14.1?3.3 37.6?6.1 ISOLATING DISTANCE (KM) Disturbed (N =6; 350 m+ 10.0 m) 6.7?3.4 15.6?3.7 Old growth (N = 6; 340 m + 11.1 m) 5.3?1.3 14.6?2.3 140 - m* Younggrowth (N = 6; 380 m+ 7.3 m) 4.8?1.7 13.0?3.9 N 120 Comparison species: Uf-Df t = 3.6, p = 0.01; UfYg t = 5.4, D100 * U * _ p = 0.003; Uf-Og t = 5.3, p = 0.003; Df-Yg t = 1.2, p = 0.26; V 80. t = = t = = 0.58. I Df-Og 1.08, p 0.32; Yg-Og 0.5, p * D 60 * Comparisonindividuals: Uf-Df t = 15.7, p = 0.001; Uf-Yg t = 20.5, p = 0.0001; Uf-Og t = 17.1, p = 0.0001; Df-Yg t = 4.4, 40 .0 A s A * s . p = 0.007; Df-Og t = 1.1, p = 0.31; Yg-Og t = 1.3, p = 0.24. * ? "* L 20 - a * * U "* S u - I I I I I I I I 00 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 ha; we detected the rest of the species in forest fragments ISOLATING DISTANCE (KM) >700 ha in size (Fig. 3). In this later case were large bodied species that areas and found in forest 4. between non mammal require large Fig. Relationship flying species rech- habitats such as Ateles ness and individuals censused and isolating distance of forest only Tayasu tajacu, geoffroyii, fragments. Eira barbara, Felis concolor and paca among others (Appendix 1).

provides a reasonably good representation of the species present in the forest remnants studied (Fig. 3). The mean number of species detected per site was 10.8 (SD ? 6.8) Species richness in undisturbed forests, and mean number of individuals censused per site was disturbed forests and in second growth 44.7 (SD ? 35.3) (Table 1). On average, forest sites had habitats 25% of the species in common. We detected a significantly higher mean number of NFM species and individuals censused per site in undisturbed than in disturbed forests (Table 2). Undisturbed forests and distance of undisturbed Area, age isolating were richer in species and individuals than young and old forest fragments and species richness second growth habitats. Young and old second growths did not differ in the mean number of detected Species richness and area of forest fragments were related species per site or in the mean number of individuals censused (rs = 0.85, p < 0.001), but species richness and number of per site Data also showed that of the 30 individuals censused per site were not found associated to (Table 2). species detected in undisturbed forests, 14 were isolating age of forest fragments (rs= -0.12, p = 0.25; rs = only (46%) pre- sent in undisturbed forests, 11 in old second -0.20, p = 0.12). However, a significant negative rela- (36%) tionship existed between site species richness and isolat- distance of forest = < and ing fragments (rs -0.77, p 0.02) Table 3. richness in ares of and between this latter variable and number Species equivalent pastures of individuals forests.Shown also is the speciesrichness detected in 100 ha of censused per site (r, = -0.50, p <0.02) (Fig. 4). agriculturalhabitats, in 1 ha forest patchand in 1.5 ha of live Partial correlation analysis showed that when the influ- fences. ences of area and distance were isolating partialled out, Habitat Numberof Numberof richness was best correlated with species isolating dis- species mammals tance (area partialled out r = -0.60, p <0.001, isolating = = distance partialled out r 0.37, p 0.01). An examina- Pastures100 ha (4 sites) 4 93 tion of new species accumulated in our sample and cumu- Forests 100 ha (4 sites) 16 175 lative area (ha), showed that c. 45% of the species were Forest 1 ha (1 site) 8 14 habitats100 ha 17 475 present in small forest < 200 ha, 60% were Agricultural (20 sites) patches pre- Live fence (8 km, 1.5 ha) 4 sites 10 81 sent in fragments < 250 ha, and 80% in fragments < 650

ECOGRAPHY 17:3 (1994) 233 -- FOREST (Table 1). These habitats had, on average, 59% of the 25 - E(Sn ) CACAO species in common (Appendix 1). 20 ---- COFFEE I Shaded man-made habitats (cacao, coffee and mixed --0--- MIXED 15 plantations) accumulated 17 species of NFM. Unshaded -,O O--- i ~~~~~~~~~~~~~~~~~~~~~-i>- --- CITRUS I- I I I - . habitats and accumulated 10 -- - ALLSPICE (citrus allspice) only eight spe- ~~ab<>- ---- LIVEFENCE i cies. These two habitats only had seven species in com- 5 mon. Arboreal mammals such as A. palliata, B sum- 0 ichrasti and P. flavus were a predominant 0 20 40 component 60 80 100 the NFM detected in shaded habitats N among (Appendix 1). We found significant differences between shaded and Fig. 5. Rarefactioncurves for all habitatsexamined. E(Sn) is the unshaded habitats in mean number of species detected per expectednumber of species in a randomsample of size n where site/habitat (shaded X = 5.8 range 3-9, unshaded X = 3.6 S is the sum of the probabilitiesthat each species will be 3-4; t = 2.34, = 0.05) and in the mean number of includedin the can be made of the ex- range p sample. Comparisons individuals censused site/habitat X = 19.7 pected numberof species at differentsample sizes. per (shaded range 2-72, unshaded X = 4.0 range 3-19; t = 4.1, p = 0.001). Mixed differed from the other shaded hab- growths and eight (26%) in young second growth hab- plantations itats. itats (cacao and coffee) in the mean (? SD) number of detected site X = 8.0 + 0.82 vs cacao The majority of the species missing from the disturbed species per (mixed X = 5.0 + 0.82; t = 3.6, = mixed vs coffee X = 4.5 forests and second growth habitats were arboreal mam- p 0.03; ? 1.2; t = 4.04; = but did not differ in the mean mals such as A. palliata, A. geoffroyii, T. mexicana, and p 0.02), number of individuals detected site X = 36.2 + C. mexicanus, carnivores such as F pardalis, F weidii, per (mixed 19.2 vs cacao X = 36.0 ? 19.2; t = 0.02, = 0.98; mixed and F concolor and other large mammals such as p vs coffee X = 26.5 + 14.4; t = 0.69, = M. americana and T. tajacu among others (Appendix 2). p 0.53) (Appendix The total NFM biomass detected at disturbed forests was 1). only c. 50% of the NFM biomass detected in undisturbed forest habitats, confirming the loss of medium and large size mammals. Regenerating forests accumulated < 15% Non flying mammals in live fences of the biomass present in undisturbed forests suggesting We censused 83 NFM of 11 in these habitats. the colonization of these forests by small size species species detected at these habitats 53% of the (Appendix 2). Species represented species detected at the man-made habitats only, exclud- ing pastures. Mean number of species detected per site was 8.0 (range 6-10) and mean number of individuals Non flying mammal species richness in censused per site was 20.2 (range 14-28) (Table 1, Ap- pastures pendix 1). Species such as S. aureogaster, D. marsupia- lis, P. mexicanus, P. opossum, D. virginiana and C. mex- Our surveys in four pasture sites totalling 100 ha resulted icanus accounted for 85% of the records in these habitats in the census of 93 NFM of four only species. One (Appendix 1). species (Sigmodon hispidus) accounted for 69% of the records. The other species were D. marsupialis, D. vir- giniana and P. opossum (Appendix 1). Mean detection used non rates of NFM species in pastures were among the lowest Vegetation types by flying mammal (X = 2.0 + 0.8 species per site; Table 1). species We recorded 49% of the species in our sample in both forests and in non forest habitats and 51% in forest habitats three (D. mar- Non flying mammals in agricultural habitats only (Appendix 1). Only species supialis, P. opossum and D. virginiana) were present in We censused at these habitats 649 NFM of 17 species. both forest habitats and in the seven man-made vegeta- Five species numerically dominated (>30 individuals tion habitats studied, 14% were present in 6-7 man-made each) the sample accounting for 60% of the records. habitats in addition to the forests and 30% were present in Among such species were habitat generalists such as forests and in 2-5 agricultural habitats (Appendix 1). D. marsupialis, P. opossum and Orthogeomys hispidus, We found no correlation between the number of vege- and forest dwellers such as Alouatta palliata and Sciurus tation types in which a species occurred and the species' aureogaster (Appendix). Mean number of NFM species mean body mass (rs = -0.19, p = 0.11), but an association detected per site/habitat ranged from 8.0 (mixed) to 3.5 existed between the number of habitats in which a species (cacao) and the mean number of individuals censused per was present at and the rank order of individuals censused site/habitat ranged from 9.5 (citrus) to 36.2 (mixed) per species (rs = 0.83, p< 0.01). This later variable was

234 ECOGRAPHY 17:3 (1994) 60 - per site in each habitat was best correlated with MNTSSH LIVEFENCE S FOREST CACAO PASTURE (FH' r = 0.65, p = 0.04; MNTSSH r = 0.88, p = 0.003). 9 I zm I I , ", The mixed, cacao and coffee plantations contributed to 0 20 40 a 60 80 100 78% of the total biomass recorded at the man made MIXED / 30 - COFFEE habitats; live fence habitats accounted for 10% and the * ALLSPICEE citrus and allspice plantations and pastures accounted for 13% (Table 1). Shaded man-made habitats (cacao, coffee, had a mean biomass site than unshaded 0 CITRUS mixed) higher per habitats (citrus, allspice and pastures) (shaded X = 24.6 + Fig. 6. Polar ordinationof PD index for the habitats studied. 9.1, unshaded X = 9.3 + 4.1; t = 6.5, p = 0.02). Note the similarityto foresthabitats in species aisssemblages of shaded man-madehabitats (cacao, coffee, mixed) and live fences andthe dissimilarityof pasturesand unshz aded man-made habitats. Discussion Our showed that a rich of forest NFM negatively correlated with the species mean body mass (rs study pool species still exists in the of Los Tuxtlas = -0.33, p = 0.02), but partial correlation analysis showed fragmented landscape (c. 70% of the Coates-Es- that when the influence of body mass and abundance rank species historically reported; trada and Estrada 1986). The majority of the species, were partialled out, only the latter variable was signif- however, were present in low numbers and their icantly correlated with the number of vegetation types in very representatives exist in isolated forest fragments scattered which species were present (rxy = 0.66, p = 0.001). in the landscape. In spite of our intensive and extensive surveys we did not detect 11 NFM species expected to appear in the forest fragments studied. These were small Undisturbed forests and habitats agricultural mammals such as Cryptotis mexicana and C. parva (Sor- Rarefaction curves showed that pastures were the poorest icidae), Liomys pictus (Heteromyidae), Oryzomysfulves- habitat in species. This analysis also showed that the most cens, Nyctomys sumichrasti, and Reithrodontomysfulves- diverse habitats were the forests followed by cacao, cof- cens (Muridae). Some of these species are arboreal, very fee and mixed plantations and the live fences. Of all the cryptic and nocturnal, features that would make them arboreal agricultural habitats examined, citrus and all- difficult to detect. It is possible that they were present in spice were the least diverse (Fig. 5). Polar ordination some of the forest habitats we studied. showed that the mixed, cacao and coffee plantations and Other larger species are known to be locally extinct in live fences were the habitats most similar to forest frag- many parts of the region. In this case we have Cyclopes ments in species assemblages. The least similar were didactylus (Myrmecophagidae), Felis onca (Felidae), Ta- citrus and allspice. Pastures were the most dissimilar with pirus bairdii (Tapiridae), Tayassu pecari (Tayassuidae) respect to the other habitats studied (Fig. 6). and Odocoileus virginianus (Cervidae). While small An examination of the vegetation data showed that (< 200 ha) forest fragments studied harboredmany mam- horizontal diversity of the vegetation (mean number of mal species, c. 45% of the species detected were present tree species detected per site/habitat (MTSSH)) and fo- only in large tracts (> 400 ha) of rain forest vegetation. In liage vertical diversity (FH') were related (rs = 0.97, p = this later case we detected large species such as Felis 0.001). Forests and shaded crops were the most complex concolor, F pardalis, M. americana, Ateles geoffroyii habitats. Partial correlation analysis showed that when and medium size carnivores such as Eira barbara and F the influence of MTSSH and FH' were partialled out, the ygouaroundi. The disappearance of some of these carni- mean number of NFM species detected per site in each vores from the remaining forests, may have important habitat and the mean number of individuals NFM per site consequences in the overall structureand dynamics of the in each habitat were associated only to the horizontal mammalian community and the ecosystem (Terborgh diversity of the vegetation (MNSSH r = 0.66, p = 0.03; 1992). MNMSH r = 0.68, p = 0.03). In our study, disturbed forest sites were significantly poorer in species and NFM biomass than undisturbed sites, probably the result of continued human activity, mainly hunting and removal of hardwoods and firewood. Non flying mammal biomass In contrast to disturbed forest sites, regenerating forests We found a negative association between total NFM contained significant numbers of mammals, with older biomass detected per site and area of forest fragments (rs regenerating forests supporting more NFM species and = -0.43, p = 0.005). Forest habitats had the highest mean biomass. biomass detected per site among the habitats investigated Proximity of regenerating forests to undisturbed forest (forest = 89 kg; man-made habitats = 6-33 kg (Table 1). fragments, the complexity of the vegetation along vertical Partial correlation analysis showed that mean biomass and horizontal dimensions as well as the presence of

ECOGRAPHY 17:3 (1994) 235 many fruiting tree species (e.g. Cecropia obtusifolia, Fi- with other troops at these plantations. We witnessed indi- cus spp., Bursera simaruba, Spondias mombin, Stemma- viduals of G. vitatta moving on the ground through the denia donnell-smithii) coupled with the presence of a pasture into agriculturalhabitats such as the cacao planta- diverse fauna of insects and small vertebrates (birds, tions and from here to nearby forest patches. We made flying mammals, small reptiles and amphibians) at these similar observations on S. aureogaster, P. lotor, N. nar- habitats (Estrada et al. 1993a,b, Estrada and Coates- ica, B. sumichrasti, P. flavus, D. marsupialis, D. virgin- Estrada pers. obs.) means the existence of protective iana and P. opossum. Since the farmers planted the trees cover, potential resting and nesting sites and food re- after clearing the forest in the case of the citrus and sources for NFM visiting and establishing residency in allspice plantations, we can assume that all the NFM these sites. The potential connection provided by live detected in these habitats probably originated from fences could enhance colonization of these habitats by nearby forest fragments or from the pasture habitat (e. g. NFM. S. hispidus). The almost total absence of forest NFM in pastures The conservation value for mammals of arboreal agri- suggests that these are unsuitable habitats for them. Even cultural islands as stepping stones and as temporary ref- species such as D. marsupialis, D. virginiana and uge habitats could be enhanced by the presence of live P. opossum that numerically dominated the total sample fences. In our study 27% of the species detected in the and were present in all habitats studied had higher detec- total sample and 53% of those detected in the man-made tion rates in the forest and in the arboreal man-made habitats were present at the live fence sites studied. The habitats than in pastures. Lack of food resources, the hundreds or thousands of linear meters of vegetation in extreme climatic conditions of pastures and the greater the form of live fences across the landscape are available exposure to predators in these habitats would make them to mammals inhabiting the many forest fragments in the inappropriate for continued mammal habitation. We ob- region. Some live fences end at the edge of forest patches served raptors such as Buteo magnirostris, Falco sparve- or interconnect with the forest vegetation remaining rius, F peregrinus and Herpethotheres cachinnans and along streams and rivers, probably enhancing biotic con- dogs chasing and preying on small NFM at the edge of nectivity in the area. For example, our records for forest forest fragments suggesting that predation may be greater mammals such as P. mexicanus, B. sumichrasti, P. flavus at the edge of the forest patches. and C. mexicanus are from those live fence sites that Shaded agricultural habitats supported not only more ended at the edge of a forest patch. species and individuals than unshaded habitats, but also The live fence trees (e.g. Bursera simaruba) also pro- contained a higher biomass of NFM represented by large duce edible fruit, adding to the opportunities available to arboreal mammals such as A. palliata and P. flavus, ab- NFM in a landscape where the forest is now a limited sent in the unshaded habitats. The presence at these resource. In those live fences where the vegetation has habitats of tall rain forest trees of plant families such as been allowed to regenerate under the trees, the presence Moraceae (Ficus spp., Poulsenia armata, Brosimum al- of high concentrations of plant species of the genera icastrum), Fabaceae (Pterocarpus rorhii, Lonchocarpus Piper (Piperaceae) Solanum (Solanaceae), Cecropia, Si- guatemalensis), Sapotaceae (Manilkara zapota, Pouteria paruna (Mominaceae), Eugenia (Myrtaceae), Psycotria campechiana), Anacardiaceae (Spondias radlkoferi), (Rubiaceae) and occasional strangler figs (Ficus spp.) Lauraceae (Nectandra ambigens, Ocotea spp.), Annona- suggests the occurrence of abundant food supplies (large ceae (Rollinia jimenezii) among others permits the exist- numbers of fruits produced per plant and year long avail- ence of a contiguous canopy cover for these mobile ability). These habitats may provide temporary suste- NFM. Some of these tree species are also an important nance not only for fruit-eating NFM, but also for those source of leaves and of fruit for the monkeys and other that consume insects and small vertebrates but comple- fruit-eating mammals (Estrada 1984, Estrada and Coates- ment their diet with important amounts of fruit. Allowing Estrada 1986). the vegetation to grow under the posts and preserving old Our data showed that isolating distance was a variable, live fences may be a favorable strategy to increase the that in conjunction with loss of area of natural habitat, number of NFM species and individuals visiting these influenced species richness and presence of NFM at the sites. forest fragments studied. While there is potential danger Our data indicated that NFM were present in habitats of predation and higher time and energy expenditure due other than forest patches, but the use of all the habitats to exposure and the distances covered, NFM species examined seemingly applies to only three species capable of reaching forest and man-made habitats outside (D. marsupialis, D. virginiana and P. opossum), the rest of the patch in which they reside may encounter a greater are less resilient. This is consistent with the generalist variety of habitats in which to find resources and meet habitat and feeding preferences of these mammals survival requirements. (Coates-Estrada and Estrada 1986, Julien-Leferriere and We observed troops of A. palliata visiting cacao, cof- Atramentowicz 1990, Emmons 1990). These species are fee and mixed plantations by moving through the trees of also characterizedby having high reproductive rates, high narrow strips of forest vegetation along streams and riv- fecundity and rapid population turnover (Atramentowicz ers. In some instances, the troops shared their presence 1986). At the other extreme were 20 species (e.g. Da-

236 ECOGRAPHY 17:3 (1994) syprocta mexicana, Tamandua mexicana, Felis weidii) will probably be the only components of the NFM com- that occurred in forest habitats only and that have low munity in the landscape of Los Tuxtlas. These three reproductive rates, low fecundity and a slow population species are hardly representatives of the once rich and turnover (Robinson and Redford 1986, Eisenberg 1989). diverse non flying mammal fauna of the northernmost Thus, species with higher tolerance to habitat frag- representation of the Amazon rain forest in the American mentation and isolation will be those with larger intrinsic continent. population densities, requiring small home ranges, with habitat and with the to Acknowledgments- We thankthe LincolnPark Zoological So- generalized requirements ability and it's Scott Fundfor Researchfor move between forest It is not then ciety Neotropical support. fragments. surprising The Centerfor Field Researchand Earthwatchprovided addi- that most of the species with high numerical representa- tionalsupport and the help of volunteers.The ExplorersClub of tion in our censuses (e.g. D. marsupialis) were those New York and the U.S. Fish and Wildlife Service gave addi- tional and UNAM aid. We thankS. species that either have small home ranges, are very support providedlogistical Sader,The Universityof Maine,for the 1990 satelliteimagery. mobile, and that seem to have generalized feeding habits. It is important to note here that the mammals present in unshaded man-made habitats were those with very gener- alized habitat and feeding requirements (Appendix 1, Coates-Estrada and Estrada 1986). Forest interior mam- References mals with more specialized habits such as P. mexicanus, Antramentowicz,M. 1986. Dynamiquede populationchez trois - H. desmarestianus, B. sumichrasti and S. deppei (Em- marsupaiuxdidelphides de Guyane. Biotropica18: 136- mons were in the shaded habitats. 149. 1990) present only Coates-Estrada,R. and Estrada,A. 1986. Manualde identif- Likewise, the exploitation of persistent resources in time icacionde campode los mamiferosde la Estacionde Biolo- and space (e.g. leaves, Estrada 1984) has allowed small gia "LosTuxtlas". - Inst.Biologia, Univ. Nac. Autonomade of A. to continue in isolated Mexico. troops palliata surviving R. M. forest Dirzo, and Garcia, C. 1992. Ratesof deforestationin Los fragments. Tuxtlas,a neotropicalarea in southernMexico. - Cons.Biol. Hence, at Los Tuxtlas, we have a small number of 6: 84-90. NFM species that seem to be able to occupy a variety of Eisenberg,J. F 1989.Mammals of the Neotropics.The northern - man-made habitats. However, the proximity of these sites Neotropics.Vol. I. Univ. of ChicagoPress. Chicago. to forest and whether are shaded with rain Emmons,L. H. 1990. Neotropicalrain forest mammals:a field fragments they guide. - Univ. of ChicagoPress, Chicago. forest trees or not seems to influence the richness and Estrada,A. 1984. Resourceuse by howlermonkeys in the rain composition of the NFM fauna existing in these habitats. forestof Los Tuxtlas,Veracruz, Mexico. - Int. J. Primat.5: For some of these a model rather than 105-131. species variegated - a model of land be more 1992. Las selvas de Los Tuxtlas:islas de extincion o de fragmentation management may conservacionde la floray faunaen Veracruz.- In:Boege, E. adequate as a conservation approach at the landscape and Rodriguez,H. (eds), Desarrolloy medio ambienteen level (McIntyre and Barrett 1992). At the other extreme Veracruz.Ciesas-Golfo, Inst. de Ecologia,A.C. y Fundacion are about half of the detected in our censuses. FriedrichEbert Stiftung, Mexico, D.E pp. 59-68. species - These intolerance and exist in a and Coates-Estrada,R. 1986. A preliminarystudy of re- display truly fragmented source overlapbetween howling monkeys(Alouatta palli- landscape. For these, the conservation of the remaining ata) and otherarboreal mammals in the tropicalrain forest large tracts of forest is fundametal to ensure their contin- of Los Tuxtlas,Mexico. - Am. J. Primat.9: 27-37. ued existence. - andCoates-Estrada, R. 1988.Tropical rain forest conversion In such as Los Tuxtlas where c. 300 km2 and perspectives in the conservationof wild primates regions only (Alouattaand Ateles) in Mexico.- Am. J. Primat.14: 315- remain of the original 2500 km2 of tropical rain forest 327. vegetation and where these remnants occur in the form of - , Coates-Estrada,R. andMartinez, M. 1985. La Estacionde isolated forest a Biologia"Los Tuxtlas": un recursopara el estudioy conser- fragments (Estrada 1992), landscape - in which some vacionde las selvasdel tropicohumedo en Mexico. In:del arrangement agricultural parcels occupy Amo, S. and Gomez-Pompa,A. (eds), Regeneracionde of the pasture separating archipelagos of forest frag- selvas II. Inst.Nac. de Investigacionessobre Recursos Biot- ments, including forests in process of regeneration, may icos. EditorialAlhambra Mexicana, S.A. de C.V. Mexico, reduce isolating distances facilitating the flow of animals pp. 379-393 Such of the land , Coates-Estrada,R. and Meritt,Jr. D. 1993a. Bat species through space. management may pre- richnessand abundance in tropicalrain forest fragments and serve selected species of forest mammals and may have in agriculturalhabitats at Los Tuxtlas,Mexico. - Ecography several important added benefits including retention of 16: 309-318. soil and of soil fertility and preservation of water re- -, Coates-Estrada,R., Meritt,Jr. D., Montiel,S. and Curiel, D. 1993b.Patterns of richnessand abun- sources (Gleissman et al. 1981). frugivorespecies dancein forestfragments and in agriculturalhabitats at Los The predominance of D. marsupialis, D. virginiana TuxtlasMexico. - In: Fleming,T. H. and Estrada,A. (eds), and P. opossum in our surveys suggests that unless we act Frugivoresand seed dispersal:ecological and evolutionary promptly with alternative scenarios of land management aspects. Kluwer Academic Publ. Dordretch,The Nether- lands, 245-257. to stop rain forest destruction and to the effects pp. mitigate Fitch,R. 1992. WinSTAT.The StatisticsProgram for Windows. of habitat destruction, fragmentation and continued isola- - KalmiaCompany, Inc. Cambridge,MA. tion on the remaining NFM species, these three species Gleissman,S. R., Garcia,R. andAmador, M. 1981.The ecologi-

ECOGRAPHY 17:3 (1994) 237 cal basis for the applicationof traditionalagriculture tech- -, Bierregaard,R. O., Rylands,A., Malcolm,J. R., Quintela, nology in the managementof tropicalecosystems. - Agro- C. F., Harper,L. H., Brown,K. S., Powell,A. H., Powell,G. ecosystems7: 173-185. V. N., Schubart,H. O. R. andHays, M. B. 1986.- In: Soule, Hall,E. R. 1981.The mammalsof NorthAmerica. Vol. I andII. M. E. (ed.), Edge and othereffects of isolationon Amazon - John Wiley and Sons, New York. forestfragments. Conservation biology: a scienceof scarcity Ibarra,G. and Sinaca,S. 1987. Listadosfloristicos de Mexico and diversity.Sinauer Ass., Sunderland,pp. 257-285. VII. Estacionde Biologia TropicalLos Tuxtlas,Veracruz. - Ludwig,J. A. and Reynolds,J. F. 1988. - Statisticalecology. Institutode Biologia, UNAM, Mexico City. John Wiley and Sons, New York. James,F C. and Rathburn,S. 1981. Rarefaction,relative abun- McIntyre,S. and Barret,G. W. 1992. Habitatvariegation, an dance and diversityof avian communities.- The Auk 98: alternativeto fragmentation.- Cons. Biol. 6: 146-147. 785-800. Pennington,T. D. and Sarukhan,J. 1968. Arbolestropicales de Julien-Laferriere,D. andAtramentowicz, M. 1990. Feedingand Mexico. - Inst. InvestigacionesForestales, Mexico D.F reproductionof threedidelphid marsupials in two Neotrop- Robinson,J. and Redford,K. H. 1986. Intrinsicrate of natural ical forests (FrenchGuiana). - Biotropica22: 404-415. increasein Neotropicalforest mammals: relationship to phy- Karr,J. 1990.Avian survival rates and the extinctionprocess on logeny and diet. - Oecologia68: 516-520. BarroColorado Island, Panama. - Cons. Biol. 4: 391-397. Saunders,D. A., Hobbs,R. J. andMargules, C. R. 1991.Biolog- Lovejoy, T. E., Bierregaard,R. O., Brown, K. S, Emmons,L. ical consecuencesof ecosystemfragmentation: a review. - and Van der Voort,M. E. 1984. - In: Niteki, M. H. (ed.), Cons. Biol. 5: 18-32. Ecosystemdecay of Amazonforest fragments. Extinctions. Terborgh,J. 1992.Maintenance of diversityin tropicalforests. - Univ. of ChicagoPress, Illinois, pp. 295-325. Biotropica24: 283-292.

238 ECOGRAPHY 17:3 (1994) Appendix1. Non flying mammalspecies and individualscensused in forest and man-madehabitats. Species Forest Cacao Coffee Mixed Citrus All- Live Pas- Total Average spice fence ture mass g

MARSUPIAWA Didelphidae Marmosa mexicana 6 6 95 (Mexicanmouse opossum) Didelphis virginiana 70 4 12 8 2 1 9 3 109 3750 (Virginiaoposum) Didelphis marsupialis 183 8 16 48 6 13 15 16 305 2000 (commonopossum) Philander opossum 154 12 5 4 8 9 10 10 212 320 (commongray four-eyedopossum) Caluromys derbianus 10 10 400 (CentralAmerican woolly opossum) PRIMATES Cebidae Allouata palliata 183 16 20 25 244 6750 (mantledhowler monkey) Ateles geoffroyii 10 10 6000 (CentralAmerican spider monkey) EDENTATA Myrmecophagidae Tamandua mexicana 2 2 5000 (northerntamandua) Dasypodidae Dasypus novemcinctus 20 7 5 9 3 2 3 49 5000 (nine-bandedlong-nosed armadillo) Leporidae Sylvilagus floridanus 2 2 (cotton-tailedrabbit) Sylvilagus brasiliensis 4 4 (Brazilianrabbit) RODENTIA Sciuridae Sciurus aureogaster 116 16 10 8 8 18 184 450 (red-belliedsquirrel) Sciurus deppei 98 2 100 213 (Deppe's squirrel) Geomyidae Orthogeomys hispidus 16 10 14 7 5 5 57 (pocketgopher) Heteromyidae Heteromys desmarestianus 108 12 120 73 (spiny pocket mice) Muridae Oryzomys palustris 26 26 65 (rice rats) Oryzomys alfaroi 10 10 63 (rice rats) Tylomys nudicaudus 4 4 215 (naked-tailedclimbing rats) Peromyscus leucopus 2 2 18 (white-footeddeer mouse) Peromyscus mexicanus 262 16 5 8 11 302 50 (Mexicandeer mouse) Sigmodon hispidus 20 4 64 88 159 (cottonrat) Erethizontidae Coendou mexicanus 8 6 14 3500 (Mexicanhairy porcupine) cont.

ECOGRAPHY 17:3 (1994) 239 Appendix1 (cont.) Species Forest Cacao Coffee Mixed Citrus All- Live Pas- Total Average spice fence ture mass g

Dasyproctidae Dasyprocta mexicana 14 14 3500 (Mexicanblack agouti) Agouti paca 6 6 7000 (pica) CARNIVORA Procyonidae Bassariscus sumichrasti 60 12 6 8 3 89 1000 (cacomistle) Procyon lotor 10 2 3 6 21 5000 (northernraccoon) Nasua narica 24 5 3 32 4000 (white-nosedcoati) Potos flavus 24 7 6 9 2 48 3500 (kinkajou) Mustelidae Mustela frenata 8 12 4 24 350 (long-tailedweasel) Eira barbara 4 4 3750 (tayra) Galictis vittata 8 3 11 3000 (grison) Conepatus semistriatus 8 2 1 3 2 4 20 2500 (stripedhog-nosed skunk) Lutra longicaudis 4 4 4500 (southernriver otter) Felidae Felis concolor 2 2 35000 (puma) Felis wiedii 2 2 4000 (margay) Felis yagouaroundi 10 10 5000 (jaguarundi) Felis pardalis 4 4 7000 (ocelot) ARTIODACTYLA Tayassuidae Tayassu tajacu 12 12 2200 (collaredpeccary) Cervidae Mazama americana 6 6 15000 (red brocketdeer) Subtotal 1518 144 106 145 38 42 83 93 2169 % 108 10 8 10 3 3 6 7 No. of species 38 16 13 13 8 7 11 4 39 % 97 41 33 33 21 18 28 10

240 ECOGRAPHY 17:3 (1994) Appendix2. Non flying mammalspecies and individualsdetected in disturbedand undisturbedforest habitats and in regenerating forests (6 sits per habitat). Species Undisturbed Disturbed Old Young forest forest growth growth

Peromyscus mexicanus 44 29 27 20 Didelphis marsupialis 30 17 16 17 Philander opossum 33 18 10 11 Sciurus aureogaster 8 6 15 10 Alouatta palliata 26 4 Didelphis virginiana 12 8 6 3 Heteromys desmarestianus 18 2 5 4 Bassariscus sumichrasti 12 1 1 1 Sciurus deppei 12 1 Oryzomys palustris 1 1 10 Sigmodon hispidus 4 1 4 1 Ateles geoffroyii 5 Potos flavus 1 1 1 Oryzomys alfaroi 2 1 Marmosa mexicana 3 Silvilagus brailiensis 1 1 1 Conepatus semistriatus 1 1 Caluromys derbianus 1 1 Silvilagus floridanius 2 Dasypus novemcinctus 1 Nasua narica 1 Procyon lotor 1 Coendou mexicanus 1 Dasyprocta mexicana 1 Tayassu tajacu 1 Felis yagouaroundi 1 Mazama americana I Agouti paca 1 Tylomys nudicaudus 1 Felis pardalis 1 Tamandua mexicana 1 Peromyscus leucopus 1 Felis wiedii 1 Felis concolor 1 Total individuals 226 94 88 79 Total species 30 17 12 10 Totalbiomass kg 388 203 56 40 Mean (? SD) biomassx site kg 73.6+45.4 23.8?14.9 9.5?3.5 6.8?4.7

16 ECOGRAPHY 17:3 (1994) 241