Available via license: CC BY 4.0
Content may be subject to copyright.
Environmental and anthropogenic inuences of
movement and foraging in a critically endangered
lemur species, Propithecus tattersalli: implications
for habitat conservation planning
Meredith A Semel ( merak91@vt.edu )
Virginia Tech https://orcid.org/0000-0003-4317-8602
Heather N Abernathy
Virginia Tech: Virginia Polytechnic Institute and State University
Brandon P Semel
Virginia Tech: Virginia Polytechnic Institute and State University
Michael J Cherry
Texas A and M University: Texas A&M University
Tsioriniaina J Ratovoson
Universite d'Antananarivo
Ignacio T Moore
Virginia Tech: Virginia Polytechnic Institute and State University
Research
Keywords: Movement, space use, foraging, resource selection, Brownian bridge modeling, home range,
road avoidance, primates, lemurs
Posted Date: October 25th, 2021
DOI: https://doi.org/10.21203/rs.3.rs-1001185/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
1
Environmental and anthropogenic influences of movement and foraging in a critically
1
endangered lemur species, Propithecus tattersalli: implications for habitat conservation
2
planning
3
Short title: Lemur space use and foraging behavior
4
5
Meredith A. Semel (0000-0003-4317-8602)
6
Department of Biological Sciences, Virginia Tech, Blacksburg, VA 24061 USA
7
Heather N. Abernathy
8
Department of Fish & Wildlife Conservation, Virginia Tech, Blacksburg, VA 24061
9
USA
10
Brandon P. Semel (0000-0003-3286-0382)
11
Department of Fish & Wildlife Conservation, Virginia Tech, Blacksburg, VA 24061
12
USA
13
Michael J. Cherry
14
Caesar Kleberg Wildlife Research Institute, Texas A&M University-Kingsville,
15
Kingsville, TX 78363 USA
16
Tsioriniaina J.C. Ratovoson
17
Department of Zoology and Animal Biodiversity, University of Antananarivo,
18
Antananarivo, Madagascar
19
Ignacio T. Moore
20
Department of Biological Sciences, Virginia Tech, Blacksburg, VA 24061 USA
21
Corresponding Author: Meredith Semel merak91@vt.edu (847) 636-0788
22
2
Abstract
23
24
Background
25
Wildlife conservation often focuses on establishing protected areas, however, these
26
conservation zones are frequently developed without adequate knowledge of the
27
movement patterns of the species they are designed to protect. Understanding movement
28
and foraging patterns of species in dynamic and diverse habitats can allow managers to
29
develop more effective conservation plans. Threatened lemurs in Madagascar are an
30
example where management plans and protected areas are typically created to encompass
31
large, remaining forests rather than the resource needs of the target species.
32
33
Methods
34
To gain an understanding of golden-crowned sifaka (Propithecus tattersalli) movement
35
patterns, including space use and habitat selection, across their range of inhabited forest
36
types, we combined behavior data with Dynamic Brownian Bridge Movement Models
37
and Resource Selection Functions. We also examined the influence of abiotic, biotic, and
38
anthropogenic factors on home range size, movement rates, and foraging patterns.
39
40
Results
41
We found that home range size and movement rates differed between seasons, with
42
increased core area size and movement in the rainy season. Forest type also played a role
43
in foraging behavior with lemur groups in humid forest avoiding roads in both seasons,
44
3
groups in the dry deciduous forest avoiding road networks in the rainy season, and groups
45
in the moderate evergreen forest displaying no selection or avoidance of road networks
46
while foraging.
47
48
Conclusion
49
Our study illustrates the importance of studying primate groups across seasons as well as
50
across forest types, as developing conservation plans as a single snapshot can give an
51
inaccurate assessment of their natural behavior and resources needs. More specifically,
52
by understanding how forest type influences golden-crowned sifaka movement and
53
foraging behavior, we can make conservation management plans specific to the
54
individual forest types they inhabit (humid, moderate evergreen, dry deciduous, littoral,
55
etc.), rather than the region as a whole.
56
57
Key Words
58
Movement; space use; foraging; resource selection; Brownian bridge modeling; home
59
range; road avoidance; primates; lemurs
60
61
4
Background
62
63
Conservation biologists have long recognized the importance of establishing protected
64
areas to facilitate population persistence in landscapes that are threatened by increasing
65
human encroachment, habitat fragmentation, and habitat loss [1–4]. However, efforts to
66
conserve wildlife and preserve biodiversity are often based on an incomplete
67
understanding of animal movement as well as variability in movement patterns among
68
groups or populations that the areas are meant to protect [5]. While a number of studies
69
have demonstrated the relevance of incorporating movement, particularly animal foraging
70
and home range size, into protected area design [6–9], integration between the disciplines
71
of conservation biology and movement (coined “conservation behavior”) is limited [10–
72
12]. Yet, knowledge of movement behavior, specifically how, when, and where animals
73
move and forage within their habitat, would illuminate how populations navigate and
74
utilize resources within their environment and thus develop better management plans
75
[13,14]. Specifically, species, populations, or even groups often respond differently to
76
factors such as seasonality, habitat characteristics, and anthropogenic pressures in
77
different ways and therefore a better understanding of their role is crucial when
78
developing management plans and establishing protected areas.
79
80
In the tropics, seasons are often divided into dry and rainy seasons, with primary
81
productivity varying seasonally as a function of rainfall. This seasonality thus influences
82
the distribution and availability of resources on the landscape and as a result animal
83
5
movement strategies shift to increase foraging efficiency [15–17]. For example, the
84
black-fronted titi monkey (Callicebus nigrifons; Nagy-Reis and Setz, 2017) and collared
85
brown lemur (Eulemur collaris; Campera et al., 2014) cope with dry season food
86
shortages by reducing movement rates, while the common bumble bee (Bombus
87
vosnesenskii; Pope and Jha, 2018) and African elephant (Loxodonta africana; Wato et al.,
88
2018), respond by increasing foraging and movement rates. Animals can also cope with
89
dry season conditions by shifting home range size or location [22] and altering time spent
90
foraging [23]. Understanding how seasonal fluctuations influence movement and
91
foraging patterns in free-living animals can allow managers to more effectively design
92
protected areas and protect critical resources [24].
93
94
In addition to abiotic factors, biotic factors such as habitat (forest) type, strongly
95
influence animal movement and foraging [25]. Various studies demonstrate that animals
96
adjust their home range size and foraging patterns in response to habitat type and
97
structure (e.g. Roe deer (Capreolus capreolus); Said and Servanty, 2005) and coyote
98
(Canis latrans); Holzman et al., 1992)) indicating that landscape heterogeneity is a key
99
factor influencing the movement of species. While studies of canids, ungulates, and
100
primates have examined the influence of habitat type on home range size, a large
101
proportion of studies are limited to examining metrics of habitat structure (e.g. forest
102
maturity, vegetation density, food scarcity, microhabitat preference) on animal movement
103
and home range size [28,29]. The benefit of understanding movement behavior across
104
6
distinct habitat types is that management strategies can be designed for each habitat type
105
a species occupies.
106
107
Importantly, anthropogenic influences affect animal movement behaviors, can have
108
deleterious effects on wildlife, and must be considered when establishing protected areas
109
[30,31]. The presence of humans and road networks may negatively influence animal
110
movement behavior by increasing human-wildlife interactions (e.g., hunting, poaching,
111
vehicle collisions) and pushing animals out of prime habitat [32,33]. Large mammals are
112
especially affected by human encroachment due to their larger home range size, lower
113
population density, more narrow geographic distributions, and large portions of their
114
distributions being shared with humans [34]. For instance, black bears (Ursus
115
americanus) have been found to avoid areas with human development during daylight
116
hours [35] and woodland caribou (Rangifer tarandus caribou) avoid high use roads,
117
mines, and cabins during months of high human activity [36]. Of large mammalian taxa,
118
few studies have examined the influence of human infrastructure on primate movement,
119
although they are often strongly affected by anthropogenic features [37,38].
120
121
The lemurs of Madagascar face significant anthropogenic threats [39]. Between 1953 and
122
2014, Madagascar lost 44% of its forests, with 46% of the remaining forests being
123
located within 100 m of a forest edge [40]. This high degree of forest destruction and
124
increasing presence of edge forest habitat influences lemur behavior and their ability to
125
meet their nutritional demands. While previous studies have examined lemur home range
126
7
size [41,42], dietary flexibility [43], species abundance [44], and reproduction in various
127
environments [45,46], our understanding of lemur movement is limited but members of
128
the genus Propithecus have provided some information. In regard to home range
129
characteristics, diademed sifakas (Propithecus diadema) in humid fragmented habitats
130
had reduced home range size and daily path length and foraged on sub-optimal food
131
items compared to sifaka groups in contiguous forest environments [41]. In contrast,
132
Milne-Edwards’ sifakas (Propithecus edwardsi) inhabiting humid logged forests traveled
133
shorter distances each day to feed in a low-quality food environment, yet maintained
134
larger home ranges than conspecifics in contiguous forests [42]. Further, Verreaux’s
135
sifakas (Propithecus verreauxi) in Madagascar’s dry deciduous forests exhibited
136
significant home range reduction from the rainy to the dry season [47]. While these
137
studies have shed light on Propithecus behavioral responses to abiotic and biotic factors
138
in extremes of the humid-dry forest gradient of forest types, we do not understand how
139
species in the genus Propithecus respond in a moderate forest type. Knowledge of
140
Propithecus movement behavior in regards to these factors would enable us to predict
141
how these lemurs would use a protected area and design a reserve accordingly.
142
143
Golden-crowned sifaka (Propithecus tattersalli) are a critically endangered lemur
144
endemic to naturally fragmented forests of northeastern Madagascar [48]. Unlike the
145
other eight species of sifaka (Propithecus spp.) on Madagascar that are restricted to dry
146
or humid forest types, P. tattersalli inhabit a range of forest types [49,50]. Variation of
147
habitable forest types makes them a unique opportunity to examine the influence of
148
8
seasonality, forest type, and anthropogenic factors on movement and foraging behavior in
149
a primate. Studies of golden-crowned sifaka have documented a major decline in the
150
population in the last decade and informed researchers of the natural fragmentation of the
151
landscape, yet no previous study has examined the influence of movement on space use
152
and foraging tree selection across their range (Quéméré et al., 2012; Salmona et al.,
153
2017). An understanding of how abiotic, biotic, and anthropogenic factors influence
154
golden-crowned sifaka space use and foraging throughout their range would allow
155
species management plans to be made for populations within each particular forest type
156
occupied rather than the species as a whole.
157
158
In this study, we analyzed location and foraging behavior of six golden-crowned sifaka
159
groups to evaluate the effects of abiotic (seasonality; rainy and dry season), biotic (forest
160
type; humid, moderate evergreen, and dry deciduous forests), and anthropogenic
161
(fragmentation; edge and interior forests) factors on their movement patterns and space
162
use. Approaches to studying nonhuman primate space use typically are limited to
163
examining daily path length and home range overlap through the use of area estimators
164
(MCP, line-based kernel density, etc.; (e.g. Lehmann and Boesch, 2002; Steiniger and
165
Hunter, 2013). More modern and sophisticated approaches including Dynamic Brownian
166
Bridge movement models (DBBMM) and Bayesian methods [55,56], reduce the
167
likelihood of both Type I and Type II errors which can bias our understanding of animal
168
space use and habitat selection [57]. Thus, to estimate space use we used DBBMM which
169
incorporated temporal and behavioral characteristics of movement trajectories into
170
9
estimation of an animal’s home range [58]. For our first objective we predicted that
171
seasonal movement rates would be greater in the rainy season, humid forests, and edge
172
forests compared to drier more interior forests. Second, we predicted that home range
173
size and core area range size would be larger in the rainy season, in edge forests, and in
174
the humid forest compared to drier and interior forests. We also predicted that sifaka
175
groups in edge forests would exhibit less core area range overlap (between the rainy and
176
dry seasons) than interior forest groups. Third, we predicted that sifakas would select the
177
largest feeding trees within their home ranges and avoid locations near human settlements
178
or manmade structures.
179
180
Methods
181
182
Study Area
183
184
Research was conducted in the Loky-Manambato Protected Area of northeastern
185
Madagascar (Fig. 1). This protected area encompasses a unique biogeographical
186
transition zone from Madagascar’s northern and western dry deciduous forests to
187
southern humid forests. The Loky-Manambato region contains a mosaic of various forest
188
types including dry deciduous, dry evergreen, humid, and littoral forests separated by
189
agricultural areas and savanna (Quéméré et al., 2012). The region experiences a four-
190
month rainy season occurring from December to March followed by an eight-month dry
191
10
season [59]. The study sites include three distinct forest types: a humid forest, moderate
192
evergreen forest, and dry deciduous forest.
193
194
Study Species and Subjects
195
196
Golden-crowned sifaka live in semi-cohesive social groups ranging in size from 3-12
197
individuals with one or more adult males, several adult females, and several immature
198
individuals of both sexes. Group members typically travel in a coordinated fashion and
199
generally remain in visual or auditory contact with at least one other group member [59].
200
Thus, we assumed that all animals within a given social group share a home range, and
201
therefore treated each group as a unit of analysis in this study. Golden-crowned sifaka are
202
frugo-folivores, but also consume seeds, petioles, buds, flowers, and bark.
203
204
We studied six groups of habituated golden-crowned sifaka distributed across the three
205
distinct forest types (two groups each in humid, moderate evergreen, and dry deciduous
206
forests) in the Loky-Manambato Protected Area. We selected three of the 11 large forest
207
fragments containing golden-crowned sifaka due to their accessibility: Binara (humid),
208
Bekaraoka (moderate evergreen), and Solanamampilana (dry deciduous) (Figure 1).
209
Within each forest type, we followed one group in primary forest towards the center of
210
the forest (hereafter interior; characterized by lemurs having a home range at least 300
211
meters from the forest edge) and one group on the edge of the forest fragment (hereafter
212
11
edge; characterized by having a home range adjacent to the forest edge). Average group
213
size was six individuals and ranged from five to eight (Table 1).
214
215
Fig. 1. Study area in the Loky-Manambato Protected Area of northern Madagascar. The
216
thin red line depicts the unpaved national road in the region. The three forest fragments
217
surveyed are denoted by hatched black lines and are colored based on forest type.
218
219
220
221
222
12
223
Table 1. Composition of focal groups within each forest fragment, forest type, and forest
224
fragmentation classification.
225
Forest Fragment
Forest Type
Forest Location
Group Size
Binara
Humid
Interior
7
Binara
Humid
Edge
5
Bekaraoka
Moderate Evergreen
Interior
7
Bekaraoka
Moderate Evergreen
Edge
8
Solanamampilana
Dry Deciduous
Interior
5
Solanamampilana
Dry Deciduous
Edge
5
226
227
Group Location Data
228
229
We collected golden-crowned sifaka group location data during two periods, February-
230
April 2019 (rainy season) and June-August 2019 (dry season). We followed groups from
231
sleep tree to sleep tree (~13 hours per day) and collected location data at 15-minute
232
intervals. In addition to daytime activity, golden-crowned sifaka are known to exhibit
233
nocturnal movements, specifically during periods of bright moon light [60], and thus
234
groups were not always located in the same sleep tree the following morning. In these
235
instances, we reestablished contact with the group as quickly as possible. Group locations
236
were recorded using a GPS receiver (Garmin 64s), using the Universal Transverse
237
13
Mercator coordinate system (zone 39L), and points were logged at the group’s
238
approximate geometric center. If no animals were visible at the 15-minute interval,
239
observers waited to establish visual contact with the social group before recording any
240
locations.
241
242
Foraging and Landscape Data
243
244
We recorded foraging data at the same 15-minute intervals using scan sampling to record
245
the behavior, height in the tree, and nearest neighbor of each individual in a group [61]. If
246
an individual was actively feeding during the scan, the plant species and part (e.g.,
247
young/mature leaf, leaf petiole, un/ripe fruit, seed, or flower) were identified, GPS
248
location recorded, and data concerning tree species, size, and current phenology
249
collected. In addition to collecting foraging data specific to each of the lemur groups, we
250
also collected general landscape data throughout each of the six lemur home ranges in
251
both the rainy and dry season. We did this by randomly generating forty GPS points
252
within each of the six home ranges (in both seasons) and collected data from potential
253
feeding trees (species, size, phenology) within 5 meters of each location. This allowed us
254
to gain an understanding of the entire landscape of all six home ranges, not just the
255
specific feeding trees utilized by each of the groups.
256
257
Home Range Estimation
258
259
14
Utilization distributions (i.e., 95% isopleth, hereafter home ranges and 50% isopleth,
260
hereafter core area) were estimated for each golden-crowned sifaka group using Dynamic
261
Brownian Bridge Movement Models (DBBMM); [58]. Home range DBBMMs use
262
behavior and movement trajectory data of the animal group that is collected in sequential
263
relocation studies. This method provides a spatially explicit model, which describes the
264
probability of the given animal group occurring in a given location during a specified
265
period. This approach also accounts for temporal autocorrelation, spatial uncertainty,
266
irregularly sampled data, and shifts in an animal’s behavior (resting, foraging,
267
thermoregulating, corridor use, etc.), making it specifically applicable to studies of group
268
living primates [57,58,62]. Using DBBMMs to estimate group home ranges requires a
269
Brownian motion variance parameter (σ2, in meters), which quantifies the degree of
270
diffusion or irregularity of an animal’s path [58]. A moving window analysis identifies
271
changes in the movement behavior and estimates σ2 for each step. The size of the moving
272
window must include an odd number of GPS locations, because the σ2 parameter is
273
estimated using a “leave-one-method”, and a margin of greater than three locations
274
bounding each end of the window in which no behavioral changes can occur [58]. We
275
parameterized the DBBMM with a 21-step window size, a 9-step margin size, and a 15 m
276
location error for all lemur groups, as visual inspection indicated these settings were
277
sufficient to identify changes in home range size and overall animal movement [58].
278
Home ranges were estimated for each lemur group using the DBBMM function in R
279
package ‘move’ [63,64]. We conducted a three-way analysis of variance (ANOVA)
280
predicting for both 50% (core area) and 95% home ranges, respectively, to determine if
281
15
season, forest type, interior or edge forests, and the interaction of forest type and season
282
influenced core area and home range size. All analyses were conducted in version 3.5.1
283
of program R [65]. We used Akaike’s Information Criterion (AICc) to identify a top
284
model from the set of candidate ANOVA models [66].
285
286
Core Area Overlap
287
288
To determine the percent of joint home range overlap between the rainy and dry season
289
home ranges, we calculated the total area of each home range and then divided the area of
290
overlap between seasons by the total home range size [67]. Possible core area overlap
291
ranged from 0% overlap, indicating no shared space use between seasons, to 100%
292
overlap, indicating the dry and rainy season ranges overlapped completely. To determine
293
if the home range overlap between the rainy and dry seasons varied as a function of
294
fragmentation types (edge or interior), we conducted an unpaired two-sample t-test
295
assuming equal variances. We tested the variance assumption of our t-test using an F-test,
296
‘var.test’, in program R. Finally, to determine if the core area overlap between the rainy
297
and dry seasons varied as a function of forest type (humid, moderate, or dry), we
298
conducted a one-way ANOVA comparing core area and 95% home range and core area
299
as a function of degree of forest type. All analyses were conducted in version 3.6.1 of
300
program R [65].
301
302
Movement Rates
303
16
304
We calculated movement rates (meters/hour) for each lemur group using the collected
305
relocation data. The step length (i.e., the distance between sequential locations) was
306
divided by the time elapsed between each sequential location to calculate speed for each
307
golden-crowned sifaka group to characterize movement rates. To determine how
308
movement varied across season (rainy and dry), forest fragmentation (edge or interior),
309
and forest types (humid, moderate evergreen, and dry deciduous) we calculated
310
movement rates at both the daily and seasonal scale.
311
312
Daily movement rates were bootstrapped to calculate a mean for each observational day.
313
Bootstrapping is a process that involves repeatedly drawing independent samples from a
314
data set (x) to create bootstrap data sets (x1, x2,…., xn). Our samples were performed with
315
replacement which allowed for the same observation to be sampled more than once such
316
that each bootstrapped sample was the same length as our raw lemur speed data (m/hour).
317
To calculate seasonal movement rates (
!"#
$), we drew 1000 independent samples
318
(
%
&
'(%
&
)(*(%
&
+,-
to calculate means and standard error (
!.
/
+
), which we then used to
319
generate 95% confidence intervals for comparison of means among seasons and groups,
320
!.
/
+
=012
3
4
56-3
4
7
8
9
7
:;<=+6',
321
where
!.
/
+-
served as our estimate of the standard error of
%
& estimated from the raw lemur
322
speed data (m/hour). We calculated seasonal movements rates using the bootstrapping
323
approach outlined above but employed the method for each observation season.
324
17
325
To determine how environmental variables influenced daily movement rates, we fit linear
326
mixed effects models to predict movement rate as a function of all combinations of
327
season (rainy or dry), forest fragmentation (edge or interior), and forest type (humid,
328
moderate evergreen, and dry deciduous), while treating forest type-forest fragmentation
329
per group intercepts as random effects [68]. We used the Satterthwaite method to
330
approximate the degrees of freedom and computed p-values for direct effects and
331
interactions using t-statistics.
332
333
Finally, to determine how environmental variables influenced seasonal movement rates,
334
we conducted a three-way ANOVA seasonal movement rates as a function of forest type
335
(dry, humid, and wet), forest fragmentation (edge and interior), and season (dry and
336
rainy). All analyses were conducted in version 3.6.1 of program R [65]. We used
337
Akaike's Information Criterion corrected for small sample size (AICc) to identify a top
338
model from the set of candidate models [69]. All analyses were conducted in version
339
3.6.1 of program R [65,68].
340
341
Habitat Selection
342
343
To quantify habitat selection of golden-crowned sifaka groups, in relation to tree size and
344
proximity to anthropogenic factors, we fit a Resource Selection Function (RSF) using a
345
use-available design. A RSF is defined as any function producing a value proportional to
346
18
the probability of selection of a given habitat [70,71]. Any estimate derived from an RSF
347
is dependent on the definition of available habitats [55,70,72]. For our RSF, selection by
348
golden-crowned sifaka availability was considered within home range selection
349
(Johnson’s third order; Johnson 1980) as defined by a 95% seasonal home range (i.e.,
350
95% isopleth) using DBBMMs. Within our seasonal home ranges, we characterized
351
availability by systematically identifying available locations at intervals of 10 m, as this
352
was the spatial resolution of all spatial data used in the RSF [73].
353
354
We created our RSFs by fitting generalized linear mixed-effects model (GLMM) with a
355
binomial link function, which included a group-specific (forest type and fragmentation
356
type) random intercept term to account for non-independence of habitat associations
357
within groups [74,75]. For our RSF, we used GPS locations of all feeding trees that
358
golden-crowned sifaka utilized during the rainy and dry field seasons and possible
359
locations within their known home ranges. We extracted tree basal area (cross-sectional
360
area of trees at breast height), Euclidian distance to village, road, and habitat fragment
361
edge for each golden-crowned sifaka feeding tree and each available location. These data
362
were generated using satellite imagery and habitat sampling of resources within each of
363
the six lemur home ranges.
364
365
To relate tree basal area and crown volume to lemur GPS location data, we created
366
continuous surfaces of tree basal area and crown volume estimates across our study area
367
by using inverse distance weighting (IDW) interpolation in the package gstat [76] in
368
19
version 3.6.1 of program R [65]. IDW uses a weighted average of estimates from nearby
369
sampling locations to predict tree basal area and crown volume estimates to the
370
surrounding pixels of a sampling location composed of user-specified areas [77]. Our
371
user-specified areas of inference were 169 m2 because it most closely matched the mean
372
distance between vegetation sampling locations (148.85 m). This interpolation process
373
provided spatially explicit estimates of tree basal area and crown volume estimates which
374
we could then associate with our lemur GPS data.
375
376
To examine if lemur habitat selection varied across forest types and seasons, we
377
developed candidate models using various combinations of distance to habitat feature
378
(i.e., village, road, and habitat fragment), and basal area, and used Akaike's Information
379
Criterion (AICc) to identify a top model from the set of candidate models [66] to
380
determine: 1) if differences in habitat selection vary as a function of forest type, and 2) if
381
differences in habitat selection vary as a function of season at each site. To account for
382
behavioral differences in lemur groups, we accounted for random effects using an
383
‘animal id’ that consisted of each lemur group’s respective forest type (humid, dry, or
384
moderate), forest fragmentation classification (edge or interior), and season (rainy or
385
dry). No environmental variables used in model development exhibited high correlation
386
(i.e., |r| > 0.7). All coefficients were estimated using the “lme4” package for R 3.0.1
387
[65,68].
388
389
390
20
Results
391
392
Home range size estimation
393
394
Overall, home range sizes (95% utilization distribution) for golden-crowned sifaka
395
groups in the Loky-Manambato Protected Area were highly variable and ranged from
396
2.78 – 31.56 hectares (Table 2). Our top ANOVA model (Table 3), revealed that golden-
397
crowned sifaka core areas (50% home range) varied with season (p = 0.003, F=14.65,
398
df=1, residual df = 10, residual SE = 0.004) with core areas being larger in the rainy
399
season (average of 1.80 hectares in the rainy season, 0.81 hectares in the dry season).
400
However, while our ANOVA model candidate set for home ranges (95% home range) did
401
include season as a top model, suggesting a trend towards increased home range size in
402
the rainy season (dry or rainy; p = 0.08, F=3.671, df=1, residual df = 10, SE = 0.08), it
403
was no better than our null model when considering a delta AIC of 2 (Table 4).
404
405
Seasonal Core Area Overlap
406
407
Seasonal overlap of the 50% home range area varied from 17% to 54% overlap (Table 2,
408
Figure 2). Core area overlap between the rainy and dry season did not vary with forest
409
type (core area – p = 0.702, F = 0.393, df = 2, residual df = 3, SE = 0.003; home range –
410
p = 0.394, F = 1.29, df = 2, residual df = 2, residual SE = 0.001) or forest fragmentation
411
21
(core area – p = 0.365, F = 1.04, df = 1, residual df =4, residual SE < 0.001; home range –
412
p = 0.219, F = 2.21. df = 1, residual df = 4, residual SE = 0.02).
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
Table 2. Home range sizes (50% and 95% isopleth (UD: utilization distribution)) of
428
sifaka groups in the rainy and dry season. The three forest fragments differ in forest type
429
(Moderate; Bekaraoka, Humid; Binara, Dry; Sola) and forest fragmentation (interior;
430
edge).
431
432
22
Lemur Group ID
50% UD (ha)
50% UD
(%
overlap)
95% UD (ha)
95% UD
(%
overlap)
Dry
Rainy
Dry
Rainy
Moderate-interior
0.63
1.87
41%
2.93
9.16
55%
Moderate-edge
0.57
0.78
17%
2.78
4.83
75%
Humid-interior
0.68
1.71
29%
3.69
25.1
38%
Humid-edge
1.18
1.90
54%
12.5
11.1
61%
Dry-interior
0.88
1.91
17%
6.04
7.95
59%
Dry-edge
0.89
2.62
51%
8.19
31.6
51%
433
434
435
436
437
438
439
440
Table 3: Models, the number of parameters (K), log-likelihood (LL), the relative
441
difference in AICc values compared to the top-ranked model (ΔAICc), and the AIC
442
model weights (W) of the model-selection procedure examining core area size (50%
443
isopleth) of lemur groups.
444
23
Model
K
LL
AICc
ΔAICc
W
season
1
48.946
-88.9
0
0.828
fragmentation + season
2
48.963
-84.2
4.68
0.08
forest type + season
2
51.882
-83.8
5.13
0.064
null
0
43.533
-81.7
7.16
0.023
fragmentation
1
43.54
-78.1
10.81
0.004
forest type
1
44.558
-75.4
13.49
0.001
forest type + season +
fragmentation
3
51.91
-75
13.87
0.001
forest type + fragmentation
2
44.566
-69.1
19.76
0
445
446
Table 4: Models, the number of parameters (K), log-likelihood (LL), the relative
447
difference in AICc values compared to the top-ranked model (ΔAICc), and the AIC
448
model weights (W) of the model-selection procedure examining home range size (95%
449
isopleth) of lemur groups.
450
Model
K
LL
AICc
ΔAICc
W
season
1
14.266
-19.5
0
0.427
null
0
12.389
-19.4
0.09
0.409
fragmentation
1
12.535
-16.1
3.46
0.076
fragmentation + season
2
14.466
-15.2
4.31
0.049
forest type
1
13.793
-13.9
5.66
0.025
forest type + season
2
16.279
-12.6
6.97
0.013
forest type + fragmentation
2
13.978
-8
11.58
0.001
forest type + fragmentation +season
3
16.562
-4.3
15.21
0
451
24
452
453
25
Fig. 2. Brownian bridge utilization distribution (50% isopleth) depicting core area use for
454
golden-crowned sifaka groups between the rainy (dark gray) and dry (light gray) season.
455
Overlapping areas are occupied during both seasons. The six figures display the seasonal
456
home ranges for all six lemur groups followed. The columns indicate forest
457
fragmentation classification (interior or edge) and the rows indicate the occupied forest
458
type (humid, moderate, or dry) of all six lemur groups.
459
460
Daily and Seasonal Movement Rates
461
462
Our top model (Table 5) indicated that seasonal movement rates varied as a function of
463
season (Sum Squares = 0.405, 95% CI [0.13, 0.65], F=11.27, df=1, p=0.007, residual df =
464
10, residual SE = 0.04), with higher rates in the rainy season (rainy season: 83.47 m/h;
465
dry season: 56.70 m/h). When investigating daily movement rates, our mixed effects
466
linear model analysis supported our seasonal movement results, as the top model
467
included season; however, there was not support for effects of forest type or forest
468
fragmentation (β = 0.36, 95% CI [0.11, 0.60]; p=0.019; Table 6, Figure 3).
469
470
471
472
473
474
475
26
Table 5: Competing models, the log-likelihood (LL), the number of parameters in the
476
model (K), the relative difference in AICc values compared to the top-ranked model
477
(ΔAICc), and the AIC model weights (W) of the model-selection procedure examining
478
seasonal movement rates of lemur groups.
479
Model
LL
AICc
ΔAICc
W
season
3.382
2.2
0
0.733
fragmentation + season
4.394
4.9
2.69
0.191
null
-1.145
7.6
5.39
0.05
fragmentation
-0.691
10.4
8.15
0.012
forest type + season
4.779
10.4
8.21
0.012
forest type
-0.529
14.8
12.54
0.001
forest type + season + fragmentation
6.086
16.6
14.39
0.001
forest type + fragmentation
-0.023
20
17.81
0
forest type + season + forest type*season
6.158
29.7
27.45
0
forest type + fragmentation + season + forest
type*season
7.856
48.3
46.05
0
480
Table 6. Competing models, the log-likelihood (LL), the number of parameters in the
481
model (K), the relative difference in AICc values compared to the top-ranked model
482
(ΔAICc), and the AIC model weights (W) of the model-selection procedure examining
483
daily movement rates of lemur groups.
484
Model
K
LL
AICc
ΔAICc
W
season
1
-27.381
63.3
0.00
0.586
null
0
-29.393
65.1
1.82
0.236
fragmentation + season
2
-28.186
67.1
3.87
0.085
fragmentation
1
-30.082
68.7
5.40
0.039
forest type + season
2
-28.022
69.1
5.87
0.031
forest type
1
-30.101
71.0
7.70
0.012
forest type + season + fragmentation
3
-28.845
73.2
9.89
0.004
forest type + season + forest
type*season
4
-27.815
73.6
10.28
0.003
forest type + fragmentation
2
-30.777
74.6
11.38
0.002
Global
5
-28.634
77.7
14.43
0.000
27
485
486
487
Fig. 3. Seasonal movement rates (meters/hour) of golden-crowned sifaka groups using
488
relocation data collected every 15 minutes. The step length (e.g., the distance between
489
sequential locations) was divided by the time elapsed between each step to calculate
490
speed for each lemur group. Data were collected during the dry season (June-August
491
2019) and the rainy season (February-April 2019). Black lines correspond to 95%
492
confidence intervals.
493
28
494
Habitat Selection
495
496
We found that lemur habitat selection varied with forest type (Table 7) and within forest
497
type, selection varied by season (Table 8, Tables S1 - S3). Thus, to make inferences
498
about differences in selection across forest type and season, we parsed the data into forest
499
type and then again into season.
500
501
We found that lemur groups in the humid forests selected feeding trees characterized with
502
greater crown volume (β = −5.0, s.e. ± 1.23, p<0.001), closer to villages (β = -0.76, s.e. ±
503
0.30, p = 0.01), and closer to the forest edge (β = −1.04, s.e. ± 0.17, p<0.001), and
504
avoided locations near roads (β = 2.79, s.e. ± 0.93, p = 0.003) in the dry season. In the
505
rainy season groups in humid forests selected locations with greater tree basal area (β =
506
0.085, s.e. ± 0.04, p = 0.049) and greater crown volume (β = 1.34, s.e. ± 0.06, p<0.001)
507
that were closer to villages (β = -1.28, s.e. ± 0.19, p<0.001), and forest edges (β = −0.24,
508
s.e. ± 0.05, p<0.001), and avoided habitats near roads (β = 3.26, s.e. ± 0.73, p<0.001).
509
While the effects of crown volume, villages, forest edges and roads were the same across
510
seasons, these effects were stronger in the rainy season (Table 8; Figure 4).
511
512
Lemur groups in moderate evergreen forests selected locations with greater crown
513
volume (β = 0.52, s.e. ± 0.07, p<0.001), greater tree basal area (β = 0.35, s.e. ± 0.17, p =
514
0.03), and locations farther from villages (β = 1.30, s.e. ± 0.32, p<0.001) in the dry
515
29
season. Selection of locations closer to the forest edge approached significance (β =
516
−1.25, s.e. ± 0.68, p = 0.065). In the rainy season, troops selected feeding trees with
517
greater tree crown volume (β = 1.24, s.e. ± 0.07, p<0.001) and greater tree basal area (β =
518
0.55, s.e. ± 0.19, p = 0.003) and avoided habitats closer to villages (β = 1.91, s.e. ± 0.38,
519
p<0.001). Avoidance of the forest edge approached significance (β = −2.76, s.e. ± 1.5, p
520
= 0.069; Table 8; Figure 4).
521
522
Finally, we found that groups in dry deciduous forests selected greater crown volume (β
523
= 1.22, s.e. ± 0.45, p<0.001). During the rainy season, groups in dry forest selected
524
locations with greater crown volume (β = 1.04, s.e. ± 0.12, p<0.001) and greater tree
525
basal area (β = 2.89, s.e. ± 0.73, p<0.001), and avoided habitat closer to villages (β =
526
3.06, s.e. ± 0.74, p<0.001) and roads (β = 2.42, s.e. ± 0.50, p< 0.001; Table 8; Figure 4).
527
528
529
530
531
532
533
534
535
536
30
Table 7. Competitive models AIC table depicting differences in foraging tree selection
537
among lemurs occupying different forest types. Competitive models, the number of
538
parameters (K), the relative difference in AIC values compared to the top ranked model
539
(ΔAIC), the AIC weights (W), and the log-likelihood (LL) of the model-selection
540
procedure examining foraging tree selection of lemurs based on occupied forest type
541
(humid, moderate, and dry). CV: Crown Volume, TBA: Tree basal area, V: Distance to
542
village, R: Distance to roads, F: Distance to forest edge, FT: Forest type. Based on the
543
models, forest types could not be grouped and were parsed to make assumptions.
544
K
AIC
ΔAIC
W
LL
(CV+ TBA + V + R + F)*FT
19
11353.62
0
1
-5657.81
CV+ TBA + V + R + F
7
11425.18
71.56
0
-5705.59
545
31
Table 8. Competitive models AIC table of lemur foraging tree selection based on occupied forest type. Competitive models, the
546
number of parameters (K), the relative difference in AIC values compared to the top ranked model (ΔAIC), the AIC weights (W), and
547
the log-likelihood (LL) of the model-selection procedure examining foraging tree selection of lemurs based on occupied forest type
548
(humid, moderate, and dry). CV: Crown Volume, TBA: Tree basal area, V: Distance to village, R: Distance to roads, F: Distance to
549
forest edge, FT: Forest type.
550
Humid Forest
K
AIC
ΔAIC
W
LL
(CV+ TBA + V + R + F)*Season
13
4158.31
0
1
-2066.15
CV+ TBA + V + R + F
7
4206.98
48.68
0
-2096.49
Moderate Forest
K
AIC
ΔAIC
W
LL
(CV+ TBA + V + R + F)*Season
13
2944.24
0
1
-1459.12
CV+ TBA + V + R + F
7
3043.27
99.04
0
-1514.64
Dry Forest
K
AIC
ΔAIC
W
LL
(CV+ TBA + V + R + F)*Season
13
4085.42
0
1
-2029.71
CV+ TBA + V + R + F
7
4105.11
19.7
0
-2045.56
551
552
553
554
32
555
Fig. 4. Selection coefficient plot for golden-crowned sifakas in the dry and rainy season within the three forest types. This coefficient
556
plot displays beta estimates for tree basal area and tree crown volume and distance to habitat fragment, roads, and villages. Blue points
557
represent habitat selection during the rainy season and red points represent habitat selection during the dry season. Solid lines above
558
and below each point represent the 95% confidence intervals around each beta estimate.
559
33
Discussion
560
561
Our study shows three primary results. First, golden-crowned sifaka movement rates are greater
562
in the rainy season and in the humid forest type. Second, variation in climatic conditions (rainy
563
vs. dry season) influences lemur movement, with core area range sizes being larger in the rainy
564
season. Third, human disturbance influences lemur spatial ecology with lemurs preferentially
565
selecting foraging locations where larger trees are present. We also detected variation in
566
behavioral responses to villages, road networks, and the forest edge. Lemurs in humid and dry
567
deciduous forest fragments specifically avoided locations near road networks in both the dry and
568
rainy seasons, while lemurs in the moderate evergreen forest did not select or avoid locations
569
near road networks. In sum, groups of golden-crowned sifaka show marked variation in
570
behavioral responses to human disturbance, but in all groups, higher-use zones consist of
571
locations closer to large trees. Thus, season, forest type, and forest fragmentation all have effects
572
on lemur space use and ranging behavior.
573
574
Using Dynamic Brownian Bridge Movement Models, home range sizes of golden-crowned
575
sifaka groups varied between 3-32 hectares, indicating an incredible amount of home range size
576
variation within the species. These home range sizes are smaller than those of diademed sifaka
577
(Propithecus diadema), a species inhabiting Madagascar’s eastern humid forests, which range
578
from 19-79 ha (95% kernel; Irwin, 2008), but larger than those of Verreaux’s sifaka (Propithecus
579
verreauxi), a species inhabiting Madagascar’s southern dry forests, which have home range sizes
580
ranging from 5-10 ha [78]. Similar to this trend, mouse lemurs (Microcebus spp) inhabiting
581
western dry forests were able to maintain higher population densities than mouse lemur species
582
34
inhabiting eastern humid forests [79]. Contrary to these previous findings and our predictions,
583
we found that golden-crowned sifaka groups in humid forest fragments do not occupy
584
significantly larger home range or core area sizes compared to groups living in moderate
585
evergreen or dry deciduous forest fragments. This finding was unexpected because Malagasy
586
humid evergreen forests are often described as being lower quality habitats compared to drier
587
forests due to decreased food availability, requiring wildlife to occupy larger ranges to meet
588
nutritional demands [80]. In sum, our study is the first determining that the same species of
589
sifaka can inhabit drastically different forest types and display great variation in home range size.
590
591
Our prediction that lemur home range sizes would vary between the rainy season and the dry
592
season was partially supported. While home range sizes (95% isopleth) were not significantly
593
different between seasons, core area range size (50% isopleth) was statistically larger for lemur
594
groups in the rainy season compared to the same groups’ core area range sizes in the dry season.
595
Similar to findings of Milne-Edwards’ sifaka, we found that golden-crowned sifaka maintained
596
similar home range (95% isopleth) locations in both seasons, but displayed considerable seasonal
597
shifts in core area (50% isopleth) locations [42]. This result demonstrates that golden-crowned
598
sifaka have significant site fidelity for their home ranges, but a lower degree of fidelity for core
599
area ranges. This difference is likely due to the non-uniform and seasonal variation in
600
distribution of resources which influenced how golden-crowned sifaka distributed their space use
601
to efficiently forage [81]. Surprisingly, the degree of core area overlap observed did not vary
602
based on the forest type or forest fragmentation level occupied.
603
604
35
Unlike our prediction that sifaka groups in more degraded habitats would occupy larger home
605
ranges, we found limited evidence that forest fragmentation (edge vs. interior) influenced home
606
range or core area size in golden-crowned sifaka. Previous studies have demonstrated varying
607
effects of disturbance on home range size of eastern sifaka species in rainforest habitats. For
608
instance, diademed sifaka (Propithecus diadema) living in edge forests occupied significantly
609
smaller home range sizes than conspecifics in contiguous forests [41] while Milne-Edwards’
610
sifaka (Propithecus edwardsi) in fragmented (logged) forests maintained larger home range sizes
611
[42]. While these studies indicate contrasting effects of forest fragmentation on lemur home
612
range size, fragmentation and other anthropogenic habitat changes are known to produce
613
negative impacts on lemurs [82]. Unfortunately, the majority of lemur studies (87%) examining
614
the influence of anthropogenic habitat changes on lemur health, genetics, biodiversity, and
615
behavior were conducted in the humid forests of eastern Madagascar [82]. Further, lemur
616
responses to habitat edges in dry forest are often highly variable, with groups avoiding, selecting,
617
or demonstrating no response in regards to feeding along forest edges. As a result, further
618
investigation of home ranges of golden-crowned sifaka and other dry forest lemurs are needed to
619
understand how increasing anthropogenic changes are influencing lemur ecology and
620
conservation.
621
622
Across seasons, regardless of forest type or forest fragmentation, golden-crowned sifaka groups
623
daily movement rates shifted, with groups moving farther per unit time in the rainy season.
624
Contrary to this finding in golden-crowned sifaka, previous studies indicate that Milne-Edward’s
625
sifaka do not increase or decrease their distance moved per day (i.e., daily path length) between
626
seasons [42]. Thus, there exists some degree of variability among sifaka species. Movement rates
627
36
in golden-crowned sifaka groups were also closely linked to home range size in that as home
628
range increased in the rainy season, so did the average distance moved per hour. This finding is
629
consistent elsewhere in highly mobile mammals as movement rates and resource availability
630
determined home range size of white-tailed deer (Odocoileus virginianus) and Iberian ibex
631
(Capra pyrenaica) [83,84].
632
633
Aligning with our predictions, sifaka habitat selection was influenced by proximity to human
634
settlements and permanent manmade structures, with golden-crowned sifaka groups
635
preferentially avoiding these areas. Studies examining the influence of road networks on other
636
mammalian species have similarly found that elk (Cervus canadensis) and caribou (Rangifer
637
tarandus) tend to avoid road crossings and seek cover when in close proximity to road networks
638
[85]. Road expansion and paving is known to increase the prevalence of vehicle collisions with
639
wildlife (e.g., Asiatic cheetah (Acinonyx jubatus venaticus; Mohammadi et al., 2018) and Florida
640
panther (Puma concolor coryi; Criffield et al., 2018). The national road that bisects the global
641
range of golden-crowned sifaka is currently being improved to enhance access to mineral
642
reserves and transportation through the region. The combination of road expansion and human
643
population growth is expected to intensify the already harmful degree of resource extraction
644
within the region (e.g., selective logging of hardwoods and gold mining) [39]. Thus, given the
645
current level of avoidance golden-crowned sifaka display towards roads, the continued road
646
paving initiatives within their global range, and evidence gained from other mammalian studies,
647
increased avoidance on golden-crowned sifaka are likely. Our results indicating sifaka avoidance
648
of human villages have been documented in other mammalian taxa; for example, caracal
649
(Caracal caracal) avoided areas visited by and adjacent to human settlements [88]. Lastly, even
650
37
low vehicle traffic (0-30 vehicles/12hrs) can lead to animal avoidance as demonstrated in
651
wolverines (Gulo gulo), where space use in regard to road networks led to avoidance and altered
652
movement patterns [89]. Consequently, increasing human activity and road prevalence has the
653
potential to impact foraging and space use behavior of wildlife species, and in the case of
654
golden-crowned sifaka, could result in an impact large enough to threaten their population.
655
656
Conclusions
657
658
Conservation implications
659
660
Our study illustrates the complex anthropogenic and ecological processes that influence
661
movement behavior of golden-crowned sifaka groups. We found clear evidence that human
662
settlements and road networks play an important role in shaping golden-crowned sifaka foraging
663
and ranging behavior. Additionally, ecological factors such as season are drivers of home range
664
size and space use in this species. In terms of conservation implications, our study illustrates the
665
importance of studying primate groups in both the rainy and dry seasons to gain an accurate
666
snapshot of their ecology and resources needs. More specifically, by understanding how forest
667
type influences golden-crowned sifaka movement and foraging behavior, we can make
668
conservation management plans specific to the individual forest types throughout the Loky-
669
Manambato Protected Area (humid, moderate evergreen, dry deciduous, littoral, etc.), rather than
670
the region as a whole. Our findings will also inform Malagasy infrastructure and road
671
development plans by working with local conservation NGOs, government officials, and
672
construction teams to limit construction nearby lemur home ranges that are most impacted by
673
38
human activity. We would advise that the national road not be re-routed towards Binara, the
674
humid forest fragment, due to the strong avoidance lemurs display towards existing road
675
networks and the increased movement of lemurs within this forest fragment. We detected the
676
least avoidance of anthropogenic activity for lemurs in the moderate evergreen forest type,
677
suggesting they are more resilient to the negative effects of human infrastructure. Overall, as
678
anthropogenic disturbance continues to alter habitat structure throughout Madagascar, a deeper
679
knowledge of how fragmentation, habitat loss, and infrastructure development influence golden-
680
crowned sifaka space use, density, and population health will be essential for wildlife managers
681
to make well informed decisions that improve conservation plans for at-risk species.
682
683
Acknowledgments
684
We would like to sincerely thank Amidou Souleimany, Aylett Lipford, Giovanni Walters, our
685
local guides (Amadou, Andre, Augiste, Bezily, Christone, Da, Edward, Ishmael, Jaojoby, John,
686
Justin, Lahimena, Laurent, Lucien, Mamoud, Michelle, Moratombo, Patrice, Pierre, Seraphin,
687
Sylvano, Theodore, Thierry, Zoky), cooking staff (Ayati, Fatomia, Francia, Jao Fera, Nicole),
688
and porters, Fanamby (Serge Rajaobelina, Celin, Narcisse, !Tiana Andriamanana, Sylvano
689
Tsialazo), and Madagascar Institute for the Conservation of !Tropical Environments (MICET;
690
Benjamin Andriamihaja, Benji Randrianambinina, Claude, Nary) for their assistance with data
691
collection and logistics.
692
693
Authors Contributions
694
39
MS and IM designed the study. MS, BS, and TR conducted the fieldwork. MS, HA, and MC
695
conducted the coding and data analyses. MS wrote the first draft of the manuscript and all
696
authors contributed substantially to revisions.
697
698
Funding
699
We acknowledge funding from the Rufford Foundation and the National Science Foundation
700
GRFP (DGE 1651272). Opinions, findings, conclusions, or recommendations expressed are
701
those of the authors and do not necessarily reflect the views of the NSF.
702
703
Availability of data and materials
704
The datasets used and/or analyzed during the current study are available from the corresponding
705
author on reasonable request.
706
707
Declarations
708
Ethics approval and consent to participate
709
This research was conducted with permission from the Ministry of Foreign Affairs of
710
Madagascar, Madagascar National Parks, the Ministry of the Environment, Forests, and Tourism
711
(MEFT), and Madagascar Institute for the Conservation of Tropical Environments (MICET).
712
40
MICET was also instrumental in permit acquisition (N015/19/MEEF/SG/DGF/DSAP/SCB) and
713
overall research coordination. Our animal follow methods were approved by the Virginia Tech
714
Institutional Animal Care and Use Committee (IACUC) office (permit #17-127).
715
716
Consent for publication
717
Not applicable.
718
719
Competing interests
720
The authors declare that they have no competing interests.
721
722
References
723
724
1. Tabarelli M, Venceslau A, Cezar M, Paul J, Peres CA. Prospects for biodiversity conservation
725
in the Atlantic Forest : Lessons from aging human-modified landscapes. Biol Conserv.
726
2010;143:2328–40.
727
2. Allen AM, Singh NJ. Linking Movement Ecology with Wildlife Management and
728
Conservation. Front Ecol Evol. 2016;3:1–13.
729
3. Delciellos AC, Ribeiro SE, Vieira M V. Habitat fragmentation effects on fine-scale
730
movements and space use of an opossum in the Atlantic Forest. J Mammal. 2017;98:1129–36.
731
4. Gastón A, Ciudad C, Mateo-Sánchez MC, García-Viñas JI, López-Leiva C, Fernández-Landa
732
41
A, et al. Species’ habitat use inferred from environmental variables at multiple scales: How much
733
we gain from high-resolution vegetation data? Int J Appl Earth Obs Geoinf. 2017;55:1–8.
734
5. Festa-Bianchet M, Apollonio M. Animal Behavior and Wildlife Conservation. Washington,
735
DC: Island Press; 2003.
736
6. Merrick MJ, Koprowski JL. Should we consider individual behavior differences in applied
737
wildlife conservation studies? Biol Conserv [Internet]. Elsevier Ltd; 2017;209:34–44. Available
738
from: http://dx.doi.org/10.1016/j.biocon.2017.01.021
739
7. Grüss A, Kaplan DM, Guénette S, Roberts CM, Botsford LW. Consequences of adult and
740
juvenile movement for marine protected areas. Biol Conserv. 2011;144:692–702.
741
8. Angeloni L, Schlaepfer MA, Lawler JJ, Crooks KR. A reassessment of the interface between
742
conservation and behaviour. Anim Behav. 2008;75:731–7.
743
9. Choi CY, Peng HB, He P, Ren XT, Zhang S, Jackson M V., et al. Where to draw the line?
744
Using movement data to inform protected area design and conserve mobile species. Biol Conserv
745
[Internet]. Elsevier; 2019;234:64–71. Available from:
746
https://doi.org/10.1016/j.biocon.2019.03.025
747
10. Berger-Tal O, Polak T, Oron A, Lubin Y, Kotler BP, Saltz D. Integrating animal behavior
748
and conservation biology: A conceptual framework. Behav Ecol. 2011;22:236–9.
749
11. Cooke SJ, Blumstein DT, Buchholz R, Caro T, Fernández-Juricic E, Franklin CE, et al.
750
Physiology, behavior, and conservation. Physiol Biochem Zool. 2014;87:1–14.
751
12. Anthony LL, Blumstein DT. Integrating behaviour into wildlife conservation: The multiple
752
ways that behaviour can reduce N(e). Biol Conserv. 2000;95:303–15.
753
13. Berger-Tal O, Blumstein DT, Carroll S, Fisher RN, Mesnick SL, Owen MA, et al. A
754
systematic survey of the integration of animal behavior into conservation. Conserv Biol.
755
42
2016;30:744–53.
756
14. Shillinger GL, Palacios DM, Bailey H, Bograd SJ, Swithenbank AM, Gaspar P, et al.
757
Persistent leatherback turtle migrations present opportunities for conservation. PLoS Biol.
758
2008;6:1408–16.
759
15. Reyna-hurtado R, Teichroeb JA, Bonnell TR, Hernández-sarabia RU, Vickers SM, Serio-
760
silva JC, et al. Primates adjust movement strategies due to changing food availability. Behav
761
Ecol. 2018;29:368–76.
762
16. Trapanese C, Meunier H, Masi S. What, where and when: spatial foraging decisions in
763
primates. Biol. Rev. 2019. p. 483–502.
764
17. Rosenzweig ML. A Theory of Habitat Selection. Ecology [Internet]. 1981;62:327–35.
765
Available from: http://www.jstor.org/stable/1936707?seq=1#page_scan_tab_contents
766
18. Nagy-Reis MB, Setz EZF. Foraging strategies of black-fronted titi monkeys (Callicebus
767
nigrifrons) in relation to food availability in a seasonal tropical forest. Primates. Springer Japan;
768
2017;58:149–58.
769
19. Campera M, Serra V, Balestri M, Barresi M, Ravaolahy M, Randriatafika F, et al. Effects of
770
Habitat Quality and Seasonality on Ranging Patterns of Collared Brown Lemur (Eulemur
771
collaris) in Littoral Forest Fragments. Int J Primatol. 2014;35:957–75.
772
20. Pope NS, Jha S. Seasonal food scarcity prompts long-distance foraging by a wild social bee.
773
Am Nat. 2018;191:45–57.
774
21. Wato YA, Prins HHT, Heitkönig IMA, Wahungu GM, Ngene SM, Njumbi S, et al.
775
Movement patterns of African Elephants (Loxodonta africana) in a Semi-arid Savanna suggest
776
that they have information on the location of dispersed water sources. Front Ecol Evol.
777
2018;6:1–8.
778
43
22. Asensio N, Schaffner CM, Aureli F. Variability in core areas of spider monkeys (Ateles
779
geoffroyi) in a tropical dry forest in Costa Rica. Primates. 2012;53:147–56.
780
23. Branco PS, Merkle JA, Pringle RM, Pansu J, Potter AB, Reynolds A, et al. Determinants of
781
elephant foraging behaviour in a coupled human-natural system: Is brown the new green? J
782
Anim Ecol. 2019;88:780–92.
783
24. Rice MB, Apa AD, Wiechman LA. The importance of seasonal resource selection when
784
managing a threatened species: Targeting conservation actions within critical habitat
785
designations for the Gunnison sage-grouse. Wildl Res. 2017;44:407–17.
786
25. Street GM, Fieberg J, Rodgers AR, Carstensen M, Moen R, Moore SA, et al. Habitat
787
functional response mitigates reduced foraging opportunity : implications for animal fitness and
788
space use. Landsc Ecol. Springer Netherlands; 2016;31:1939–53.
789
26. Said S, Servanty S. The influence of landscape structure on female roe deer home-range size.
790
Landsc Ecol. 2005;20:1003–12.
791
27. Holzman S, Conroy MJ, Pickering J. Home Range , Movements , and Habitat Use of Coyotes
792
in Southcentral Georgia. J Wildl Manage. 1992;56:139–46.
793
28. McLean KA, Trainor AM, Asner GP, Crofoot MC, Hopkins ME, Campbell CJ, et al.
794
Movement patterns of three arboreal primates in a Neotropical moist forest explained by LiDAR-
795
estimated canopy structure. Landsc Ecol. Springer Netherlands; 2016;31:1849–62.
796
29. Campos FA, Bergstrom ML, Childers A, Hogan JD, Jack KM, Melin AD, et al. Drivers of
797
home range characteristics across spatiotemporal scales in a Neotropical primate, Cebus
798
capucinus. Anim Behav. Elsevier Ltd; 2014;91:93–109.
799
30. Haddad NM, Brudvig LA, Clobert J, Davies KF, Gonzalez A, Holt RD, et al. Habitat
800
fragmentation and its lasting impact on Earth ’ s ecosystems. Sci Adv. 2015;1–10.
801
44
31. Menchaca A, Rossi NA, Froidevaux J, Dias-freedman I, Caragiulo A, Wultsch C, et al.
802
Population genetic structure and habitat connectivity for jaguar (Panthera onca) conservation in
803
Central Belize. BMC Genet. BMC Genetics; 2019;20:1–13.
804
32. Peaden JM, Nowakowski AJ, Tuberville TD, Buhlmann KA, Todd BD. Effects of roads and
805
roadside fencing on movements, space use, and carapace temperatures of a threatened tortoise.
806
Biol Conserv. Elsevier; 2017;214:13–22.
807
33. Rogan MS, Miller JRB, Lindsey PA, Mcnutt JW. Socioeconomic drivers of illegal bushmeat
808
hunting in a Southern African Savanna. Biol Conserv. Elsevier; 2018;226:24–31.
809
34. Cardillo M, MacE GM, Gittleman JL, Jones KE, Bielby J, Purvis A. The predictability of
810
extinction: Biological and external correlates of decline in mammals. Proc R Soc B Biol Sci.
811
2008;275:1441–8.
812
35. Zeller KA, Wattles DW, Conlee L, Destefano S. Black bears alter movements in response to
813
anthropogenic features with time of day and season. Mov Ecol. Movement Ecology; 2019;7:1–
814
14.
815
36. Polfus JL, Hebblewhite M, Heinemeyer K. Identifying indirect habitat loss and avoidance of
816
human infrastructure by northern mountain woodland caribou. Biol Conserv [Internet]. Elsevier
817
Ltd; 2011;144:2637–46. Available from: http://dx.doi.org/10.1016/j.biocon.2011.07.023
818
37. Isaac NJB, Cowlishaw G. How species respond to multiple extinction threats. Proc R Soc B
819
Biol Sci. 2004;271:1135–41.
820
38. Almeida-Rocha JM d., Peres CA, Oliveira LC. Primate responses to anthropogenic habitat
821
disturbance: A pantropical meta-analysis. Biol Conserv [Internet]. Elsevier; 2017;215:30–8.
822
Available from: http://dx.doi.org/10.1016/j.biocon.2017.08.018
823
39. Estrada A, Garber PA, Rylands AB, Roos C, Fernandez-duque E, Fiore A Di, et al.
824
45
Impending extinction crisis of the world’s primates: Why primates matter. Sci Adv. 2017;3:1–
825
16.
826
40. Vieilledent G, Grinand C, Rakotomalala FA, Ranaivosoa R, Rakotoarijaona J, Allnutt TF, et
827
al. Combining global tree cover loss data with historical national forest cover maps to look at six
828
decades of deforestation and forest fragmentation in Madagascar. Biol Conserv. Elsevier;
829
2018;222:189–97.
830
41. Irwin MT. Feeding ecology of Propithecus diadema in forest fragments and continuous
831
forest. Int J Primatol. 2008;29:95–115.
832
42. Gerber BD, Arrigo-Nelson S, Karpanty SM, Kotschwar M, Wright PC. Spatial Ecology of
833
the Endangered Milne-Edwards’ Sifaka (Propithecus edwardsi): Do Logging and Season Affect
834
Home Range and Daily Ranging Patterns? Int J Primatol. 2012;33:305–21.
835
43. Erhart EM, Tecot SR, Grassi C. Interannual Variation in Diet, Dietary Diversity, and Dietary
836
Overlap in Three Sympatric Strepsirrhine Species in Southeastern Madagascar. Int J Primatol.
837
International Journal of Primatology; 2018;39:289–311.
838
44. Herrera JP, Borgerson C, Tongasoa L, Andriamahazoarivosoa P, Rasolofoniaina BJR,
839
Rakotondrafarasata ER, et al. Estimating the population size of lemurs based on their mutualistic
840
food trees. J Biogeogr. 2018;1–18.
841
45. Baden AL. A description of nesting behaviors , including factors impacting nest site selection
842
, in black-and-white ruffed lemurs (Varecia variegata). Ecol Evol. 2019;9:1010–28.
843
46. Lahann P, Dausmann KH. Live fast, die young: flexibility of life history traits in the fat-
844
tailed dwarf lemur (Cheirogaleus medius ). Behav Ecol Sociobiol. 2011;65:381–90.
845
47. Norscia I, Carrai V, Borgognini-Tarli SM. Influence of Dry Season and Food Quality and
846
Quantity on Behavior and Feeding Strategy of Propithecus verreauxi in Kirindy, Madagascar. Int
847
46
J Primatol. 2006;27:1001–22.
848
48. Quéméré E, Champeau J, Besolo A, Rasolondraibe E, Rabarivola C, Crouau-Roy B, et al.
849
Spatial variation in density and total size estimates in fragmented primate populations: The
850
golden-crowned sifaka (Propithecus tattersalli). Am J Primatol. 2010;72:72–80.
851
49. Quéméré E, Crouau-Roy B, Rabarivola C, Louis EE, Chikhi L. Landscape genetics of an
852
endangered lemur (Propithecus tattersalli) within its entire fragmented range. Mol Ecol.
853
2010;19:1606–21.
854
50. Goodman SM, Raherilalao MJ, Wohlhauser S. Site 6: Loky Manambato. Terr Prot Areas
855
Madagascar Their Hist Descr Biot. 2018.
856
51. Quemere E, Amelot X, Pierson J, Crouau-Roy B, Chikhi L. Genetic data suggest a natural
857
prehuman origin of open habitats in northern Madagascar and question the deforestation
858
narrative in this region. Proc Natl Acad Sci [Internet]. 2012;109:13028–33. Available from:
859
http://www.pnas.org/cgi/doi/10.1073/pnas.1200153109
860
52. Salmona J, Heller R, Quéméré E, Chikhi L. Climate change and human colonization
861
triggered habitat loss and fragmentation in Madagascar. Mol Ecol. 2017;26:5203–22.
862
53. Lehmann J, Boesch C. Social influences on ranging patterns among chimpanzees (Pan
863
troglodytes verus) in the Taı National Park, Cote d ’ Ivoire. Behav Ecol. 2002;14:642–9.
864
54. Steiniger S, Hunter AJS. A scaled line-based kernel density estimator for the retrieval of
865
utilization distributions and home ranges from GPS movement tracks. Ecol Inform. 2013;13:1–8.
866
55. Avgar T, Potts JR, Lewis MA, Boyce MS. Integrated step selection analysis : bridging the
867
gap between resource selection and animal movement. Methods Ecol Evol. 2016;7:619–30.
868
56. Kranstauber B. Modelling animal movement as Brownian bridges with covariates. Mov Ecol.
869
Movement Ecology; 2019;7:1–10.
870
47
57. Fieberg J, Borger L. Could you please phrase ‘“ home range ”’ as a question? J Mammal.
871
2012;93:890–902.
872
58. Kranstauber B, Kays R, Lapoint SD, Wikelski M, Safi K. A dynamic Brownian bridge
873
movement model to estimate utilization distributions for heterogeneous animal movement. J
874
Anim Ecol. 2012;81:738–46.
875
59. Meyers DM. The Effects of Resource Seasonality on Behavior and Reproduction in the
876
Golden-crowned Sifaka (Propithecus Tattersalli, Simons, 1988) in Three Malagasy Forests.
877
Duke University; 1993.
878
60. Erkert HG, Kappeler PM. Arrived in the light : diel and seasonal activity patterns in wild
879
Verreaux s sifakas ( Propithecus v . verreauxi ; Primates : Indriidae ). Behav Ecol Sociobiol.
880
2004;57:174–86.
881
61. Altmann J. Observational Study of Behavior: Sampling Methods. Behaviour. 1974;49:227–
882
67.
883
62. Gurarie E, Andrews RD, Laidre KL. A novel method for identifying behavioural changes in
884
animal movement data. Ecol Lett. 2009;12:395–408.
885
63. Horne JS, Garton EO, Krone SM, Lewis JS. Analyzing animal movements using Brownian
886
bridges. Ecology. 2007;88:2354–63.
887
64. Kranstauber B, Smolla M, Scharf A. Visualizing and Analyzing Animal Track Data. 2013.
888
65. R Core Team. Vienna, Austria: R Foundation for Statistical Computing; 2020.
889
66. Burnham KP, Anderson DR. Model selection and multimodel inference: a practical
890
information-theoretic approach. New York, New York: Springer-Verlag; 2002.
891
67. Atwood TC, Weeks HP. Spatial home-range overlap and temporal interaction in eastern
892
coyotes: The influence of pair types and fragmentation. Can J Zool. 2003;81:1589–97.
893
48
68. Bates D, Mächler M, Bolker BM, Walker SC. Fitting Linear Mixed-Effects Models Using
894
lme4. J Stat Softw. 2015;67:1–48.
895
69. Hurvich CM, Tsai CL. Regression and time series model selection in small samples.
896
Biometrika. 1989;76:297–307.
897
70. Lele SR, Merrill EH, Keim J, Boyce MS. Selection, use, choice and occupancy: clarifying
898
concepts in resource selection studies. J Anim Ecol. 2013;82:1183–91.
899
71. Manly BF, McDonald L, Thomas D, McDonald TL, Erickson WP. Resource Selection by
900
Animals. 2nd ed. Springer Netherlands; 2002.
901
72. Johnson DJ. The comparison of usage and availability measurements for evaluating resource
902
preference. Ecology. 1980;61:65–71.
903
73. Benson JF. Improving rigour and efficiency of use-availability habitat selection analyses with
904
systematic estimation of availability. Methods Ecol Evol. 2013;4:244–51.
905
74. Gillies CS, Hebblewhite M, Nielsen SE, Krawchuk MEGA, Aldridge CL, Jacqueline L, et al.
906
Application of random effects to the study of resource. J Anim Ecol. 2006;75:887–98.
907
75. Hebblewhite M, Merrill E. Modelling wildlife – human relationships for social species with
908
mixed-effects resource selection models. J Appl Ecol. 2008;45:834–44.
909
76. Pebesma E, Graeler B, Pebesma ME. Package ‘gstat’. 2019.
910
77. De Smith MJ, Goodchild MF, Longley PA. Geospatial Analysis: A Comprehensive Guide to
911
Principles Techniques and Software Tools. 6th ed. Ingram Publisher Services; 2018.
912
78. Benadi G, Fitchel C, Kappeler P. Intergroup Relations and Home Range Use in Verreaux’s
913
Sifaka (Propithecus verreauxi ). Am J Primatol. 2008;70:956–65.
914
79. Setash CM, Zohdy S, Gerber BD, Karanewsky CJ. A biogeographical perspective on the
915
variation in mouse lemur density throughout Madagascar. Mamm Rev. 2017;47:212–29.
916
49
80. Wright PC. Lemur traits and Madagascar ecology: coping with an island environment. Am J
917
Phys Anthropol [Internet]. 1999;Suppl 29:31–72. Available from:
918
http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&id=10601983&retmode=
919
ref&cmd=prlinks%5Cnpapers2://publication/uuid/E29CB817-D796-49CD-81E0-
920
0405F2237A27
921
81. Ostro LET, Silver SC, Koontz FW, Young TP. Ranging behavior of translocated and
922
established groups of black howler Ranging behavior of translocated and established groups of
923
black howler monkeys Alouatta pigra in Belize, Central America. Biol Conserv. 1999;87:181–
924
90.
925
82. Kling KJ, Yaeger K, Wright PC. Trends in forest fragment research in Madagascar:
926
Documented responses by lemurs and other taxa. Am J Primatol. 2020;1–14.
927
83. Rhoads CL, Bowman JL, Eyler B. Home Range and Movement Rates of Female Exurban
928
White-Tailed Deer. J Wildl Manage. 2010;74:987–94.
929
84. Viana DS, Granados JE, Fandos P, Pérez JM, Cano-manuel FJ, Burón D, et al. Linking
930
seasonal home range size with habitat selection and movement in a mountain ungulate. Mov
931
Ecol. Movement Ecology; 2018;6:1–11.
932
85. Prokopenko CM, Boyce MS, Avgar T. Characterizing wildlife behavioural responses to
933
roads using integrated step selection analysis. J Appl Ecol. 2017;54:470–9.
934
86. Mohammadi A, Almasieh K, Clevenger AP, Fatemizadeh F. Road expansion: A challenge to
935
conservation of mammals , with particular emphasis on the endangered Asiatic cheetah in Iran. J
936
Nat Conserv. Elsevier; 2018;43:8–18.
937
87. Criffield M, Van De Kerk M, Leone E, Cunningham MW, Lotz M, Oli MK, et al. Assessing
938
impacts of intrinsic and extrinsic factors on Florida panther movements. J Mammal.
939
50
2018;99:702–12.
940
88. Ünal Y, Pekin BK, Oğurlu I, Süel H, Koca A. Human, domestic animal, Caracal (Caracal
941
caracal), and other wildlife species interactions in a Mediterranean forest landscape. Eur J Wildl
942
Res. 2020;66:1–10.
943
89. Scrafford MA, Avgar T, Heeres R, Boyce MS. Roads elicit negative movement and habitat-
944
selection responses by wolverines (Gulo gulo luscus). Behav Ecol. 2018;29:534–42.
945
946
