Table 2. Method of calculation, description, and variance inflation factor (VIF) scores for all fragment (explanatory) variables.

Fragment variables	Method of calculation	Description	VIF score
Area (ha)	Area tool in Google Earth©	Fragment size may positively correlate with fragment quality: fragmentation alters tree species composition in fragments, and a greater proportion of smaller fragments is affected by edge effects	2.3
Shape index	Shape index (SI) = fragment perimeter/(√(fragment area) × π)	SI = 1 indicates a perfect circle; the higher the index, the more irregular the shape (Arroyo-Rodriguez et al., 2013b). SI negatively correlates with habitat quality for edge-averse species (Ewers & Didham, 2007), but complex shapes may contribute to connectivity (Ewers & Didham, 2006)	1.9
Stem density	Stem density = (number of trees sampled of DBH ≥ 10 cm)/(area sampled in transects (1,000 m2))	In mature, primary forest, lower stem density is expected, because there more large, mature trees and fewer trees per given area. In secondary/new growth forest, in forest with more disturbance, or forest with more new edge habitat, stem density would be higher (Wright, 2005; Santos et al., 2008)	3.4
Minimum tree DBH	Smallest diameter of tree (DBH ≥ 10 cm) sampled in fragment transects	Indicative of age/maturity of the forest or successional stage	2.3
Maximum tree DBH	Largest diameter of tree (of DBH ≥ 10 cm) sampled in fragment transects	Indicative of age/maturity of the forest	3.0
Minimum tree height	Smallest height of tree (DBH ≥ 10 cm) found in fragment transects	Indicative of age/maturity of the forest or successional stage. Alouatta spp. are also known to prefer tall trees (Gonzalez-Kirchner, 1998)	2.8
Maximum tree height	Largest height of tree (DBH ≥ 10 cm) found in fragment transects		2.1
Number of stumps	Number of stumps (DBH ≥ 10 cm) sampled in transects	Indicative of natural or human disturbance (e.g., tree mortality/logging; McClennan & Plumptre, 2012)	3.2
Mean DBH of stumps	Mean DBH of stumps (DBH ≥ 10 cm) sampled in fragment transects	Indicates whether dead trees in the fragment were mature, large trees, or young, small trees	1.7
Shannon Index H′ (genera)	${\mathrm{H}}^{\mathrm{\prime }}=\sum _{i=1}^S-(\mathrm{P}\mathrm{i}\ast \mathrm{I}\mathrm{n}\mathrm{P}\mathrm{i})$ Pi = fraction of population made up of type i; S = total numbers of types	A quantitative measure that reflects how many different genera there are in the dataset, and simultaneously considers how evenly individual trees are distributed among those types. A higher H′ indicates a highly diverse (and evenly distributed) dataset (Morris et al., 2014)	5.5
Simpson’s Evenness E (genera)	E = (1/ΣPi2)/S Pi = the proportion of individuals belonging to type i; S = total number of types	Indicative of the degree to which a fragment is dominated by a single or a few genera, with high values indicating that relatively equal numbers of individual trees belong to each genus (Morris et al., 2014)	4.2
Distance to the nearest fragment	Distance (m) to nearest fragment >1 ha in size, regardless of fragment occupancy. Measured with the distance tool in Google Earth©	Measures how close the monkeys in fragment X are to another patch of forest.	2.6
Distance to PNP	Distance (m) to edge of core area of PNP. Measured with the distance tool in Google Earth©	Measures how close the monkeys in fragment X are to a potential “source population” in a relatively large, undisturbed forest (Pulliam, 1988)	2.4
Number of tree lines extending from fragment	Count of number of tree lines (e.g., living fences) of any length extending from fragment	Measure of the difficulty for monkeys to leave or enter the fragment, and their ability to minimize time forced to walk on the ground while traversing the matrix	2.4
Proportion of forest cover in the buffer	Proportion of forest within a pre-defined buffer zone around a fragment; buffer radius = dispersal distance = 7 × (diameter of home range sizea) (Bowman et al., 2002)	An area-based isolation metric; this alternative measure of isolation may be more appropriate for arboreal species that are able, but reluctant, to travel terrestrially (Arroyo-Rodriguez & Dias, 2010)	3.9

Note:

^aIn this formula, we used the diameter of the mean home range size of five groups in PNP: Motiepa group (7 ha), Pakal group (10.8 ha), Naha group (16.1 ha), Unites group (8.6 ha), Balam group in 2016 (6.5 ha), Balam group in 2017 (6.7 ha). Mean home range size = 9.3 ha.