American Journal of Primatology 70:69–77 (2008)

RESEARCH ARTICLE
Landscape Attributes Affecting Patch Occupancy by Howler Monkeys (Alouatta
palliata mexicana) at Los Tuxtlas, Mexico
VÍCTOR ARROYO-RODRÍGUEZ1, SALVADOR MANDUJANO2, AND JULIETA BENÍTEZ-MALVIDO3
1
División de Posgrado, Instituto de Ecologı´a A.C., Congregación el Haya, Xalapa, Veracruz, Mexico
2
Departamento de Biodiversidad y Ecologı´a Animal, Instituto de Ecologı´a A.C., Xalapa, Veracruz, Mexico
3
Centro de Investigaciones en Ecosistemas, Universidad Nacional Autónoma de Me´xico, Morelia, Michoacán, Mexico

Understanding how forest cover is related to patch attributes such as size, shape, and isolation, and how
this influences the occurrence of a species in fragmented landscapes is an important question in landscape
ecology and conservation biology. To study the effects of fragmentation on patch occupancy by the
critically endangered Mexican mantled howler monkey (Alouatta palliata mexicana) in the Los Tuxtlas
region of Mexico, we selected three landscapes of ca. 5,000 ha, which differed in their remaining forest
cover (24, 11, and 4%). For each landscape, we related patch occupancy to forest cover, patch size and
shape, and four isolation parameters. Landscape attributes varied according to forest cover, and the
percentage of occupation was greater in landscapes with more forest cover. The attributes affecting the
probability of occupancy differed among landscapes. Occupancy was positively related to patch size in all
landscapes, but in the northernmost landscape, shape irregularity had a negative effect on occupancy,
whereas in the southernmost landscape, occupancy was favored by greater distances to the nearest
village. The results show that not only the total amount of forest cover but also patch configuration need
to be taken into consideration when designing management strategies for the conservation of the
Mexican mantled howler monkey. Am. J. Primatol. 70:69–77, 2008.
c 2007 Wiley-Liss, Inc.

Key words: fragmentation; landscape structure; Mexican mantled howler monkey; Neotropical
primates

INTRODUCTION ments [Chapman et al., 2003; Marsh, 2003; Mbora &

Fragmentation results in the reduction of forest
coverage and in changes of landscape attributes, such
as an increase in the number of forest patches, a
decrease in patch size and increased isolation [An-
drén, 1994]. These shifts in landscape structure affect
other ecological processes such as population dy-
namics and dispersal mechanisms [e.g., Fahrig,
1998]. In this context, identification of the factors
influencing the presence of species in fragmented
landscapes may be fundamental to the development of
effective strategies of conservation and management.
Metapopulation theory predicts that the prob-
ability of local extinction in fragmented habitats
increases with decreasing patch size and increasing
isolation, whereas the probability of colonization
follows the opposite trend [Hanski, 1999]. Several
studies with primates support these predictions
[Anzures-Dadda & Manson, 2007; Chapman et al.,
2003; Mandujano et al., 2006; Swart & Lawes, 1996].
Initially, changes in forest cover result in a random
occupancy of the remaining patches, but over time,
primates may disappear from some fragments owing
to their small size or inadequate characteristics,
whereas they may recolonize others suitable frag-
Meikle, 2004]. However, in landscapes with a large
proportion of remaining habitat, the effects of
fragmentation may be primarily a consequence of
habitat loss, whereas in highly fragmented habitats,
the decline in population size can be greater than
expected from habitat loss alone because of the
synergistic effects of habitat loss, reduced patch size,
and increased isolation [Andrén, 1994; Fahrig, 1998,
2003; Flather & Bevers, 2002; Harrison & Bruna,
1999].
We analyzed patterns of patch occupancy by
Mexican mantled howler monkeys (Alouatta palliata
mexicana) in the Los Tuxtlas region, southeastern
Contract grant sponsors: Department of Biodiversity and Animal
Ecology at the Instituto de Ecologı́a A.C.; Dirección General de
Relaciones Internacionales-SEP.
Correspondence to: Vı́ctor Arroyo-Rodrı́guez, División de
Posgrado, Instituto de Ecologı́a A.C., Km 2.5 Antigua Carretera
Coatepec No. 351, Congregación el Haya, Xalapa 91070,
Veracruz, México. E-mail: victorarroyo_rodriguez@hotmail.com
Received 1 February 2007; revised 30 April 2007; revision
accepted 21 May 2007
DOI 10.1002/ajp.20458
Published online 21 June 2007 in Wiley InterScience (www.
interscience.wiley.com).

r 2007 Wiley-Liss, Inc.

70 / Arroyo-Rodrı́guez et al.
Mexico. Los Tuxtlas region has suffered severe
fragmentation [Dirzo & Garcia, 1992; Guevara
et al., 2004] and the remaining howler populations
are located in forest patches that vary in size and
degree of isolation [e.g., Cristóbal-Azkarate et al.,
2005]. Previous research has been focused on a single
landscape unit and has shown that both patch size
and isolation affect the presence and abundance of
howlers [Estrada & Coates-Estrada, 1996; Estrada
et al., 1999; Mandujano et al., 2006]. In this study, a
regional approach was adopted, in which landscapes
with different amounts of forest cover were analyzed
comparatively, and with reference to important
additional variables such as patch shape, distance
to roads and villages. This approach may improve
our understanding of how primates respond to
habitat fragmentation, given that these variables
may be related to: (1) food resources and vegetation
structure [Arroyo-Rodrı́guez & Mandujano, 2006];
and (2) hunting pressure and other anthropogenic
disturbances [Chiarello & de Melo, 2001; Peres,
2001]. Particularly, as fragments become smaller
and more irregularly shaped, their floristic composi-
tion and vegetation structure are increasingly mod-
ified [Benı́tez-Malvido, 1998; Harrison & Bruna,
1999; Laurance et al., 2002], decreasing the quality
and quantity of food resources for howlers [Arroyo-
Rodrı́guez & Mandujano, 2006].
In this paper, we address the effects of forest
cover, and patch size, shape, and isolation, on the
presence of howler monkey populations in 208 forest
fragments in three landscapes representing a gra-
dient of forest loss and fragmentation. Specifically,
we (1) analyze the relationships between forest cover
and six landscape attributes (patch size and shape,
and four isolation parameters); (2) identify the
landscape attributes affecting patch occupancy with-
in each landscape unit; and (3) discuss implications
for the conservation of howler monkeys. This is a
timely study for the conservation of this endangered
primate, given that its populations have declined by
90% since 1940 in Los Tuxtlas [Estrada & Coates-
Estrada, 1994], the northernmost area of the
distribution of howler monkeys.
METHODS
Study Site
The Los Tuxtlas region is in the southeast of the
state of Veracruz, Mexico (181 80 –181450 N, 941
370 –951 220 W; Fig. 1). The region was decreed a
Special Biosphere Reserve in 1998 owing to its
biodiversity and its cultural importance [Diario
Oficial de la Federación, 1998; Laborde, 2004]. The
reserve covers 155,122 ha [Laborde, 2004]. The
original dominant vegetation type at altitudes below
700 m asl was tropical rain forest, but this region has
been severely fragmented over the past 60 years and
the remaining forest is surrounded by a matrix of
Am. J. Primatol. DOI 10.1002/ajp
pastures and croplands [Castillo-Campos & Laborde,
2004; Dirzo & Garcia, 1992; Guevara et al., 2004].
The climate is humid and warm, with a mean annual
temperature of 251C and annual rainfall that
oscillates between 3,000 and 4,600 mm [Soto, 2004].
The Study Landscapes
We selected three landscapes that: (1) represent
a gradient of forest loss and fragmentation; (2) are
situated between 0 and 400 m asl; (3) are naturally
delimited by the coast of the Gulf of Mexico and large
rivers; and (4) encompass approximately the same
area (ca. 5,000 ha; Fig. 1, Table I). The fact that the
landscapes are delimited by large rivers and isolated
by 30 km (northern-central and central-southern;
Fig. 1) permit us to consider them as distinct units
for howler populations. Taking into account the rates
of deforestation at Los Tuxtlas between 1972 and
1993 [Guevara et al., 2004], we assumed that all
patches were of approximately the same age and
therefore this variable was not considered in the
analysis.
Landscape Attributes
We digitized the three landscapes with ArcView
3.2 (Environmental System Research Institute, Inc.,
USAs) software, using aerial photographs [1:20,000;
INEGI, 1999], orthophotos [INEGI, 1996], and
digital [INEGI, 1990] and field data. We defined
‘‘patch’’ as a remnant of the original forest, with a
mean canopy height Z10 m, and a surface area
Z0.5 ha, which appear to be the minimum para-
meters for patch occupation by A. p. mexicana in Los
Tuxtlas [Cristóbal-Azkarate et al., 2005; Rodrı́guez-
Toledoet al., 2003]. We calculated size, isolation, and
shape for all patches using the Patch Analyst 2.2
extension. Four isolation parameters were used:
distance to the nearest patch, occupied patch, road,
and village. We assigned each patch paffiffiffiffiffiffi
shape
ffi index
[Forman & Godron, 1986]: IF ¼ P= Ap; where P
and A are the perimeter and area, in meters. Index
values vary from 1, for a circular shape, to 5 for a
highly irregular shape. All forest patches were
considered in the analysis.
Preliminary analysis showed that the study
landscapes represent a gradient of loss and fragmen-
tation of forest cover (Table I, Fig. 1). The northern
landscape had a larger proportion of remaining cover
(24%, in 74 patches) followed by the southern (11%,
in 88 patches) and central (4%, in 46 patches)
landscapes. Mean patch size (7SD) in the northern
landscape (15.2740.3 ha) was significantly greater
(F2,205 5 3.3, P 5 0.04) than in the southern
(6.2712.8 ha) and central (4.7710.3 ha) landscapes
(Table I). With respect to size, 80% of the patches in
the northern landscape were smaller than 8 ha, and
11% larger than 32 ha. In the southern and central
landscapes, 89% of the patches were smaller than
 Patch Occupancy by Howler at Los Tuxtlas / 71

Fig. 1. Location of the three landscapes units studied at the Los Tuxtlas Biosphere Reserve, southeastern Veracruz, Mexico. Continuous
forest, and patches occupied by howler monkeys are indicated in the map.

TABLE I. Attributes of the Three Landscapes Studied at Los Tuxtlas, Mexico

Northern Southern Central F2,205

Landscape attributes
Total area (ha) 4,656 4,965 5,046 NT
Total forest cover (ha) 1,107 542 216 NT
Percentage of forest cover 23.8% 10.9% 4.3% NT
Number of forest patches 74 88 46 NT
Mean (7SD) patch size (ha) 15.2 (40.3)A 6.1 (12.7)B 4.7 (10.3)B 3.3
Mean (7SD) shape index 2.2 (1.0)A 1.9 (0.7)B 1.8 (0.6)B 4.1
Mean (7SD) distance (m) to:
Nearest patch 102.7 (172.0)B 112.7 (99.9)B 288.7 (299.8)A 17.0
Nearest occupied patch 173.3 (193.1)B 373.2 (314.9)B 1496.3 (1186.6)A 74.5
Nearest road 423.5 (519.0) 443.9 (518.3) 333.0 (435.1) 0.8 NS
Nearest village 1462.5 (1240.4)B 889.8 (662.6)A 1725.3 (1175.7)B 12.2
Number of villages 4 8 3 NT
Patches occupied by Alouatta palliata
Number of occupied patches 20 18 4 NT
Percentage of occupancy 27.0% 20.5% 8.7% NT
Differences among landscapes for each attribute were analyzed with analyses of variance. Different capital letters used as superscript indicate which means
significantly differ among landscapes (based on Tukey’s post hoc comparison).
Po0.05; **Po0.001.
NS, not significant; NT, not tested.

Am. J. Primatol. DOI 10.1002/ajp

72 / Arroyo-Rodrı́guez et al.

8 ha, and only 5 and 2%, respectively, were larger
than 32 ha (Table II).
Larger, patches in the northern landscape were
significantly more irregular in shape (mean7SD,
IF 5 2.271.0; F2,205 5 4.1, P 5 0.001) than in the
southern (1.970.7) and central (1.870.6) landscapes
(Tables I and II). Mean patch isolation (7SD), as
measured by the distance to the nearest patch and
distance to the nearest occupied patch, in the
northern landscape (1037172 and 1737193 m,
respectively) was significantly lower than in the
other landscapes (Table I); whereas patches in the
central landscape were the most isolated (2897300
and 1,49671,187 m, respectively). Distances be-
tween patches and villages were significantly lower
in the southern landscape (Tables I and II).
Sampling of Howler Monkey Populations
To determine the occurrence of howler monkeys
within the three landscape units, all forest patches
were sampled between 2004 and 2005 using stan-
dardized procedures [Rodrı́guez-Toledo et al., 2003].
Patch occupancy was defined as the presence of at
least one individual in a given fragment. To confirm
occupancy, we made slow treks around and inside
each patch, generally from 6 in the morning to
midday and from 4 to 7 in the afternoon. The typical
howling vocalization of the species allowed us to
locate the groups within the fragments. Search time
depended on patch size, with 5–6 hr being dedicated
to the survey of small fragments (o5 ha), 8 hr in
medium-sized patches (5–10 ha), and 24–48 hr in
larger fragments. Nevertheless, as larger patches
tended to have larger howler populations [Cristóbal-
Azkarate et al., 2005; Estrada & Coates-Estrada,
1996], confirmation of the presence of howlers often
took less time in larger patches. To prevent failure in
detection owing to small population size or density
[Gu & Swihart, 2004], we sampled each patch two or
three times per year.
Data Analysis
To test attribute differences among landscapes,
we used analysis of variance with Tukey’s post hoc

TABLE II. Distribution of Forest Patches by Size, Shape and Isolation in the Three Focal Landscapes of Los
Tuxtlas, Veracruz, Mexico, and Occupancy by Howler Monkeys (Alouatta Palliata)

Northern landscape Southern landscape Central landscape

Total Occupied Total Occupied Total Occupied

(N 5 74) (n 5 20) (N 5 88) (n 5 18) (N 5 46) (n 5 4)

Patch size (ha)

0–1 12 0 12 0 10 0
41–2 17 1 25 2 13 0
42–4 13 2 23 2 14 2
44–8 17 6 18 5 4 0
48–16 5 2 5 4 3 1
416–32 2 1 1 1 1 0
432–64 3 3 2 2 0 0
465–128 3 3 2 2 1 1
4128 2 2 0 0 0 0
Shape index
1–2 43 7 57 9 34 2
42–3 17 8 25 5 9 2
43–4 6 2 5 3 3 0
44–5 8 3 1 1 0 0
Distance to nearest forest patch (m)
0–50 43 17 25 8 8 0
450–100 14 0 29 7 10 1
4100–200 9 1 20 3 5 2
4200–400 3 0 12 0 10 0
4400–800 3 1 2 0 10 1
4800 2 1 0 0 3 0
Distance to nearest occupied forest patch (m)
0–50 22 9 7 4 0 0
450–100 13 2 13 4 1 0
4100–200 18 2 9 1 3 0
4200–400 15 4 23 6 7 2
4400–800 4 2 25 3 4 0
4800 2 1 11 0 31 2

Am. J. Primatol. DOI 10.1002/ajp


comparisons. To identify the landscape attributes
that influence the probability of occupancy (pre-
sence/absence of howlers), we used multiple logistic
regression analysis with generalized linear models
[GLM; Crawley, 2002], except in the case of the
central landscape, which presented highly unba-
lanced data (four occupied patches vs. 42 unoccupied
ones). The GLM permit us to elaborate statistical
models analogous to multiple regression analysis.
However, whereas in a regression analysis the data
must have a normal distribution, that is, the
distribution of residuals must be normal, with
GLM the structure of the error distribution is
analyzed by a link-function, which is related to a
specific distribution function (e.g., normal, Poisson,
gamma, binomial). We fixed a binomial error (to a
binary response variable), and corrected for over-
dispersion [Crawley, 2002], and used Akaike’s
information criterion to select the most parsimo-
nious model, that is, the combination of independent
variables (i.e., patch size, shape, and four isolation
distances) that best explained the probability of a
patch being occupied [Motulsky & Christopoulos,
2003]. To reduce collinearity between the predictor
variables and multivariate models, we checked the
variance inflation factor. Redundant variables were
deleted where variance inflation factor values were
above 4, which indicate possible multicollinearity
[Chatterjee et al., 2000]. Specifically, the distance to
the nearest forest patch was not included in the
model for the northern landscape, because this
variable was highly correlated with the distance to
the nearest occupied patch (r 5 0.83, Po0.001) and
nearest village (r 5 0.77, Po0.001). Similarly, the
distance to the nearest road was not included in the
model for the southern landscape because it was
correlated with the distance to the nearest village
(r 5 0.71, Po0.001).
After generating the models for each landscape
independently, we tested their accuracy by using the
model for each landscape to predict patch occupancy
TABLE III. Landscape Attributes Influencing Patch Occupancy by Howler Monkeys Within Two Landscapes at Los
Tuxtlas, Mexico
Parameter
Northern landscape’s generalized linear model:
Y 5 1/[1+exp(0.66910.294  Patch size1.100  Shape index)]
Patch size 0.294
Shape index 1.100
Intercept 0.669
Southern landscape’s generalized linear model:
Y 5 1/[1+exp(7.65910.632  Patch size10.003  Distance to nearest village)]
Patch size 0.632
Distance to nearest village 0.003
Intercept 7.659
The attributes were selected through generalized linear models. We indicated the mathematical formula of each model; where Y is the response variable
(occupancy). The sign of each parameter indicates the relationship (positive or negative) between each factor and the response variable.
Patch Occupancy by Howler at Los Tuxtlas / 73
in the other landscapes, and estimating the propor-
tion of patches classified correctly by each model.
Following Mbora and Meikle [2004] protocol, we
considered patches to be occupied where the pre-
dicted probability of occupancy was Z0.50. All
analyses were performed using S-Plus 2000 for
Windows [Anon, 1999].
RESULTS
Patch Occupancy by Howlers
A larger proportion of patches in the northern
landscape were occupied (27.0%) than in the southern
(20.5%) and central (8.7%) landscapes (Table I, Fig. 1).
All patches with more than 32 ha in size were
occupied by howlers, whereas 43, 83, and 25% of
medium-sized patches (8–32 ha) were occupied in the
northern, southern, and central landscapes, respec-
tively. Occupation rates in the small fragments
(o8 ha) were 19, 14, and 7%, respectively. No patches
smaller than 1 ha were occupied. Regarding the
isolation and shape of the patches, around 75–100%
of occupied patches were at distances below 200 m to
the nearest patch, and presented more regular
shapes—IFo3 (Table II). Overall, considering the
three landscapes 33% of less isolated patches (o50 m)
were occupied by howlers, in contrast to only 7% of
the most isolated patches (4200 m). If we considered
the distance to the nearest occupied patch, these two
values (%) are greater, with 45% of occupation in the
less isolated patches, and with 16% of occupation in
the most isolated patches.
Landscape Attributes Affecting Patch
Occupancy
For the northern and southern landscapes, the
probability of patch occupancy increased with patch
size (Table III). However, different factors influenced
patch occupancy in the two landscapes (Table III). In
the northern landscape occupation decreased in the
SE w2 P
0.095 9.57 0.002
0.515 4.56 0.033
0.787 0.72 0.395
0.115 30.08 0.000
0.000 31.70 0.000
1.084 49.89 0.000
Am. J. Primatol. DOI 10.1002/ajp
74 / Arroyo-Rodrı́guez et al.

TABLE IV. Correctly Classified Patches as changes in total habitat amount induce shifts in
Occupied or Unoccupied by Howler Monkeys landscape structure [Andrén, 1994; Fahrig, 2003;
Flather & Bevers, 2002; Turner et al., 1989]. Several
Predicteda studies support the idea that below a certain
proportion of suitable habitat—e.g.,o10–30% [An-
Occupied Unoccupied % correct
drén, 1994], o20% [Fahrig, 1998], o30–50%
Northern-southern [Flather & Bevers, 2002]—landscape attributes such
Actually occupied 7 11 38.9 as patch size and isolation become very important
Actually unoccupied 0 70 100.0 factors in explaining the distribution, abundance,
Overall classification 87.5 and richness of a species. As forest covered in this
accuracy study was at the lower end of this range of values, we
Northern-central expected landscape attributes to influence strongly
Actually occupied 2 2 50.0
the pattern of howler monkey occupancy in Los
Actually unoccupied 3 39 92.9
Overall classification 89.1
Tuxtlas.
accuracy Our results suggest that there is a positive
Southern-northern relationship between the total amount of forest cover
Actually occupied 16 4 80.0 and the proportion of occupied patches, which is
Actually unoccupied 19 35 64.8 consistent with Venier and Fahrig [1996]. Because
Overall classification 68.9 the proportion of occupied sites can be positively
accuracy correlated with the abundance of individuals, the
Southern-central landscapes with a greater proportion of suitable
Actually occupied 3 1 75.0 habitat are predicted to support more individuals.
Actually unoccupied 16 26 61.9
Our results support this prediction, as 316 indivi-
Overall classification 63.0
accuracy
duals are reported to inhabit 21 patches in the
northern landscape [Cristóbal-Azkarate et al., 2005];
Classification was done throughout logistic regression models generated 73 individuals are reported to inhabit 18 patches in
for each of two landscapes. The information shows the landscape used to
generate each model, and the landscape pair used to make the predictions
the southern landscape [Mandujano et al., 2006];
(indicated by -). For example, northern-southern means that the whereas approximately 40 individuals inhabit four
model calculated for the northern landscape was used to predict the patches in the central landscape (25 inhabit a single
occupied and unoccupied patches of the southern landscape.
a
Fragments with a probability Z0.50 were considered occupied. patch; Arroyo-Rodrı́guez, personal observation).
The number of larger and less isolated patches
in the northern landscape could explain the higher
occupation rate recorded in this landscape, by
patches with the most irregular shapes (Wald’s reducing local rates of extinction and favoring
w2 5 4.56, P 5 0.03), whereas in the southern land- dispersal between patches [e.g., Hanski, 1999]. By
scape the probability of occupancy increased with contrast, the lowest proportion of occupied patches
increasing distance to the nearest village (w2 5 31.70, was recorded in the central landscape, which also
Po0.001) (Table III). had a greatly reduced forest cover, consisting of
Using the classification rule, 63 to 89% of the relatively small and isolated patches.
patches were correctly classified by the models
(Table IV). Taking into account the percentage of
correctly classified occupied patches, the model for Landscape Attributes Affecting Patch
the southern landscape was the most accurate. Occupancy by Howler Monkeys
However, the model for the northern landscape best Patch size appears to be the main factor
predicted which patches were unoccupied (93 to influencing the presence of howler monkeys in Los
100% accuracy). Depending on the model, the Tuxtlas. Our findings agree with metapopulation
predicted probability of occupancy for 20 to 61% of theories [e.g., Hanski, 1999], and with other studies
the patches occupied by howlers was o50%. Con- of primate species from the Old World [Chapman
versely, between 0 and 38% of unoccupied patches et al., 2003; Swart & Lawes, 1996] and the New
had a predicted probability of occupancy higher than World [Anzures-Dadda & Manson, 2007; Chiarello,
50% (Table IV). 2003; Cristóbal-Azkarate et al., 2005; Gilbert, 2003;
Estrada et al., 1999]. Larger fragments can sustain
DISCUSSION larger populations [Anzures-Dadda & Manson, 2007;
Cristóbal-Azkarate et al., 2005; Estrada & Coates-
Forest Cover at the Landscape Level Estrada, 1996; Wahungu et al., 2005; Wieczkowski,
Our results show that as the total amount 2004], decreasing the probability of extinction
of forest cover decreases, mean patch size also that results from demographic and environmental
decreases, the shape of the patches becomes less stochasticity [Akc- akaya et al., 1999; Hanski, 1999].
irregular and isolation increases, indicating that At Los Tuxtlas, fragment size may also limit food

Am. J. Primatol. DOI 10.1002/ajp


availability, given that fragment size has been
positively related to home range size [Cristóbal-
Azkarate & Arroyo-Rodrı́guez, 2007] and the quan-
tity and quality of food resources [Arroyo-Rodrı́guez
& Mandujano, 2006].
The influence of patch shape on occupancy has
received less attention. In the northern landscape—
where patches were more irregular overall—the
probability of occupation decreased as shape
became more irregular. As fragments become smal-
ler and more irregularly shaped, the perimeter to
area ratio increase [Laurance & Yensen, 1991],
promoting numerous environmental changes at the
patch edge leading to alterations in vegetation
structure and composition [Arroyo-Rodrı́guez &
Mandujano, 2006; Benı́tez-Malvido, 1998; Harrison
& Bruna, 1999; Laurance et al., 2002]. For instance,
tree mortality rate can increase at the edge of the
fragments [e.g., Laurance et al., 1998, 2002].
Consistently, studies at Los Tuxtlas showed that
the smallest fragments had lower basal area of top
food resources for howlers and less large primary
tree species in the canopy [Arroyo-Rodrı́guez &
Mandujano, 2006]. Overall, vegetation changes can
affect the distribution of howlers in highly fragmen-
ted landscapes, as the presence of howlers in small
patches of Los Tuxtlas is highly associated with
the distribution and density of large trees [Arroyo-
Rodrı́guez et al., 2007].
The positive association between occupancy and
distance to the nearest village in the southern
landscape may reflect the influence of anthropogenic
pressures (e.g., hunting, logging), which have been
reported as one of the main causes for the decline of
howler populations in the study area [Rodrı́guez-
Luna et al., 1996]. The southern landscape has twice
as many villages as the other landscapes, and the
shortest distances between patches and villages. The
combination of these factors could increase anthro-
pogenic pressure, thus explaining why the patches
that occur at greater distances from the villages were
more likely to be occupied by howlers [see Chiarello
& de Melo, 2001; Peres, 2001]. However, direct
studies on anthropogenic activities should be incor-
porated to clearly understand the patterns of howler
abundance and distribution in the highly fragmented
landscape of Los Tuxtlas.
Models that incorporate information on land-
scape attributes may be useful tools for identifying
conservation priorities in the region (Table IV).
Patch size and distance to the nearest village
(southern landscape model) correctly predicted ap-
proximately three-quarters of patch occupancy in the
other landscapes, whereas patch size and shape
(northern landscape model) predicted the absence
of howlers in almost all the unoccupied patches in
the central and southern landscapes. In contrast,
between 20 and 67% of the patches occupied at the
time of this study had a predicted probability of
Patch Occupancy by Howler at Los Tuxtlas / 75
occupancy of o50%, implying that these patches
were probably not as suitable for occupation as
others and indicating the possibility of source-sink
dynamics in the system [Pulliam, 1988; Shmida &
Wilson, 1985]. Howlers in these patches may be at a
high risk of local extinction [see Mandujano et al.,
2006; Mbora & Meikle, 2004].
Implications for Conservation
Overall, our results show that at Los Tuxtlas: (1)
total forest cover is related to other landscape
attributes, which may determine the population
dynamics of howler monkeys; and (2) as the amount
of forest cover decreases, the proportion of occupied
patches decreases. Patch occupancy in the three
landscapes was associated with different landscape
attributes owing to differences in their spatial
structure. This implies that the efficacy of the
biosphere reserve will be improved if landscape-
specific strategies are incorporated in management
planning. Models incorporating information on land-
scape attributes such as patch size, shape, and
distance to villages may also serve as a useful tool
for identifying important potential conservation
areas in the region.
ACKNOWLEDGMENTS
The authors thank R. Mateo-Gutierrez and
family for their hospitality and invaluable help and
M. Ordano for his assistance with the statistical
analyses. We also appreciate the help of R. Palacios-
Silva, A. Escobedo-Morales, C. Cuende-Fantón, A.
González-Zamora, C. Domingo-Balcells and C. Perez-
Moscoso. K. J. Feeley and two anonymous reviewers
made valuable comments and suggestions on the
final version of the manuscript. The Department of
Biodiversity and Animal Ecology at the Instituto de
Ecologı́a A.C., and the Dirección General de Rela-
ciones Internacionales-SEP, provided financial sup-
port for the completion of this research. The research
complied with protocols approved by the appropriate
institutional animal care committee of SEMARNAT-
the Mexican Office for the Environment and Natural
Resources, and adhered to the legal requirements of
Mexico.
REFERENCES
Akc- akaya HR, Burgman MA, Ginzburg LR. 1999. Applied
population ecology. Principles and computer exercises using
RAMASs Ecolab 2.0. Sunderland: Sinauer Associates, Inc.
Andrén H. 1994. Effects of habitat fragmentation on birds and
mammals in landscapes with different proportions of
suitable habitat: a review. Oikos 71:340–346.
Anon. 1999. S-Plus 2000 User’s Guide. MathSoft, Inc.
Anzures-Dadda A, Manson RH. 2007. Patch- and landscape-
scale effects on howler monkey distribution and abundance
in rainforest fragments. Anim Conserv 10:69–76.
Am. J. Primatol. DOI 10.1002/ajp
76 / Arroyo-Rodrı́guez et al.
Arroyo-Rodrı́guez V, Mandujano S. 2006. Forest fragmenta-
tion modifies habitat quality for Alouatta palliata. Int J
Primatol 27:1079–1096.
Arroyo-Rodrı́guez V, Mandujano S, Benı́tez-Malvido J,
Cuende-Fantón C. 2007. The influence of large tree density
on howler monkey (Alouatta palliata mexicana) presence in
very small rainforest fragments. Biotropica (DOI 10.1111/j.
1744-7429.2007.00330.x).
Benı́tez-Malvido J. 1998. Impact of forest fragmentation on
seedling abundance in a tropical rain forest. Conserv Biol
2:380–389.
Castillo-Campos G, Laborde D. 2004. La vegetación. In:
Guevara S, Laborde J, Sánchez-Rı́os G, editors. Los Tuxtlas.
El paisaje de la sierra. Xalapa: Instituto de Ecologı́a A.C.,
and Unión Europea. p 231–265.
Chapman CA, Lawes MJ, Naughton-Treves L, Gillespie T.
2003. Primate survival in community-owned forest frag-
ments: Are metapopulation models useful amidst intensive
use? In: Marsh LK, editor. Primates in fragments: ecology
and conservation. New York: Kluwer Academic/Plenium
Publishers. p 63–78.
Chatterjee S, Hadi AS, Price B. 2000. Regression analysis by
example. New York: Wiley.
Chiarello AG. 2003. Primates of the Brazilian Atlantic
forest: the influence of forest fragmentation on survival.
In: Marsh LK, editor. Primates in patches: ecology and
conservation. New York: Kluwer Academic/Plenum Press.
p 99–121.
Chiarello AG, de Melo FR. 2001. Primate population densities
and sizes in Atlantic forest remnants of northern Espı́rito
Santo, Brazil. Int J Primatol 22:379–396.
Crawley MJ. 2002. Statistical computing. An introduction to
data analysis using S-Plus. New York: Wiley.
Cristóbal-Azkarate J, Arroyo-Rodrı́guez V. 2007. Diet and
activity pattern of howler monkeys (Alouatta palliate) in Los
Tuxtlas, Mexico: effects of habitat fragmentation and
implications for conservation. Am J Primatol 69 (DOI
10.1002/ajp.20420).
Cristóbal-Azkarate J, Veà JJ, Asensio N, Rodrı́guez-Luna E.
2005. Biogeographical and floristic predictors of the pre-
sence and abundance of mantled howlers (Alouatta palliata
mexicana) in rainforest fragments at Los Tuxtlas, Mexico.
Am J Primatol 67:209–222.
Diario Oficial de la Federación. 1998. Tomo DXLII No. 16.
Decreto Presidencial de la Reserva de la Biosfera ‘‘Los
Tuxtlas’’.
Dirzo R, Garcı́a MC. 1992. Rates of deforestation in Los
Tuxtlas, a Neotropical area in Veracruz, Mexico. Conserv
Biol 6:84–90.
Estrada A, Coates-Estrada R. 1994. La contracción y fragmen-
tación de las selvas y las poblaciones de primates silvestres:
El caso de Los Tuxtlas, Veracruz. La Ciencia y El Hombre
18:45–70.
Estrada A, Coates-Estrada R. 1996. Tropical rain forest
fragmentation and wild populations of primates at Los
Tuxtlas, Mexico. Int J Primatol 17:759–783.
Estrada A, Anzurez A, Coates-Estrada R. 1999. Tropical rain
forest fragmentation, howler monkeys (Alouatta palliate),
and dung beetles at Los Tuxtlas, Mexico. Am J Primatol
48:253–262.
Fahrig L. 1998. When does fragmentation of breeding habitat
affect population survival? Ecol Model 105:273–292.
Fahrig L. 2003. Effects of habitat fragmentation on biodiver-
sity. Ann Rev Ecol Evol Syst 34:487–515.
Flather CH, Bevers M. 2002. Patchy reaction-diffusion and
population abundance: the relative importance of habitat
amount and arrangement. Am Nat 159:40–56.
Forman RTT, Godron M. 1986. Landscape ecology. New York:
Wiley.
Gilbert KA. 2003. Primates and fragmentation of the Amazon
forest. In: Marsh LK, editor. Primates in patches: ecology
Am. J. Primatol. DOI 10.1002/ajp
and conservation. New York: Kluwer Academic/Plenum
Press. p 145–157.
Gu W, Swihart RK. 2004. Absent or undetected? Effects of
non-detection of species occurrence on wildlife-habitat
models. Biol Conserv 116:195–203.
Guevara S, Sánchez-Rı́os G, Landgrave R. 2004. La deforesta-
ción. In: Guevara S, Laborde J, Sánchez-Rı́os G, editors. Los
Tuxtlas. El paisaje de la sierra. Xalapa: Instituto de Ecologı́a
A.C., and Unión Europea. p 85–108.
Hanski I. 1999. Metapopulation ecology. Oxford: Oxford
University Press.
Harrison S, Bruna E. 1999. Habitat fragmentation and large-
scale conservation: what do we know for sure? Ecography
22:225–232.
INEGI. 1990. Conjuntos de datos vectoriales de la carta San
Juan Volador E15A74, escala 1:75000. Xalapa: Instituto
Nacional de Estadı́stica Geográfica e Informática.
INEGI. 1996. Ortofotos digitales E15A74b, E15A74c, E15A74e
y E15A74f de la carta San Juan Volador E15A74, escala
1:75000. Xalapa: Instituto Nacional de Estadı́stica Geográ-
fica e Informática.
INEGI. 1999. Fotografı́as aéreas escala 1:20000. Vuelo Espe-
cial Sierra de Los Tuxtlas, Xalapa: Instituto Nacional de
Estadı́stica Geográfica e Informática.
Laborde J. 2004. La Reserva de la Biosfera. In: Guevara S,
Laborde J, Sánchez-Rı́os G, editors. Los Tuxtlas. El paisaje
de la sierra. Xalapa: Instituto de Ecologı́a A.C., and Unión
Europea. p 271–279.
Laurance WF, Yensen E. 1991. Predicting the impacts of edge
effects in fragmented habitats. Biol Conserv 55:77–92.
Laurance WF, Ferreira LV, Rankin-de Merona JM, Laurance
SG, Hutchings RW, Lovejoy TE. 1998. Effects of forest
fragmentation on recruitment patterns in Amazonian tree
communities. Conserv Biol 12:460–464.
Laurance WF, Lovejoy TE, Vasconcelos HL, Bruna EM,
Didham RK, Stouffer PC, Gascon C, Bierregaard RO,
Laurance SG, Sampaio E. 2002. Ecosystem decay of
Amazonian forest fragments: a 22-year investigation. Con-
serv Biol 16:605–618.
Mandujano S, Escobedo LA, Palacios-Silva R, Arroyo-Rodrı́-
guez V, Rodrı́guez-Toledo EM. 2006. A metapopulation
approach to conservation of howler monkeys in highly
altered landscape in Mexico. In: Estrada A, Garber PA,
Pavelka M, Luecke L, editors. New perspectives in the study
of Mesoamerican primates: distribution, ecology, behavior,
and conservation. New York: Kluwer Academic/Plenium
Publishers. p 513–538.
Marsh LK. 2003. The nature of fragmentation. In: Marsh LK,
editor. Primates in fragments: ecology in conservation. New
York: Kluwer Academic/Plenum Publishers. p 1–10.
Mbora DMN, Meikle DB. 2004. Forest fragmentation and the
distribution, abundance and conservation of the Tana River
red colobus (Procolobus rufomitratus). Biol Conserv
118:67–77.
Motulsky HJ, Christopoulos A. 2003. Fitting models to
biological data using linear and nonlinear regression. A
practical guide to curve fitting. San Diego: GraphPad
Software Inc. www.graphpad.com. Downloaded on 20 July
2005.
Peres CA. 2001. Synergistic effects of subsistence hunting and
habitat fragmentation on Amazonian forest vertebrates.
Conserv Biol 15:1490–1505.
Pulliam HR. 1988. Sources, sinks, and population regulation.
Am Nat 132:652–661.
Rodrı́guez-Luna E, Cortés-Ortiz L, Miller PH, Ellis S. 1996.
Population and habitat viability assessment for the mantled
howler monkey (Alouatta palliata mexicana). Apple Valley:
IUCN/SSC Conservation Breeding Specialist Group.
Rodrı́guez-Toledo EM, Mandujano S, Garcı́a-Orduña F. 2003.
Relationships between characteristics of forest patches and
howler monkeys (Alouatta palliata mexicana) in southern

Veracruz, Mexico. In: Marsh LK, editor. Primates in
patches: ecology and conservation. New York: Kluwer
Academic/Plenum Press. p 79–97.
Shmida A, Wilson MV. 1985. Biological determinants of
species diversity. J Biogeogr 12:1–20.
Soto A. 2004. El clima. In: Guevara S, Laborde J, Sánchez-
Rı́os G, editors. Los Tuxtlas. El paisaje de la sierra.
Xalapa: Instituto de Ecologı́a A.C., and Unión Europea.
p 195–198.
Swart J, Lawes MJ. 1996. The effect of habitat patch
connectivity on the samango monkey (Cercopithecus mitis)
metapopulation persistence. Ecol Model 93:57–74.
Patch Occupancy by Howler at Los Tuxtlas / 77
Turner MG, Gardner RH, Dale HH, Merriam G, O’Neil RV.
1989. Predicting the spread of disturbance across hetero-
geneous landscapes. Oikos 55:121–129.
Venier LA, Fahrig L. 1996. Habitat availability causes the species
abundance-distribution relationship. Oikos 76:564–570.
Wahungu GM, Muoria PK, Moinde NN, Oguge NO, Kirathe
JN. 2005. Changes in forest fragment sizes and primate
population trends along the River Tana floodplain, Kenya.
Afr J Ecol 43:81–90.
Wieczkowski J. 2004. Ecological correlates of abundance in the
Tana Mangabey (Cercocebus galeritus). Am J Primatol
63:125–138.
Am. J. Primatol. DOI 10.1002/ajp