ArticlePDF Available

Abstract and Figures

South Africa’s decentralized approach to conservation entails that wildlife outside formally protected areas inhabit complex multi-use landscapes, where private wildlife business (ecotourism and/or hunting) co-exist in a human-dominated landscape matrix. Under decentralized conservation, wildlife is perceived to benefit from increased amount of available habitat, however it is crucial to understand how distinct management priorities and associated landscape modifications impact noncharismatic taxa, such as small mammals. We conducted extensive ink-tracking-tunnel surveys to estimate heterogeneity in rodent distribution and investigate the effect of different environmental factors on abundance patterns of two size-based rodent groups (small- and medium-sized species), across three adjacent management contexts in NE KwaZulu-Natal, South Africa: a private ecotourism game reserve, mixed farms and traditional communal areas (consisting of small clusters of houses interspersed with grazing areas and seminatural vegetation). Our hypotheses were formulated regarding the (1) area typology, (2) vegetation structure, (3) ungulate pressure and (4) human disturbance. Using a boosted-regression-tree approach, we found considerable differences between rodent groups’ abundance and distribution, and the underlying environmental factors. The mean relative abundance of medium-sized species did not differ across the three management contexts, but small species mean relative abundance was higher in the game reserves, confirming an influence of the area typology on their abundance. Variation in rodent relative abundance was negatively correlated with human disturbance and ungulate presence. Rodent abundance seems to be influenced by environmental gradients that are directly linked to varying management priorities across land uses, meaning that these communities might not benefit uniformly by the increased amount of habitat promoted by the commercial wildlife industry.
Content may be subject to copyright.
animals
Article
Patterns and Drivers of Rodent Abundance across a South
African Multi-Use Landscape
Beatriz C. Afonso 1, * , Lourens H. Swanepoel 2, Beatriz P. Rosa 1, Tiago A. Marques 3,4, Luís M. Rosalino 1,
Margarida Santos-Reis 1and Gonçalo Curveira-Santos 1


Citation: Afonso, B.C.; Swanepoel,
L.H.; Rosa, B.P.; Marques, T.A.;
Rosalino, L.M.; Santos-Reis, M.;
Curveira-Santos, G. Patterns and
Drivers of Rodent Abundance across
a South African Multi-Use Landscape.
Animals 2021,11, 2618. https://
doi.org/10.3390/ani11092618
Academic Editor: Emiliano Mori
Received: 2 August 2021
Accepted: 31 August 2021
Published: 7 September 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
cE3c—Centre for Ecology, Evolution and Environmental Changes, Faculdade de Ciências da Universidade de
Lisboa, Campo Grande, 1749-016 Lisboa, Portugal; beatrizrprosa@hotmail.com (B.P.R.);
lmrosalino@fc.ul.pt (L.M.R.); mmreis@fc.ul.pt (M.S.-R.); gcurveirasantos@gmail.com (G.C.-S.)
2Department of Zoology, School of Mathematical & Natural Sciences, University of Venda,
Thohoyandou 0950, Limpopo, South Africa; lourens.swanepoel@univen.ac.za
3Centre for Research into Ecological and Environmental Modelling, The Observatory, University of St
Andrews, St Andrews KY16 9LZ, UK; tiago.marques@st-andrews.ac.uk
4Centro de Estatística e Aplicações, Departamento de Biologia Animal, Faculdade de Ciências, Universidade
de Lisboa, 1749-016 Lisboa, Portugal
*Correspondence: beatrizcardosoafonso@gmail.com
Simple Summary:
Wildlife ecological patterns are driven not only by environmental and biological
contexts, but also by landscape-management schemes that shape those contexts. The present study
aims to determine the effect of different environmental factors (including management schemes)
on the occurrence patterns of a southern African small mammal community. Based on a landscape
where three land-use contexts that differ in their levels of human presence and/or where activities
coexist (private ecotourism reserve, mixed farms and traditional communal areas), and by using
a body-size-based approach (i.e., using two size-based rodent groups—medium and small—as
models), we found that the mean relative abundance of medium-sized species did not differ across
the management contexts, but small species’ mean relative abundance was higher in the game reserve.
The overall variation in rodent abundance was negatively affected by ungulate presence (possibly
linked to a decrease in food availability) and by human presence (increased disturbance). Rodent
abundance seems to be influenced by environmental gradients that are directly linked to varying
management priorities across land uses, meaning that these communities might not benefit uniformly
by the increased amount of habitat promoted by the commercial wildlife industry.
Abstract:
South Africa’s decentralized approach to conservation entails that wildlife outside formally
protected areas inhabit complex multi-use landscapes, where private wildlife business (ecotourism
and/or hunting) co-exist in a human-dominated landscape matrix. Under decentralized conservation,
wildlife is perceived to benefit from increased amount of available habitat, however it is crucial
to understand how distinct management priorities and associated landscape modifications impact
noncharismatic taxa, such as small mammals. We conducted extensive ink-tracking-tunnel surveys
to estimate heterogeneity in rodent distribution and investigate the effect of different environmental
factors on abundance patterns of two size-based rodent groups (small- and medium-sized species),
across three adjacent management contexts in NE KwaZulu-Natal, South Africa: a private ecotourism
game reserve, mixed farms and traditional communal areas (consisting of small clusters of houses
interspersed with grazing areas and seminatural vegetation). Our hypotheses were formulated
regarding the (1) area typology, (2) vegetation structure, (3) ungulate pressure and (4) human
disturbance. Using a boosted-regression-tree approach, we found considerable differences between
rodent groups’ abundance and distribution, and the underlying environmental factors. The mean
relative abundance of medium-sized species did not differ across the three management contexts,
but small species mean relative abundance was higher in the game reserves, confirming an influence
of the area typology on their abundance. Variation in rodent relative abundance was negatively
correlated with human disturbance and ungulate presence. Rodent abundance seems to be influenced
by environmental gradients that are directly linked to varying management priorities across land
Animals 2021,11, 2618. https://doi.org/10.3390/ani11092618 https://www.mdpi.com/journal/animals
Animals 2021,11, 2618 2 of 18
uses, meaning that these communities might not benefit uniformly by the increased amount of habitat
promoted by the commercial wildlife industry.
Keywords: non-invasive sampling; ecological modelling; management options; conservation
1. Introduction
In South Africa, agricultural intensification, and overgrazing have led to profound
land use changes [
1
]. Historically, most landscapes were converted into livestock farms and
farmlands, either as intensive, extensive, or communally managed areas [
2
], leading to the
destruction, degradation and/or fragmentation of natural ecosystems [
3
]. Consequently,
such habitat destruction led to declines in wildlife populations and distribution in much of
South African nonprotected areas [4].
However, the establishment of national policies attributing custodial rights over
wildlife to landowners, prompted a transition in the governance of natural resources from
the state to privates [
5
]. This political option led to widespread conversion of rangelands,
i.e., farmlands and livestock farms, into areas dedicated to commercial wildlife industries,
such as game ranching and private game/ecotourism reserves [
6
]. The positive conserva-
tion outcomes of these policies for economically valuable and charismatic species [
7
] is
believed to have an umbrella effect on other taxa, mainly through the increased coverage,
representativeness and connectivity of protected/restored habitats [
6
,
8
]. However, the ef-
fect of such management approaches is unexplored for most overlooked—but functionally
important—taxa, such as rodents [
9
]. Thus, information on the ecological responses of
less-charismatic taxa is needed to better gauge the complementary conservation role of
South Africa’s private land.
In South Africa, game farms and private game reserves often coincide across relatively
small scales, rooted in human-dominated landscapes (e.g., communal lands) [
10
]. These
land uses have contrasting management priorities and, consequently, distinct impacts on
the landscape structure and wildlife ecological patterns. In game farms, the main objective
is to maximize the production of ungulates for meat or hunting, while in private game
reserves the goal is to maintain charismatic species, promoting ecotourism-based activi-
ties [
11
]. Often, these wildlife-oriented land uses are surrounded by human-dominated
areas with high levels of anthropogenic disturbance. The regional co-existence of all these
land uses generates complex multi-tenured landscapes, usually divided by semi-permeable
wildlife fences, influencing the biodiversity supported by each of these land uses [12].
Management actions directed to charismatic or valuable species may have cascading
effects on rodents, usually overlooked and handled like pests [
13
15
]. However, it is crucial
to understand the effect of human-induced land-use changes on rodent spatial patterns,
as well as the underlying ecological mechanisms thereof, since rodents are fundamental
for some ecosystem functions [
16
]. Rodents are primary consumers [
16
] and support
a large community of predators [
17
,
18
], which makes them a vital link in food-chain
structuring [
19
]. Moreover, they are considered useful indicators of ecosystem functioning
as they are valuable tools to the description and monitoring of habitat integrity. For these
reasons, rodents have been used as model species to understand how land use changes
affects wildlife [16].
Several factors have been identified as influential in shaping rodent community and
population structures, many of which are often determined by the landscape management
options [
20
]. Some studies have indicated that vegetation type and structure are fundamen-
tal drivers of rodent occurrence and abundance [
21
23
]. For example, areas with greater
herbaceous coverage favor rodents by providing shelter against predators, food, and ade-
quate microclimatic conditions [
24
]. Studies have shown negative effects of overgrazing on
small mammals’ abundance, by reducing the herbaceous stratum, increasing trampling risk
and feeding competition with ungulates [
9
,
24
27
]. Regarding rodent distribution, it tends
Animals 2021,11, 2618 3 of 18
to be uniform when the habitat is favorable and resources are abundant. However, when
disturbances increase the level of habitat heterogeneity, causing landscape fragmentation,
their distribution is mostly clumped [28,29].
Rodents are not a homogeneous group, since different species may establish distinct
relationships with the environmental and biotic components of the ecosystem. For example,
larger rodents’ range over larger spatial scales than smaller rodents [
30
] and, therefore,
are more susceptible to changes at this landscape level [31].
Changes in management priorities across South African multi-tenured landscapes will
have a direct impact on these environmental drivers and, ultimately, in the distribution and
abundance of small mammal species across and within management contexts. For instance,
when management measures promote the abundance of ungulates (e.g., as prey for large
carnivore populations in ecotourism reserves, or as hunting assets in game farms), grazing
pressure will increase, negatively influencing the herbaceous strata [
26
]. Alongside with
long dry and hot seasons [
32
], these conditions may lead to shrub encroachment, known to
reduce food availability (leaves, seeds, and arthropods) for ground dwelling rodents [
33
].
Nevertheless, some rodent species are usually considered efficient colonizers of human
shaped environments [
13
,
34
], as they are able to use human-related food resources due to
their omnivore character [35].
Although the processes that regulate small mammals’ spatial distribution are known
for some landscapes (e.g., woodland [
29
] and mixed forest [
36
]), there is a lack of informa-
tion regarding the drivers of rodent-abundance patterns in African savannas (but see [
9
,
37
]),
as well as how these vary across different management schemes. Here, we evaluated the
variation in rodent abundance across three adjacent management contexts, spanning a
private ecotourism game reserve, mixed farms and communally owned land, managed
by Zulu tribal authorities [
12
], under the following two main objectives: (1) to estimate
heterogeneity in small-mammal-abundance distribution (mean abundance and patchiness)
across management contexts (game reserve, mixed farms and communal lands); and (2) to
determine the main, fine-scale environmental factors affecting small-mammal-abundance
patterns across land-use types. These objectives were tested in two size-based rodent
groups, for a more detailed assessment of ecological responses.
Linked to these two goals, we tested four hypothetical drivers of rodent communities:
(i)
An area-typology hypothesis, i.e., cumulative effect of management-induced changes
to vegetation, grazing pressure, etc., creates area-specific differences in rodent abun-
dance. Patchiness will also be tested to acknowledge in which area each group is
more or less clumped, regarding their abundance values. Although the exact effect of
area on rodent abundance is not fully predictable [
37
] (given the disturbance gradient)
we expected the communal lands to have the lowest values of abundance and highest
patchiness (i.e., more clumped), followed by mixed farms and the game reserve, with
higher abundances and lower patchiness;
(ii)
A vegetation-structure hypothesis, i.e., areas with higher herbaceous cover will have
a positive influence on both rodent size-based groups, since it shapes the ability of
the landscape to provide protection against potential predators [2123,25,27,38];
(iii)
An ungulate-pressure hypothesis, i.e., rodent species abundance is negatively influ-
enced by the abundance of ungulates, since higher grazing pressure tends to decrease
herbaceous land cover, increase disturbance due to the trampling effect, and increase
landscape fragmentation [9,24];
(iv)
A human-disturbance hypothesis, i.e., rodent species’ distribution is negatively influ-
enced by human disturbance factors, such as the presence of domestic animals and
households that may constrain species’ presence [14,39].
2. Materials and Methods
2.1. Study Area
This study was implemented in the Maputaland–Pondoland–Albany Biodiversity
Hotspot [
40
] in northern KwaZulu-Natal, South Africa. Our specific study area is char-
Animals 2021,11, 2618 4 of 18
acterized by a spatial gradient of human intervention, ranging from the Mun-ya-wana
private game reserve (less subject to human associated activities), to mixed game farms
and to communally managed lands, where two distinct Zulu communities are settled
(Figure 1b). The Mun-ya-wana private game reserve (27
40
0
S–27
55
0
S; 31
12
0
E–32
26
0
E)
represents the union of several properties without internal fences, managed by private
owners whose goal is to explore eco-touristic products, therefore promoting wildlife and
habitat conservation. Those management objectives are commonly related with a more
sustainable use of wildlife, typically wildlife-viewing tourism [
41
]. The reserve is sur-
rounded, to the South, by a mosaic of commercial game ranches for the production of wild
ungulate species, occasionally mixed with domestic cattle [
42
] (hereafter mixed farms) and
represents large expanses of natural habitat with low human density. Communal lands to
the east are composed of households, interspersed with pasture areas and semi-natural
vegetation. The region is characterized by a warm-temperature climate, with a humid
and hot summer (October to April), according to the Köppen–Geiger classification. Mean
monthly temperatures range from 19
C in July to 31
C in January, and the average annual
precipitation is 800 mm [
43
,
44
]. Elevation ranges from 3 m to 304 m above sea level [
45
],
dominated by a similar mixture of vegetation throughout the area (bushveld, woodland
and grassland) [
46
] (Figure 1b). Nevertheless, the game reserve hosts a higher diversity and
abundance of pristine habitats, such as indigenous forests, while mixed farms are mainly
composed of pasture areas (low shrubland and grassland–Figure 1). Contrarily, communal
lands have the lowest proportion of vegetation and the highest cover of urban–village
occupation (Figure 1).
2.2. Rodent Sampling
Rodents were sampled between October and November 2017 (the southern hemi-
sphere’s spring) using ink-tracking tunnels [
42
], left active in the field for four consecutive
nights (open circles in Figure 1c). Ink-tracking tunnels were made of robust corrugated
plastic (55
×
10
×
10 cm), open on both ends to allow rodents to enter. Both entrances of
the tunnel are equipped with an adhesive paper with the glue side up, and an ink pad
(
12 ×10 cm
) was placed in the floor center [
47
] (Figure S1B). In the middle of the tunnel,
a small PVC-pipe section, hanging from the ceiling, was installed, and contained bait
composed of a mixture of peanut butter, oatmeal and sunflower oil [
46
]. The pipe was
used to prevent the consumption of the bait by the animals entering/crossing the tunnel.
The ink tunnels were placed on the ground, grouped in clusters of nine, in a Y formation,
10 m apart from each other (Figure 1c). The arms of the Y formation were 120 degrees
apart (
Figure 1c
). This design provided an adequate spatial coverage in relation to the
home-ranges of the rodent species, also ensuring some level of independence between
sampling units, considering the mean distance between sites (see below). After the four-
day sampling period, the plates of each ink tunnel (containing footprints and tracks) were
photographed individually, always at the same distance and with a reference scale.
The footprint data was used to estimate rodent relative abundance, using the propor-
tion of the tunnels with records (track index; TI–for more details see Supplementary Materi-
als) [
48
]. To ensure that this approach captured spatial heterogeneity in relative abundance,
we conducted a small trial, comparing the abundance indices derived from ink tunnels
to those obtained from live-trapping (see Supplementary Materials, PART A). As track
identification at the species level is very time consuming and not viable in large-scale
studies, and as distinguishing footprints from similar-sized species is very difficult and
bias prone, we opted for dividing tracks into groups based on track size (for more details
see Supplementary Materials PART A; Figures S1A and S2A, Table S1A). Rodent footprints
were grouped into three different size-based groups per body length/weight, assuming
a relation between rodent body length/weight and footprint sizes [
49
,
50
]: small (body
length: 50–100 mm), medium (100–150 mm) and large rodents (150–200 mm) (
Figure S2B
).
Sampling intentionally took place outside the breeding season (which peaks in the wet
season, [
51
]), in order to avoid grouping juveniles in the wrong size-based group. However,
Animals 2021,11, 2618 5 of 18
considering the low number of detections of large rodents in ink-tracking tunnels, we only
analyzed the data from small- and medium-sized rodents (see Results). The most common
species captured during live trapping and linked to each group were Mus minutoides and
Dendromus melanotis for small rodents, Mastomys natalensis and Saccostomus campestris for
medium rodents and Otomys angoniensis and Rattus rattus for large rodents (Table S2A).
Animals 2021, 11, x 5 of 18
2. Materials and Methods
2.1. Study Area
This study was implemented in the Maputaland–Pondoland–Albany Biodiversity
Hotspot [40] in northern KwaZulu-Natal, South Africa. Our specific study area is charac-
terized by a spatial gradient of human intervention, ranging from the Mun-ya-wana pri-
vate game reserve (less subject to human associated activities), to mixed game farms and
to communally managed lands, where two distinct Zulu communities are settled (Figure
1b). The Mun-ya-wana private game reserve (27°40 S–27°55 S; 31°12 E–32°26 E) repre-
sents the union of several properties without internal fences, managed by private owners
whose goal is to explore eco-touristic products, therefore promoting wildlife and habitat
conservation. Those management objectives are commonly related with a more sustaina-
ble use of wildlife, typically wildlife-viewing tourism [41]. The reserve is surrounded, to
the South, by a mosaic of commercial game ranches for the production of wild ungulate
species, occasionally mixed with domestic cattle [42] (hereafter mixed farms) and repre-
sents large expanses of natural habitat with low human density. Communal lands to the
east are composed of households, interspersed with pasture areas and semi-natural veg-
etation. The region is characterized by a warm-temperature climate, with a humid and hot
summer (October to April), according to the Köppen–Geiger classification. Mean monthly
temperatures range from 19 °C in July to 31 °C in January, and the average annual precip-
itation is 800 mm [43,44]. Elevation ranges from 3 m to 304 m above sea level [45], domi-
nated by a similar mixture of vegetation throughout the area (bushveld, woodland and
grassland) [46] (Figure 1b). Nevertheless, the game reserve hosts a higher diversity and
abundance of pristine habitats , such as indigenous forests, while mixed farms are mainly
composed of pasture areas (low shrubland and grassland–Figure 1). Contrarily, commu-
nal lands have the lowest proportion of vegetation and the highest cover of urban–village
occupation (Figure 1).
Figure 1.
Location of the study area in South Africa, with the black dot representing the location
of the study area in the Maputaland region of northern KwaZulu-Natal (
a
); landscape composition
of the three studied areas with distinct management schemes–Mun-ya-wana private game reserve,
mixed farms and communal land (Zulu tribal land)–with the location of the sampling points and the
number of sampling points per area (in parenthesis) (
b
); each sampling point included a camera trap
in the center and nine ink tunnels, distributed in a Y shape (open circles represent ink tunnels) (c).
2.3. Environmental Variables
Vegetation structure variables were collected using two different approaches: field
measures and remote-sensed products [
52
]. All variables collected have been previously
detected as influential to rodent presence elsewhere (e.g., vegetation cover) [
21
,
23
]. Shrub-
and-grass cover were visually estimated and assigned the corresponding Edwards classifi-
cation category [
53
] (see Table 1for details), within a 30 m radius buffer, centered on the ink
tunnel’s Y formation. Regarding the land use, the predominant categories were selected
(thicket, grassland, sand forest and urban villages) and, for each buffer, was assigned
the category with the highest cover. According to the type of crops present in the study
area, the harvesting season occurs mostly between April and June [
54
], not coinciding
with the study period. Therefore, we assumed that there would be no influence of crop
productivity on the distribution/abundance of rodents in our study. The percentage of
tree cover was assessed based on the Global Forest Watch database (Table 1). We also
selected the Normalized Difference Vegetation Index (NDVI), widely used as a vegetation
productivity proxy, collected from Landsat 8 Images [55].
Animals 2021,11, 2618 6 of 18
Variables of ungulate pressure and human disturbance were collected from Curveira-
Santos et al. [
12
] camera-trap surveys. Cameras, located in the center of the Y formation,
were active for 60–90 days, and attached to a tree or metal stake, 30 cm above the ground,
without any bait and set to photograph at minimum delay (1 s for daytime and 30 s for
night-time) (see [
12
] for details). Each of the defined ink-tunnel clusters (i.e., one cluster
includes nine ink tunnels and one camera-trap; Figure 1c) were spaced approximately
1.4 km apart (Figure 1b). In total, were sampled 196 points: 100 points in Mun-ya-wana
eco-tourism/game reserve, 50 points in mixed farms and 46 points in communal lands.
Capture rates, expressed as the number of independent camera records (>1 h interval
between photographs of the same species, per 100 trap-days) for livestock (cows and goats),
wild ungulates and human disturbance, were used as surrogates of disturbance in the
modeling procedure (Table 1).
Wild ungulates were grouped according to two criteria: weight, since trampling is
one of the main negative impacts of ungulates on rodents [
26
], and/or the fact that they
are actively managed in all studied areas (Table S1B). Only ungulates weighing between
45–200 kg and actively managed were used in the analysis, since they are more abundant
than other ungulates, as they are present throughout the areas under study, and because
they have a greater impact on rodents, due to their weight (Table S1B). Livestock were also
separated in two weight classes: i.e., goats and cows.
Table 1.
Environmental variables used in the modeling procedure used to assess the determinants of rodent abundance,
collected in the field, from camera-trapping or based on remote-sensing data (GIS-based variables). The variable description,
acronym, range, resolution and source, as well the reference that support their influence on rodent presence/abundance, are
listed. H1—Hypothesis 1; H2—Hypothesis 2; H3—Hypothesis 3, H4—Hypothesis 4.
Variable
Acronym Description Mean/Range Resolution Source Supporting
References
AREA TYPE (H1)
Area Managment context
Mixed farms
Mun-ya-wana
Communal lands
Collected at point - [37]
VEGETATION STUCTURE (H2)
Tree_Cover % Tree Cover 30.80/6–72% 30 ×30 m
Global Forest Watch
https://www.
globalforestwatch.org/
(16 April 2019)
[27,56]
Shrub_Cover % of Shrub cover
Continuous (C)—76–100%
Semi-continuous
(SC)—51–74%
Moderated closed
(MC)—26–50%
Semi-open (SO)—11–25%
Open (O)—0–10%
30 m buffer Visually estimated [22,27,39,5760]
Grass_Cover % of Grass cover
Continuous (C)—76–100%
Semi-continuous
(SC)—51–74%
Moderated closed
(MC)—26–50%
Semi-open (SO)—11–25%
Open (O)—0–10%
30 m buffer Visually estimated [22,25,38,57]
Land_use Land use categories
Thicket
Grassland
Sand Forest
Urban Villages
30 m buffer
2013–2014 National Land
Cover South Africa-SASDI
http://www.sasdi.net/
(16 April 2019)
[2123]
NDVI
Normalized
difference vegetation
index calculated from
Landsat images
0.48/0.28–0.67 30 ×30 m
Landsat 8
https:
//earthexplorer.usgs.gov/
(18 April 2019)
[60,61]
Animals 2021,11, 2618 7 of 18
Table 1. Cont.
Variable
Acronym Description Mean/Range Resolution Source Supporting
References
UNGULATE PRESSURE (H3)
Goats
Capture rate of goats
(number of records
per 100 days of
trapping)
0.16/0–1.88 Collected at point Camera-trapping survey
[9,24,26]
Livestock
Capture rate of cows
(number of records
per 100 days of
trapping)
0.20/0–3.17 Collected at point Camera-trapping survey
Wild
Ungulates
Capture rate of
ungulates (number of
records per 100 days
of trapping)
0.750/0–3.48 Collected at point Camera-trapping survey
DISTURBANCE VARIABLES (H4)
HUMANS Capture rate of
humans 0.84/0–10 Collected at point Camera-trapping survey
[39]
DIST Distance to houses 2.738/0.031–9.867 km Collected at point Camera-trapping survey
2.4. Data Analyses/Modelling
2.4.1. Spatial Patterns of Rodent Relative Abundance Across Areas and Size-Based Groups
Differences in mean abundance values of size-based groups (small and medium)
between study areas (Mun-ya-wana game reserve, mixed farms and communal lands) were
tested using GLM with 3-level area covariate and binomial error distribution. The magni-
tude of patchiness in each area was ascertained by spatial-point pattern analysis of count
data using Lloyd’s index of patchiness [
62
]. A Lloyd’s index of 1 indicates a random
distribution, whilst one <1 suggests uniformity and >1 patchiness.
2.4.2. Influence of Environmental Variables on Rodent Relative Abundance
Due to the high number of candidate variables and to avoid multicollinearity bias,
we first estimated the nonparametric Spearman’s correlation (r
s
) using the “psych” R
package [
63
]. When a high correlation between two covariates was detected (r
s
0.7; [
62
]),
the variable that was less correlated with the dependent variable was excluded from the
analysis [64].
The influence of all candidate variables on rodent relative abundance was tested
using a boosted-regression-tree (BRT) approach, implemented with the “gbm” package [
65
]
in R [
66
,
67
]. This modelling technique encompasses the advantages of regression trees
(e.g., predictor variables can be of any type, analysis is insensitive to outliers and can
accommodate missing data [
68
]), overcoming their low predictive capacity through the
boosting algorithm [
69
]. The final model is a linear addition of several regression models
in which the simplest term is a tree [68,70].
Boosted-regression-tree models are resilient to model overfitting but, to have a bet-
ter predictive performance, we defined, a priori, the model’s input parameters based on
Carslaw and Taylor’s suggestions [
70
]. In BRT, learning rate (lr) is the shrinkage parameter
that controls the contribution of each tree to the model, and tree complexity (tc) determines
the number of nodes in a tree and, consequently, its size. These two parameters control the
number of trees in the model, while the bag fraction (0.5) selects the proportion of data be-
ing used at each step [
61
,
70
,
71
]. All models were fitted to allow interactions using a ten-fold
cross validation to determine the optimal number of trees for each model. The largest learn-
ing rate and the smallest tree complexity were selected to allow a minimum of 1000 trees
in the BRT fitting process (see [
68
]). Non-informative variables were removed during the
fitting process, allowing the simplification of the set of variables [
68
]. This simplification
consisted of defining how many variables the function can test to remove, based on relative
Animals 2021,11, 2618 8 of 18
influence and total number of variables. Then, a graph was produced showing differences
in the predicted deviance according to several scenarios, each one with a different number
of variables removed. Next, the number of variables to eliminate was decided, and they
were removed in order of minor relative influence. We defined a threshold value and only
reported the interactions with relative influence values >10%. The final relative influence
of each variable was calculated by averaging the number of times a covariate is used for
splitting, weighted by the squared improvement to the model as the result of each split.
It is then scaled, such that the values sum to 100 [
72
]. Fitted values were plotted in relation
to the most important predictors, revealing their effects on rodent abundance. Explained
deviance was calculated using the following formula from Abeare (2009) [73]
D2=1residual deviance
total deviance
The 95% confidence intervals of each variable were estimated for the fitted function
by taking 500 bootstrap samples of the input data, with the same size as the original data.
A BRT was fitted to each sample, and the 5th and 95th percentiles were calculated for
the points of each function. Models were built separately for small- and medium-sized
rodents. For each model performed, interactions between typology and the other influential
independent variables (i.e., relative importance above >10%) were estimated, to evaluate
context-dependency in the influence in the effect environmental variable associated with
the management context. All analyses were implemented in R via R Studio Version
1.1.463 [66,67].
3. Results
3.1. Spatial Patterns of Rodent Abundance Across Areas and Size-Based Groups
From the 192 sampling points monitored, 85% presented small rodent tracks, while
76% detected the occurrence of medium rodents, with an overlap in 35% of sites and
inter-area variation in detection (i.e., number of tunnels with signs/total number of tunnels,
Table S2B). Mean abundance in Mun-ya-wana game reserve was 0.52
±
0.26 (mean
±
SD)
for small rodents and 0.43
±
0.34 for medium rodents; in mixed farms, 0.31
±
0.21 for small
rodents and 0.52
±
0.32 for medium rodents; and in communal lands was 0.26
±
0.23 for
small rodents and 0.36
±
0.24 for medium rodents (Figure 2). Regarding the GLM result
for size-based groups, it revealed significant differences in relative abundances only for
small rodents, between Mun-ya-wana game reserve and the remaining areas (Table S3B,
Supplementary Materials). No significant differences were detected in relative abundances
of medium rodents between areas (Figure 3). Between groups, significant differences were
only found in mixed farms (Table S3B, Supplementary Materials), with medium rodents
being more abundant (0.52
±
0.37) than small-size rodents (0.31
±
0.26) (Figure 3). Based
on these results, the effect of environmental drivers on rodent abundance was evaluated
separately for each of the size-based groups.
Rodent Patchiness
Lloyd’s Index of Patchiness revealed that for every area and size-based group, all abun-
dance values were aggregated (
γ
> 1; Table 2). Both medium and small rodents are heteroge-
neously distributed within the three study areas (Figure 3), demonstrating a heterogeneity
gradient. According to Table 2, we can observe that the highest values for small rodents are
in communal lands, followed by mixed farms and finally, the game reserve. For medium
rodents, there is a greater clustering pattern in the game reserve, followed by communal
lands and mixed farms. With these results, it is possible to state that the abundance patterns
differ between the size-based groups, and within each area.
Animals 2021,11, 2618 9 of 18
Animals 2021, 11, x 10 of 18
Figure 2. Map of the study area showing rodent distributions: small-size rodents are in orange and medium-size rodents
in yellow. The size of each point is equivalent to abundance value, as indicated in the respective legend.
Figure 3. Boxplot of medium and small rodents’ relative abundance in the three management-type
zones monitored: game (mixed) farms, Mun-ya-wana game reserve and communal lands. Based on
the GLM test, * indicates a significant difference between size-based groups in mixed farms (p =
0.011), + indicates a significant difference between Mun-ya-wana game reserve and remaining areas
for small rodents (p = 0.016).
Figure 2.
Map of the study area showing rodent distributions: small-size rodents are in orange and medium-size rodents in
yellow. The size of each point is equivalent to abundance value, as indicated in the respective legend.
Animals 2021, 11, x 10 of 18
Figure 2. Map of the study area showing rodent distributions: small-size rodents are in orange and medium-size rodents
in yellow. The size of each point is equivalent to abundance value, as indicated in the respective legend.
Figure 3. Boxplot of medium and small rodents’ relative abundance in the three management-type
zones monitored: game (mixed) farms, Mun-ya-wana game reserve and communal lands. Based on
the GLM test, * indicates a significant difference between size-based groups in mixed farms (p =
0.011), + indicates a significant difference between Mun-ya-wana game reserve and remaining areas
for small rodents (p = 0.016).
Figure 3.
Boxplot of medium and small rodents’ relative abundance in the three management-type
zones monitored: game (mixed) farms, Mun-ya-wana game reserve and communal lands. Based
on the GLM test, * indicates a significant difference between size-based groups in mixed farms
(
p= 0.011
), + indicates a significant difference between Mun-ya-wana game reserve and remaining
areas for small rodents (p= 0.016).
Animals 2021,11, 2618 10 of 18
Table 2.
Results of Lloyd’s Index of Patchiness per study area and rodent size-based group (small
and medium) (γ).
Area Lloyd’s Index of Patchiness (γ)
Small Medium
Mun-ya-wana game reserve 1.128 1.529
Mixed farms 1.372 1.296
Communal lands 1.528 1.306
3.2. Drivers of Abundance
Capture rate of goats and cows were both correlated with human presence (p= 0.75;
p= 0.76, respectively), and intercorrelated (p= 0.79). Therefore, both former variables were
removed from the analysis.
3.2.1. Small-Size Rodents
The predictive deviance for the BRT model produced for small rodents was 38.8%.
After the simplification of the model, and consequent removal of two variables, predictive
deviance increased to 50%, indicating that the final model explained an important part of
the total variability [
68
]. Distance to houses, wild ungulates, human presence, NDVI, grass
cover and area were identified as the most influential drivers of small rodent abundance
(Figure 4). Small rodents were more abundant in areas far from human settlements,
with lower abundances of wild ungulates and low presence of humans. Regarding the
NDVI, values between 0.29 and 0.35 affect positively the abundance of small rodents.
Semi-open grass cover had the most positive effect on small rodent abundance, as well
as the Mun-ya-wana ecotourism/game reserve. Interactions with area typology within
this model were found for wild ungulates (0.20, interaction size) and NDVI (0.34). As it
is possible to see, in the Figure 5, that the most evident and distinct responses for both
variables occur in Mun-ya-wana game reserve, revealing a clear influence of this area on
wild ungulates and NDVI.
3.2.2. Medium-Size Rodents
For this rodent group, the initial predictive deviance of the model was 40.6%, but, after
the removal of one variable during the model simplification, the predictive deviance
increased to 50%. The set of variables identified as important for this group was very
similar to that described for the previous rodent groups (Figure 4). Medium-size rodents’
abundance was also higher in areas with low abundance of wild ungulates, human presence
and which were far from human settlements. However, this group seems to thrive in
more continuous grass cover and it is positively affected by low values of NDVI (0–0.18).
Contrarily to the small-rodents group, this model did not include the area variable, which
may indicate a lower relevance of area typology in shaping the abundance patterns of
these rodents.
Animals 2021,11, 2618 11 of 18
Animals 2021, 11, x 11 of 17
Figure 4. Variation in abundance (fitted function) predicted from the boosted-regression-tree (BRT) models, for the most
important predictors of rodent abundance (relative importance > 10%). The 95% confidence intervals of each variable are
represented in grey and the red dotted line represents the boundary between the positive and negative effects. Functions
are continuous for all variables, except for grass cover and areagrass cover: C continuous, SC sub-continuous, MC
moderately closed, SO semi-open, O open; area: MF Mixed farms, MW Mun-ya-wana, CL Communal lands. A
common scale is used on the vertical axis for all plots (see Table 1 for variable units).
Figure 5. Interaction between area typology and (a) wild ungulates (b) NDVI. Each line represents the variation of small-
rodent abundance in the respective area (see color legend).
4. Discussion
Rodent abundance, although often an unheeded aspect of conservation management,
is crucial to understand ecosystem functioning, since rodents are primary consumers [16]
and support a large community of predators [17,18], making them a vital link in food-
chain structuring [19]. In our study area, spatial heterogeneity in rodent-abundance pat-
terns appears to be influenced by environmental gradients that are directly linked to var-
ying management priorities across land uses (e.g., ungulate pressure associated with wild
game), which means that these rodent communities, and groups within these communi-
ties, might not benefit uniformly from the increased amount of habitat promoted by the
commercial wildlife industry.
Figure 4.
Variation in abundance (fitted function) predicted from the boosted-regression-tree (BRT) models, for the most
important predictors of rodent abundance (relative importance > 10%). The 95% confidence intervals of each variable are
represented in grey and the red dotted line represents the boundary between the positive and negative effects. Functions
are continuous for all variables, except for grass cover and area–grass cover: C—continuous, SC—sub-continuous, MC—
moderately closed, SO—semi-open, O—open; area: MF—Mixed farms, MW—Mun-ya-wana, CL—Communal lands.
A common scale is used on the vertical axis for all plots (see Table 1for variable units).
Animals 2021, 11, x 12 of 18
Figure 4. Variation in abundance (fitted function) predicted from the boosted-regression-tree (BRT) models, for the most
important predictors of rodent abundance (relative importance > 10%). The 95% confidence intervals of each variable are
represented in grey and the red dotted line represents the boundary between the positive and negative effects. Functions
are continuous for all variables, except for grass cover and area–grass cover: C continuous, SC sub-continuous, MC
moderately closed, SO semi-open, O open; area: MF Mixed farms, MW Mun-ya-wana, CL Communal lands. A
common scale is used on the vertical axis for all plots (see Table 1 for variable units).
Figure 5. Interaction between area typology and (a) wild ungulates (b) NDVI. Each line represents the variation of small-
rodent abundance in the respective area (see color legend).
4. Discussion
Rodent abundance, although often an unheeded aspect of conservation management,
is crucial to understand ecosystem functioning, since rodents are primary consumers [16]
and support a large community of predators [17,18], making them a vital link in food-
chain structuring [19]. In our study area, spatial heterogeneity in rodent-abundance pat-
terns appears to be influenced by environmental gradients that are directly linked to var-
ying management priorities across land uses (e.g., ungulate pressure associated with wild
game), which means that these rodent communities, and groups within these communi-
ties, might not benefit uniformly from the increased amount of habitat promoted by the
commercial wildlife industry.
This image cannot currently be displayed.
Figure 5.
Interaction between area typology and (
a
) wild ungulates (
b
) NDVI. Each line represents the variation of
small-rodent abundance in the respective area (see color legend).
4. Discussion
Rodent abundance, although often an unheeded aspect of conservation management,
is crucial to understand ecosystem functioning, since rodents are primary consumers [16]
and support a large community of predators [
17
,
18
], making them a vital link in food-chain
structuring [
19
]. In our study area, spatial heterogeneity in rodent-abundance patterns
appears to be influenced by environmental gradients that are directly linked to varying
management priorities across land uses (e.g., ungulate pressure associated with wild game),
which means that these rodent communities, and groups within these communities, might
not benefit uniformly from the increased amount of habitat promoted by the commercial
wildlife industry.
Animals 2021,11, 2618 12 of 18
4.1. Context-Specific Responses and Variation Across Management Schemes
Area typology was an important abundance driver for small rodents (thus, just par-
tially supporting our first hypothesis-H1), with higher abundances being estimated for
Mun-ya-wana game reserve than for the remaining areas. Medium-size rodents did not
show any significant differences in their abundance between areas (Figure 3).
The difference in small rodent abundance between areas (Figure 3) is supported by the
interactions of the NDVI and wild ungulates abundance with the area typology (Figure 5).
Overall, small rodent abundance decreased with an increase in wild ungulate abundance,
irrespective of the management scheme, as predicted (H3: ungulate pressure hypothesis).
Similarly, small rodent abundance increased with an increase in NDVI. However, the game
reserve displayed a higher small rodent abundance, relative to the other land uses, and there
is a differential effect of wild ungulates and NDVI on abundance between areas. Within
the game reserve, these variables have a greater influence on this group probably due
to the applied management practices. The greater variation in small rodent abundance
in response to variation in wild ungulate abundance in Mun-wa-wana game reserve
may be driven by the greater vegetation spatial heterogeneity of this area. The game
reserve has a greater habitat heterogeneity compared to the other study areas due to better
conservation derived from its protection status. This habitat heterogeneity results in a
heterogeneous distribution of wild ungulates, owing to differences in habitat preference
or selection (e.g., [
74
]). Thus, this wider variation of ungulate abundance across the
reserve induces a more pronounced response in rodents, leading to the detected typology
effect. Regarding the NDVI, the response may be influenced by the same factor (better
conservation status of native forests-sand forests), which assure a lower disturbance regime,
and thus create conditions to support a more abundant rodent community. However,
the conservation character of some environments may induce the opposite trend in other
taxa. Studies that analyzed the influence of protected areas in the conservation of small
mammals found that these areas exhibit lower abundances compared to neighboring areas,
since their conservation aims is mostly focused on wild ungulates and predators [
37
].
This induces small mammals’ movement to nearby areas, such as farms and agriculture
lands, where they can find more resources (e.g., food) [
9
], and sometimes lower predation
pressure. A study conducted in the same studied game reserve, based on live trapping
measures, revealed a higher abundance of small mammals in adjacent farms and former
cattle farms [
9
]. This pattern seems to be corroborated by our study data, but only for
medium rodents that are less abundant in the more protected area (i.e., Mun-ya-wana game
reserve). Small rodents respond differently, and the pattern may be associated with the
environmental conditions provided by the game reserve, that seem to promote this group
abundance. As mentioned above, the game reserve has a greater habitat heterogeneity
derived from its protection status. This allows the conservation of certain vegetation
patches that do not thrive in the other two areas. In this case, the NDVI values that promote
a higher abundance of small rodents (between 0.28–0.35, Figure 4) correspond to native
forest that exist in greater coverage in the game reserve (i.e., sand forests, Figure 1). Despite
a greater abundance of wild ungulates and possible predators, the presence of these native
habitats establishes more favorable conditions for small rodents. Considering that these
rodents use the landscape on a smaller scale due to their size [
30
], these minor patches of
vegetation create a significant difference in the abundance of this group.
Rodent abundances vary not only between areas (linked to areas specificities, and small
mammals’ requirements), but also show an inter-group variation within areas. The spatial
variation of abundances within-areas seems to be linked to the type of management im-
plemented in each area that affects the vegetation structure and thus may have important
implication in species conservation [
9
]. Lloyd’s Index supports that aggregation levels
differ between size-based groups, since rodents preferentially aggregate in different areas
(medium-size rodents in mixed farms and small rodents in Mun-ya-wana game reserve),
which supports an allopatric distribution of both rodent groups. Furthermore, the highest
abundances of each size-based group occurred in distinct areas (small in Mun-ya-wana,
Animals 2021,11, 2618 13 of 18
medium in mixed farms). Places where rodents occur in a more regular pattern, usually
have better conditions (i.e., higher, and more regularly distributed resources), while sites
where rodent distribution is more aggregated/clustered, indicate a more heterogeneous
distribution of resources [
75
]. Our results show that the lowest values of Lloyd’s Index i.e.,
less patchy distribution, match the highest abundance values for both groups. This pat-
tern is verified for small rodents in the game reserve and medium rodents in mixed farms
(
Table 2
). Area typology influences the patchiness, since conditions will be more or less suit-
able for rodents according to the type of management applied (e.g., reserve and communal
lands; [
76
]). A greater patchiness may lead to isolated populations, causing more sensitive
species to disappear [
75
]. Thus, it is crucial to determine which type of management best
promotes rodent abundance.
4.2. Fine-Scale Environmental Drivers of Rodent Abundance Across the Landscape
Our data also reveals that the abundance of both rodents groups is overall promoted
by grass cover, which supports our second hypothesis (H2). However, the type of grass
cover that enhances rodent abundance varies between groups. While medium-size ro-
dents reached higher densities in continuous grass cover, small rodents are more abundant
in semi-open grass cover. Grass cover, especially continuous layers, can provide protec-
tion against potential predators [
25
,
27
], reducing predation risk, and therefore allowing
medium-size rodents to reach higher abundances. The different results might be associated
to habitat preferences. Small rodents occurred predominantly in forested savanna areas
(ex. Mun-ya-wana game reserve center area), while medium rodents occurred predominantly
in open savanna areas (ex. north and south areas of the game reserve–see
Figures 1and 2
).
The continuous grass cover patches may be more important in these open areas, since they
provide an efficient protection against predators [
24
]. In forested regions (where small-size
rodents seem to be more abundant), grass cover may be less important compared to its
potential cover under better conservation of native forests, which guarantees a greater
diversity of microhabitats and assures a lower disturbance regime, thus creating conditions
to support a more abundant rodent community.
The presence of ungulates (wild or domestic) has been associated with a reduction
of habitat quality for rodents, by decreasing the availability of food and shelter for these
small mammals [
26
,
37
]. This general pattern is reflected in our results, corroborating our
third hypothesis (H3), i.e., species abundance is negatively influenced by the abundance
of ungulates. This negative impact of ungulates may be linked to their impact on vege-
tation [
26
], since higher grazing pressure tend to decrease herbaceous land cover [
9
,
24
].
A study conducted in central Kenya showed an increase in small mammals’ abundance
in the absence of ungulates, revealing the existence of food competition between ungu-
lates and African rodents [
77
]. Although being omnivores, rodents feed mainly on seeds
and grasses [
78
], which are highly depleted when ungulates are present. Furthermore,
the ungulates trampling impacts on small mammals are also a possible explanation for
this negative influence, since the soil compaction due to ungulates movements hampers
burrows maintenance [
26
,
79
]. Other studies highlighted the impact of a reduction of the
herbaceous layer, as it decreases refuge availability and increases predation risk by improv-
ing small mammals’ detection by predators [
31
,
80
82
]. Therefore, these two-fold effects
(decrease in food and shelter availability), acting in isolation or in synergy, may be the
underlying processes that constrain rodent abundance in the presence of ungulates.
The distance to human settlements and human presence are also two factors that we
identified as having a negative effect on both rodent groups’ abundance, which corroborates
our fourth hypothesis (H4). Rodents revealed lower abundances in areas closer to houses,
especially in communal lands, the area with the highest density of settlements (while
houses are almost absent from the other two areas). Thus, the effect of this variable
cannot be linearly interpreted as a distance to the nearest house, but probably as a distance
to the communal lands themselves, as both groups’ abundances are low in this area
(see Figure 2). The average abundance values confirm that the least preferred zone for
Animals 2021,11, 2618 14 of 18
both groups of rodents are the communal lands, as it is the place where the lowest values
of abundance were estimated (Figure 3). However, these negative effects of anthropic
disturbance may also be linked to the presence of domestic animals (livestock and goats),
that occur concomitantly with settlements, and that also negatively affect rodent abundance,
due to the same processes described above for wild ungulates [26].
This different patterns between rodent groups, as well as the variation of the drivers
and their importance on the abundance variation of both species, supports the division of
our dataset into size-based groups. This means that not only rodents should be taken into
consideration, but also heterogeneity within rodent communities, which is important given
their different functional roles (e.g., as prey, as consumers–granivory and insectivory–and
seed dispersers).
Although we acknowledge some limitations of this approach, based on footprint size,
we have tried to minimize this by sampling only in seasons where the misclassification
effect of juveniles’ presence is negligible. Nevertheless, this time-limited sampling hampers
the validity of extrapolating results. Interpretation of the overall (annual) pattern of
abundances’ spatial distribution must be done with care. Rodents numerically respond
to variations in rainfall and food availability, which vary throughout the year. Thus,
by sampling in only one season, we may have gotten a partial image of the processes
shaping rodent abundance. However, in terms of wildlife management and conservation,
it is always better to have a partial understanding of the ecological patterns and processes
than having none.
5. Conclusions
Our study contributes to the current view that landscape-management options shape
the ecological patterns of species, by modifying the composition and structure of habitats.
Moreover, responses to land composition are species/group-specific. These results high-
light the need to expand conservation actions beyond protected areas. For biodiversity
conservation to succeed in these habitat mosaics, landscape-level policies and management
are required to integrate both protected and managed areas, as the later also host a large
number of species, acting as a metapopulation source-sink. We encourage future work that
evaluates the transferability of our findings to other southern African multi-use landscapes.
Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10
.3390/ani11092618/s1. PART A—Comparison of ink-tracking tunnels with live-trapping for track
index validation. Figure S1A: example of forefeet tracks of the three functional groups;
Figure S2A
:
track measures in mm from the three functional groups; Table S1A: Dunn’s test for the four track
measurements between groups; Table S2A: list of species occurring or possibly occurring in the
region. PART B—Figures and tables additional to the manuscript. Figure S1B: ink-tracking tunnel
scheme; Figure S2B: scheme of the method used to measure the 100 random tracks; Table S1B:
categories used to describe the abundance of wild ungulates detected during the camera-trapping
campaigns;
Table S2B
: percentage of rodent detection in each area per functional group; Table S3B:
linear regression models between size-based groups and areas. References [
83
89
] are cited in the
Supplementary Materials.
Author Contributions:
Conceptualization, B.C.A., G.C.-S., L.H.S., M.S.-R.; Methodology, B.C.A.,
B.P.R., G.C.-S., L.M.R.; Formal Analysis, B.C.A., T.A.M.; Investigation, B.C.A., B.P.R.; Resources,
L.H.S.; Data Curation, B.C.A., B.P.R., G.C.-S.; Writing—Original Draft Preparation, B.C.A.; Writing—
Review & Editing, L.M.R., G.C.-S., M.S.-R.; Visualization, B.C.A.; Supervision, L.M.R., M.S.-R.; Project
Administration, G.C.-S., M.S.-R., L.H.S.; Funding Acquisition, L.H.S. All authors have read and
agreed to the published version of the manuscript.
Funding:
This research was funded by FCT/MCTES, through national funds, and the co-funding by the
FEDER, within the PT2020 Partnership Agreement and Compete 2020 (cE3c: UIDB/00329/2020), and by
the South African National Research Foundation, South Africa (UID 107099&115040). TAM thanks
partial support by CEAUL (funded by FCT-Fundação para a Ciência e a Tecnologia, Portugal, through
the project UIDB/00006/2020).
Animals 2021,11, 2618 15 of 18
Institutional Review Board Statement:
The study was conducted according to the guidelines of the
Declaration of Helsinki, and approved by the Research Ethics Committee of University of Venda
(protocol SMNS/17/Z00/04/0905 from 11/05/2017 and 13/11/2018), and under the permit number
OP 1391/2018 from the Ezemvelo KZN Wildlife.
Data Availability Statement:
The datasets generated during the current study are available from the
corresponding author on reasonable request.
Acknowledgments:
We thank the Mun-ya-wana Conservancy’s manager and staff, ranch owners
and tribal authorities for granting permission to conduct and for supporting our study. We are
grateful to everyone who assisted with fieldwork.
Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript;
or in the decision to publish the results.
References
1.
McCusker, B. Land use and cover change as an indicator of transformation on recently redistributed farms in Limpopo Province,
South Africa. Hum. Ecol. 2004,32, 49–75. [CrossRef]
2.
Smith, N.; Wilson, S.L. Changing Land Use Trends in the Thicket Biome: Pastoralism to Game Farming; Blackwell Publishing: Oxford,
UK, 2002.
3.
Maitima, J.M.; Mugatha, S.M.; Reid, R.S.; Gachimbi, L.N.; Majule, A.; Lyaruu, H.; Pomery, D.; Mathai, S.; Mugisha, S. The linkages
between land use change, land degradation and biodiversity across East Africa. Afr. J. Environ. Sci. Technol. 2009,2, 310–325.
4.
Bender, D.J.; Contreras, T.A.; Fahrig, L. Habitat loss and population decline: A meta-analysis of the patch size effect. Ecology
1998
,
79, 517–533. [CrossRef]
5.
Pitman, R.T.; Fattebert, J.; Williams, S.T.; Williams, K.S.; Hill, R.A.; Hunter, L.T.B.; Slotow, R.; Balme, G.A. The conservation costs
of game ranching. Conserv. Lett. 2016,10, 403–413. [CrossRef]
6.
Lindsey, P.A.; Romanach, S.S.; Davies-Mostert, H.T. The importance of conservancies for enhancing the value of game ranch land
for large mammal conservation in southern Africa. J. Zool. 2009,277, 99–105. [CrossRef]
7.
Child, B.A. The economics and governance of the wildlife economy in drylands in Southern Africa. In Natural Resources, Tourism
and Community Livelihoods in Southern Africa; Routledge: Oxfordshire, UK, 2019; pp. 161–175.
8.
Kerley, G.I.; Pressey, R.L.; Cowling, R.M.; Boshoff, A.F.; Sims-Castley, R. Options for the conservation of large and medium-sized
mammals in the Cape Floristic Region hotspot, South Africa. Biol. Conserv. 2003,112, 169–190. [CrossRef]
9.
Rautenbach, A. Patterns and processes of rodent and shrew assemblages in the Savanna Biome of KwaZulu-Natal, South Africa.
Ph.D. Thesis, University of KwaZulu-Natal, KwaZulu-Natal, South Africa, 2013.
10.
Di Minin, E.; Macmillan, D.C.; Goodman, P.S.; Escott, B.; Slotow, R.; Moilanen, A. Conservation businesses and conservation
planning in a biological diversity hotspot. Conserv. Biol. 2013,27, 808–820. [CrossRef]
11.
Jones, B.T.; Stolton, S.U.E.; Dudley, N. Private protected areas in East and southern Africa: Contributing to biodiversity
conservation and rural development. Parks 2005,15, 67–76.
12.
Curveira-Santos, G.; Sutherland, C.; Santos-Reis, M.; Swanepoel, L.H. Responses of carnivore assemblages to decentralized
conservation approaches in a South African landscape. J. Appl. Ecol. 2021,58, 92–103. [CrossRef]
13.
Makundi, R.H.; Oguge, N.O.; Mwanjabe, P.S. Rodent pest management in East Africa—an ecological approach. In Book Ecologically-
Based Rodent Management, 1st ed.; Singleton, G., Hinds, L., Leirs, H., Zhang, Z., Eds.; Australian Center for International
Agricultural Research: Camberra, Australia, 1999; pp. 460–476.
14.
Mahlaba, T.A.A.; Monadjem, A.; McCleery, R.; Belmain, S.R. Domestic cats and dogs create a landscape of fear for pest rodents
around rural homesteads. PLoS ONE 2017,12, e0171593. [CrossRef]
15.
Thibault, M.; Brescia, F.; Jourdan, H.; Vidal, E. Invasive rodents, an overlooked threat for skinks in a tropical island hotspot of
biodiversity. N. Z. J. Ecol. 2017,41, 74–83. [CrossRef]
16.
Avenant, N.L.; Cavallini, P. Correlating rodent community structure with ecological integrity, Tussen-die-Riviere Nature Reserve,
Free State province, South Africa. Integr. Zool. 2007,2, 212–219. [CrossRef]
17. Andersson, M.; Erlinge, S. Influence of predation on rodent populations. Oikos 1977,29, 591–597. [CrossRef]
18.
Jonsson, P.; Koskela, E.; Mappes, T. Does risk of predation by mammalian predators affect the spacing behaviour of rodents?
Two large-scale experiments. Oecologia 2000,122, 487–492. [CrossRef]
19.
Cameron, G.N.; Scheel, D. Getting warmer: Effect of global climate change on distribution of rodents in Texas. J. Mammal.
2001
,
82, 652–680. [CrossRef]
20.
Panzacchi, M.; Linnell, J.D.; Melis, C.; Odden, M.; Odden, J.; Gorini, L.; Andersen, R. Effect of land-use on small mammal
abundance and diversity in a forest–farmland mosaic landscape in south-eastern Norway. For. Ecol. Manag.
2010
,259, 1536–1545.
[CrossRef]
21.
Williams, S.E.; Marsh, H.; Winter, J. Spatial scale, species diversity, and habitat structure: Small mammals in Australian tropical
rain forest. Ecology 2002,83, 1317–1329. [CrossRef]
Animals 2021,11, 2618 16 of 18
22.
Layme, V.M.G.; Lima, A.P.; Magnusson, W.E. Effects of fire, food availability and vegetation on the distribution of the rodent
Bolomys lasiurus in an Amazonian savanna. J. Trop. Ecol. 2004,20, 183–187. [CrossRef]
23.
Holland, G.J.; Bennett, A.F. Differing responses to landscape change: Implications for small mammal assemblages in forest
fragments. Biodivers. Conserv. 2009,18, 2997–3016. [CrossRef]
24.
Hoffmann, A.; Zeller, U. Influence of variations in land use intensity on species diversity and abundance of small mammals in the
Nama Karoo, Namibia. Belg. J. Zool. 2005,135, 91–96.
25.
Bond, W.; Ferguson, M.; Forsyth, G. Small mammals and habitat structure along altitudinal gradients in the southern Cape
mountains. Afr. Zool. 1980,15, 34–43. [CrossRef]
26.
Keesing, F. Impacts of ungulates on the demography and diversity of small mammals in central Kenya. Oecologia
1998
,116,
381–389. [CrossRef]
27.
Delcros, G.; Taylor, P.J.; Schoeman, M.C. Ecological correlates of small mammal assemblage structure at different spatial scales in
the savannah biome of South Africa. Mammalia 2015,79, 1–14. [CrossRef]
28.
Massawe, A.W.; Leirs, H.; Rwamugira, W.P.; Makundi, R.H. Effect of land preparation methods on spatial distribution of rodents
in crop fields. In Book Rats, Mice and People: Rodent Biology and Management, 1st ed.; Australian Centre for International Agricultural
Research: Camberra, Australia, 2003.
29.
Mortelliti, A.; Amori, G.; Annesi, F.; Boitani, L. Testing for the relative contribution of patch neighborhood, patch internal
structure, and presence of predators and competitor species in determining distribution patterns of rodents in a fragmented
landscape. Can. J. Zool. 2009,87, 662–670. [CrossRef]
30.
Sutherland, G.D.; Harestad, A.S.; Price, K.; Lertzman, K.P. Scaling of natal dispersal distances in terrestrial birds and mammals.
Conserv. Ecol. 2000,4, 16. [CrossRef]
31.
Pacifici, M.; Rondinini, C.; Rhodes, J.R.; Burbidge, A.A.; Cristiano, A.; Watson, J.E.M.; Woinarski, J.C.Z.; Di Marco, M. Global
correlates of range contractions and expansions in terrestrial mammals. Nat. Commun. 2020,11, 2840. [CrossRef] [PubMed]
32.
Roques, K.G.; O’connor, T.G.; Watkinson, A.R. Dynamics of shrub encroachment in an African savanna: Relative influences of
fire, herbivory, rainfall and density dependence. J. Appl. Ecol. 2001,38, 268–280. [CrossRef]
33.
Blaum, N.; Rossmanith, E.; Jeltsch, F. Land use affects rodent communities in Kalahari savannah rangelands. Afr. J. Ecol.
2007
,45,
189–195. [CrossRef]
34.
Hurst, Z.M.; McCleery, R.A.; Collier, B.A.; Silvy, N.J.; Taylor, P.J.; Monadjem, A. Linking changes in small mammal communities
to ecosystem functions in an agricultural landscape. Mamm. Biol. 2014,79, 17–23. [CrossRef]
35.
Mulungu, L.S.; Mahlaba, T.A.; Massawe, A.W.; Kennis, J.; Crauwels, D.; Eiseb, S.; Monadjem, A.; Makundi, R.H.; Katakweba,
A.A.S.; Leirs, H.; et al. Dietary differences of the multimammate mouse, Mastomys natalensis (Smith, 1834), across different
habitats and seasons in Tanzania and Swaziland. Wildl. Res. 2011,38, 640–646. [CrossRef]
36.
Dalmagro, A.D.; Vieira, E.M. Patterns of habitat utilization of small rodents in an area of Araucaria forest in Southern Brazil.
Austral Ecol. 2005,30, 353–362. [CrossRef]
37.
Caro, T.M. Species richness and abundance of small mammals inside and outside an African national park. Biol. Conserv.
2001
,98,
251–257. [CrossRef]
38. Monadjem, A. Habitat preferences and biomasses of small mammals in Swaziland. Afr. J. Ecol. 1997,35, 64–72. [CrossRef]
39.
Dunstan, C.; Fox, B. The effects of fragmentation and disturbance of rainforest on ground-dwelling small mammals on the
Robertson Plateau, New South Wales, Australia. J. Biogeogr. 1996,23, 187–201. [CrossRef]
40.
Balme, G.A.; Slotow, R.O.B.; Hunter, L.T. Edge effects and the impact of non-protected areas in carnivore conservation: Leopards
in the Phinda–Mkhuze Complex, South Africa. Anim. Conserv. 2010,13, 315–323. [CrossRef]
41.
Krug, W. Private supply of protected land in southern Africa: A review of markets, approaches, barriers and issues. In Workshop
Paper, World Bank/OECD International Workshop on Market Creation for Biodiversity Products and Services, Paris; OECD: Paris, France,
2001; Volume 25.
42.
King, C.M.; Edgar, R.L. Techniques for trapping and tracking stoats (Mustela erminea); a review, and a new system. N. Z. J. Zool.
1977,4, 193–212. [CrossRef]
43.
Fick, S.E.; Hijmans, R.J. WorldClim 2: New 1km spatial resolution climate surfaces for global land areas. Int. J. Climatol.
2017
,37,
4302–4315. [CrossRef]
44.
Harris, I.; Jones, P.D.; Osborn, T.J.; Lister, D.H. Updated high-resolution grids of monthly climatic observations—The CRU TS3.10
Dataset. Int. J. Climatol. 2014,34, 623–642. [CrossRef]
45. ASTER GDEM Data Version 2. Available online: https://search.earthdata.nasa.gov/search/ (accessed on 27 April 2019).
46.
Department of Environmental Affairs (DEA). 2013-14 SA National Land-Cover—Broad Parent Classes. Available online:
http://www.sasdi.net/metaview.aspx?uuid=20c6336b75031c73041ec8b88372c3b0 (accessed on 27 April 2019).
47.
Glennon, M.J.; Porter, W.F.; Demers, C.L. An alternative field technique for estimating diversity of small-mammal populations.
J. Mammal. 2002,83, 734–742. [CrossRef]
48.
Hughes, J.J.; Ward, D.; Perrin, M.R. Predation risk and competition affect habitat selection and activity of Namib Desert gerbils.
Ecology 1994,75, 1397–1405. [CrossRef]
49.
Wilkinson, E.B.; Branch, L.C.; Miller, D.L.; Gore, J.A. Use of track tubes to detect changes in abundance of beach mice. J. Mammal.
2012,93, 791–798. [CrossRef]
Animals 2021,11, 2618 17 of 18
50.
Palma, A.R.T.; Gurgel-Gonçalves, R. Morphometric identification of small mammal footprints from ink tracking tunnels in the
Brazilian Cerrado. Rev. Bras. Zool. 2007,24, 333–343. [CrossRef]
51.
Greene, D.U.; Oddy, D.M.; Gore, J.A.; Gillikin, M.N.; Evans, E.; Gann, S.L.; Leone, E.H. Differentiating footprints of sympatric
rodents in coastal dune communities: Implications for imperiled beach mice. J. Fish Wildl. Manag. 2018,9, 593–601. [CrossRef]
52. Delany, M.J. Ecology of small rodents in Africa. Mammal. Rev. 1986,16, 1–41. [CrossRef]
53.
QGIS Development Team. QGIS Geographic Information System. Open Source Geospatial Foundation Project. Available online:
http://qgis.osgeo.org (accessed on 12 March 2019).
54. Edwards, E. A broad-scale structural classification of vegetation for practical purposes. Bothalia 1983,14, 705–712. [CrossRef]
55.
FAO Country Analysis (GIEWS). Available online: http://www.fao.org/giews/countrybrief/country.jsp?code=ZAF&lang=ru&
fbclid=IwAR3fdjZN3f-NmB1RDGa5Z7Q01OdiKy4ndt3xuFmGrWsHHiYisK_pWEdY30Y (accessed on 24 August 2021).
56. Earth Explorer (Landsat 8 Images). Available online: https://earthexplorer.usgs.gov/ (accessed on 18 April 2019).
57.
Freitas, S.R.; Cerqueira, R.; Vieira, M.V. A device and standard variables to describe microhabitat structure of small mammals
based on plant cover. Braz. J. Biol. 2002,62, 795–800. [CrossRef]
58.
Martin, G.H.G.; Dickinson, N.M. Small mammal abundance in relation to microhabitat in a dry sub-humid grassland in Kenya.
Afr. J. Ecol. 1985,23, 223–234. [CrossRef]
59. Dueser, R.D.; Shugart, H.H., Jr. Microhabitats in a forest-floor small mammal fauna. Ecology 1978,59, 89–98. [CrossRef]
60.
Monadjem, A. Geographic distribution patterns of small mammals in Swaziland in relation to abiotic factors and human land-use
activity. Biodivers. Conserv. 1999,8, 223–237. [CrossRef]
61.
Christie, J.E.; Wilson, P.R.; Taylor, R.H.; Elliott, G. How elevation affects ship rat (Rattus rattus) capture patterns, Mt Misery,
New Zealand. N. Z. J. Ecol. 2017,41, 113–119. [CrossRef]
62.
Miranda, C.S.; Gamarra, R.M.; Mioto, C.L.; Silva, N.M.; Conceição Filho, A.P.; Pott, A. Analysis of the landscape complexity and
heterogeneity of the Pantanal wetland. Braz. J. Biol. 2018,78, 318–327. [CrossRef]
63. Lloyd, M. Mean crowding’. J. Anim. Ecol. 1967,36, 1–30. [CrossRef]
64.
Revelle, M.W. Package ‘psych’: The Comprehensive R Archive Network. Version 1.9.12. 2015. Available online: https:
//cran.r-project.org/web/packages/psych/psych.pdf (accessed on 16 April 2019).
65.
Zuur, A.; Ieno, E.N.; Walker, N.; Saveliev, A.A.; Smith, G.M. Mixed Effects Models and Extensions in Ecology with R; Springer Science
& Business Media: New York, NY, USA, 2009.
66.
Filipe, A.F.; Cowx, I.G.; Collares-Pereira, M.J. Spatial modelling of freshwater fish in semi-arid river systems: A tool for
conservation. River Res. Appl. 2002,18, 123–136. [CrossRef]
67. Package ‘gbm’. Available online: https://cran.r-project.org/web/packages/gbm/index.html (accessed on 16 April 2019).
68.
R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria;
Available online: https://www.R-project.org/ (accessed on 12 March 2019).
69.
RStudio Team. RStudio: Integrated Development for R; RStudio, Inc.: Boston, MA, USA. Available online: http://www.rstudio.com
(accessed on 12 March 2019).
70.
Elith, J.; Leathwick, J.R.; Hastie, T. A working guide to boosted regression trees. J. Anim. Ecol.
2008
,77, 802–813. [CrossRef]
[PubMed]
71. De’Ath, G. Boosted trees for ecological modeling and prediction. Ecology 2007,88, 243–251. [CrossRef]
72.
Carslaw, D.C.; Taylor, P.J. Analysis of air pollution data at a mixed source location using boosted regression trees. Atmos. Environ.
2009,43, 3563–3570. [CrossRef]
73.
Williams, G.J.; Aeby, G.S.; Cowie, R.O.; Davy, S.K. Predictive modeling of coral disease distribution within a reef system. PLoS
ONE 2010,5, e9264. [CrossRef]
74.
Colin, B.; Clifford, S.; Wu, P.P.; Rathmanner, S.; Mengersen, K. Using boosted regression trees and remotely sensed data to drive
decision-making. Open J. Stat. 2017,7, 859–875. [CrossRef]
75.
Abeare, S. Comparisons of Boosted Regression Tree, GLM and GAM Performance in the Standardization of Yellowfin Tuna
Catch-Rate Data from the Gulf of Mexico. Master’s Thesis, Louisiana State University, Baton Rouge, LA, USA, 2009.
76.
Averbeck, C.; Apio, A.; Plath, M.; Wronski, T. Environmental parameters and anthropogenic effects predicting the spatial
distribution of wild ungulates in the Akagera savannah ecosystem. Afr. J. Ecol. 2009,47, 756–766. [CrossRef]
77.
Swihart, R.K.; Atwood, T.C.; Goheen, J.R.; Scheiman, D.M.; Munroe, K.E.; Gehring, T.M. Patch occupancy of North American
mammals: Is patchiness in the eye of the beholder? J. Biogeogr. 2003,30, 1259–1279. [CrossRef]
78.
Lagesse, J.V.; Thondhlana, G. The effect of land-use on small mammal diversity inside and outside the Great Fish River Nature
Reserve, Eastern Cape, South Africa. J. Arid Environ. 2016,130, 76–83. [CrossRef]
79. Keesing, F. Cryptic consumers and the ecology of an African savanna. BioScience 2000,50, 205–215. [CrossRef]
80.
Iwuala, M.O.; Braide, E.I.; Maduka, N. Observations on the food habits of some African rodents. Rev. Biol. Trop.
1979
,28, 227–236.
81.
Torre, I.; Díaz, M.; Martínez-Padilla, J.; Bonal, R.; Vinuela, J.; Fargallo, J.A. Cattle grazing, raptor abundance and small mammal
communities in Mediterranean grasslands. Basic Appl. Ecol. 2007,8, 565–575. [CrossRef]
82.
Tew, T.E.; Macdonald, D.W. The effects of harvest on arable wood mice Apodemus sylvaticus. Biol. Conserv.
1993
,65, 279–283.
[CrossRef]
83.
Pounds, C.J. Niche overlap in sympatric populations of stoats (Mustela erminea) and weasels (M. nivalis) in Northeast Scotland.
Ph.D. Thesis, University of Aberdeen, Aberdeen, UK, 1981.
Animals 2021,11, 2618 18 of 18
84.
Carrilho, M.; Teixeira, D.; Santos-Reis, M.; Rosalino, L.M. Small mammal abundance in Mediterranean Eucalyptus plantations:
How shrub cover can really make a difference. For. Ecol. Manag. 2017,391, 256–263. [CrossRef]
85.
Razali, N.M.; Wah, Y.B. Power comparisons of Shapiro-Wilk, Kolmogorov-Smirnov, Lilliefors and Anderson-Darling tests. J. Stat.
Model. Anal. 2011,2, 21–33.
86.
Ali, Z.; Bhaskar, S.B. Basic statistical tools in research and data analysis. Indian J. Anaesth.
2016
,60, 662–669. [CrossRef] [PubMed]
87.
Ostertagová, E.; Ostertag, O.; Kováˇc, J. Methodology and Application of the Kruskal-Wallis Test. Appl. Mech. Mater.
2014
,611,
115–120. [CrossRef]
88.
Kingdon, J.; Happold, D.; Butynski, T.; Hoffmann, M.; Happold, M.; Kalina, J. Mammals of Africa; A&C Black: London, UK, 2013;
Volume 1.
89.
Havlicek, L.L.; Peterson, N.L. Robustness of the Pearson Correlation against Violations of Assumptions. Percept. Mot. Skills
1976
,
43, 1319–1334. [CrossRef]
... Nevertheless, not all human-derived environmental changes have such dramatic outcomes, with effects ranging from population declines due to habitat loss/degradation (Wiegand et al. 2005), resource depletion (Chaikina and Ruckstuhl 2006), or human-wildlife conflict (Nyhus 2016) to behavioral changes in how species use the landscape (e.g., landscapes of coexistence; Oriol-Cotterill et al. 2015) or their 24-h activity patterns (Gaynor et al. 2018). Furthermore, anthropic disturbances affect a wide range of mammals having distinct functional roles in the ecosystems in which they live, from small granivores (e.g., Afonso et al. 2021) to large carnivores (e.g., Nakamura et al. 2021). However, the overall scale of impact greatly depends on a species' life history, population demography, and biology (i.e., distribution, density, and sensitivity to environmental change) (e.g., Mendes et al. 2017;Piña et al. 2019). ...
Article
Humans have negatively impacted most ecosystems on Earth, altering how species use habitats and resources available to them, but also their circadian rhythms. Among many factors affecting animal activity patterns, conversion of native habitat into production forests and hunting are critical, and their impacts may be exacerbated by seasonal weather conditions. Both these factors are issues in Mediterranean Europe, especially in Portugal. Nevertheless, their impact on native species behavior remains largely unknown, even for generalist species often resilient to human disturbance such as wild boar (Sus scrofa). Therefore, we assessed temporal activity patterns of wild boar in Portugal to explore differences between habitats (native forest versus Eucalyptus plantations), hunting and non-hunting periods, or between seasons. We found that wild boars were primarily nocturnal, with greater nighttime activity more evident during the dry season and within plantations. Furthermore, they appeared to avoid dusk during the hunting season. Together, these patterns indicate that even a species typically resilient to disturbance and environmental change, such as wild boar, tends to avoid adverse weather conditions and disturbance factors, namely high daytime temperatures in summer and probability of encountering forestry workers or hunters. Accordingly, other more sensitive or resource-demanding mammals may also be affected, so evaluating the impacts of the studied anthropic drivers on other mammals should be prioritized to provide information for sustainable plantation management and hunting.
... Variable detection rates for various species in the riparian and grassland landscapes suggest that most mesocarnivore species in the Telperion Nature Reserve employ spatial partitioning rather than temporal or trophic niche partitioning as a coexistence mechanism (Donadio & Buskirk, 2006). With its mixture of riparian and grassland vegetation types, the heterogeneity of the nature reserve, and the related diversity of available prey species (Afonso et al., 2021), was likely the main contributor to the rate of species detected, despite its relatively small size. Landscape heterogeneity has been shown to facilitate niche partitioning and enable coexistence (Fisher et al., 2013) and is particularly important in increasingly humandominated landscapes (Manlick et al., 2020). ...
Article
Despite the diversity of mesocarnivores and the broad geographic ranges of these species, our understanding of their behaviour and ecology at multi-species and community levels is limited. Our study was conducted between April and mid-July 2015 and used data collected over 105 days from 39 camera traps to quantify activity patterns of sympatric meso-carnivores in riparian and grassland habitats of Telperion Nature Reserve, South Africa. A total of 13 mesocarnivore species were detected within this relatively small (~7350 ha) reserve. Sufficient records (≥10 records) were obtained for rusty-spotted genet (Genetta maculata), black-backed jackal (Canis mesomelas), otter species (African clawless otter, Aonyx capensis, and spotted-necked otter, Hydrictis maculicollis), serval (Leptailurus serval), slender mongoose (Galerella sanguinea), yellow mongoose (Cynictis penicillata) and marsh mongoose (Atilax paludinosus). Generalized linear models were used to investigate whether species ID, temperature, vegetation characteristics or moon phase best predicted temporal activity. To assess which species had the highest potential for competitive interaction, we also quantified the coefficient of activity overlap. Our results show that species ID and temperature were the best predictors of mesocarnivore activity. Slender and yellow mongooses displayed the highest coefficient of activity overlap (0.90), followed by marsh mongoose and rusty-spotted genet (0.80), and serval and rusty-spotted genet (0.79). These species are likely to have the highest potential for competitive interactions, but preferences for different vegetation characteristics and variations in the estimated relative abundance may point to coexistence through spatial and fine-scale temporal partitioning. The other species exhibited lower coefficients of activity overlap with each other, suggesting they may coexist through temporal partitioning of resources.
Article
Full-text available
Understanding changes in species distributions is essential to disentangle the mechanisms that drive their responses to anthropogenic habitat modification. Here we analyse the past (1970s) and current (2017) distribution of 204 species of terrestrial non-volant mammals to identify drivers of recent contraction and expansion in their range. We find 106 species lost part of their past range, and 40 of them declined by >50%. The key correlates of this contraction are large body mass, increase in air temperature, loss of natural land, and high human population density. At the same time, 44 species have some expansion in their range, which correlates with small body size, generalist diet, and high reproductive rates. Our findings clearly show that human activity and life history interact to influence range changes in mammals. While the former plays a major role in determining contraction in species’ distribution, the latter is important for both contraction and expansion. Understanding why many species ranges are contracting while others are stable or expanding is important to inform conservation in an increasingly human-modified world. Here, Pacifici and colleagues investigate changes in the ranges of 204 mammals, showing that human factors mostly explain range contractions while life history explains both contraction and expansion.
Article
Full-text available
Identifying techniques for conducting frequent, effective, and inexpensive monitoring of small mammals can be challenging. Traditional approaches such as livetrapping can be laborious, expensive, detrimental to animal health, and ineffective. Passive approaches such as tracking (e.g., from tracks on the ground or footprints collected at a tracking station) have been shown to lessen those burdens, but a problem with tracking, particularly for rodents, is the uncertainty in identifying species from footprints. To address the need for a more accurate method of identifying small mammal tracks, we measured footprints from live-captured rodents and developed a classification tree for distinguishing between subspecies and species using footprint widths treated as having known or unknown identification. We captured rodents within or near the coastal dunes of Florida and Alabama with a focus on areas occupied by threatened and endangered beach mice Peromyscus polionotus subspp., whose populations warrant regular monitoring but whose tracks are not easily distinguished from those of some sympatric species. We measured 6,996 front and hind footprints from 540 individuals across eight species. The overall accuracy of our classification tree was 82.6% and we achieved this using only the front footprint width. Footprint width cutoffs for species identification were, 5.5 mm for house mice Mus musculus, 5.5–6.7 mm for beach mice, and 6.7–8.3 mm for cotton mice Peromyscus gossypinus. We were most successful in confirming the identity of beach mice: we correctly classified approximately 94% of beach mice, while we misclassified fewer than 6% as house mice and fewer than 1% as cotton mice. When we input a beach mouse individual into the classification tree as of an unknown species, we correctly identified 78.1% of individuals as beach mice from their tracks, and most incorrect identifications were of house mouse tracks. Our study demonstrates that researchers can identify sympatric rodent species in coastal dune communities from tracks using quantitative classification based on footprint width. Accurate identification of beach mice or other imperiled species from tracks has important management implications. Not only can wildlife managers determine the presence of a species accurately, but they can monitor populations with considerably less effort than livetrapping requires. Although our study was specific to coastal dune communities, our methods could be adapted for the creation of a classification tree for identifying tracks from suites of species in other areas.
Article
Full-text available
This is the first report on analysis of habitat complexity and heterogeneity of the Pantanal wetland. The Pantanal encompasses a peculiar mosaic of environments, being important to evaluate and monitor this area concerning conservation of biodiversity. Our objective was to indirectly measure the habitat complexity and heterogeneity of the mosaic forming the sub-regions of the Pantanal, by means of remote sensing. We obtained free images of Normalized Difference Vegetation Index (NDVI) from the sensor MODIS and calculated the mean value (complexity) and standard deviation (heterogeneity) for each sub-region in the years 2000, 2008 and 2015. The sub-regions of Poconé, Canoeira, Paraguai and Aquidauana presented the highest values of complexity (mean NDVI), between 0.69 and 0.64 in the evaluated years. The highest horizontal heterogeneity (NDVI standard deviation) was observed in the sub-region of Tuiuiú, with values of 0.19 in the years 2000 and 2015, and 0.21 in the year 2008. We concluded that the use of NDVI to estimate landscape parameters is an efficient tool for assessment and monitoring of the complexity and heterogeneity of the Pantanal habitats, applicable in other regions.
Article
Full-text available
We created a new dataset of spatially interpolated monthly climate data for global land areas at a very high spatial resolution (approximately 1 km 2). We included monthly temperature (minimum, maximum and average), precipitation, solar radiation, vapour pressure and wind speed, aggregated across a target temporal range of 1970–2000, using data from between 9000 and 60 000 weather stations. Weather station data were interpolated using thin-plate splines with covariates including elevation, distance to the coast and three satellite-derived covariates: maximum and minimum land surface temperature as well as cloud cover, obtained with the MODIS satellite platform. Interpolation was done for 23 regions of varying size depending on station density. Satellite data improved prediction accuracy for temperature variables 5–15% (0.07–0.17 ∘ C), particularly for areas with a low station density, although prediction error remained high in such regions for all climate variables. Contributions of satellite covariates were mostly negligible for the other variables, although their importance varied by region. In contrast to the common approach to use a single model formulation for the entire world, we constructed the final product by selecting the best performing model for each region and variable. Global cross-validation correlations were ≥ 0.99 for temperature and humidity, 0.86 for precipitation and 0.76 for wind speed. The fact that most of our climate surface estimates were only marginally improved by use of satellite covariates highlights the importance having a dense, high-quality network of climate station data.
Article
Conservation efforts in South Africa play out across multi‐use landscapes where formal protected areas coexist with private wildlife business (ecotourism and/or hunting) in a human‐dominated matrix. Despite the persistence of highly diverse carnivore guilds, management idiosyncrasies are often orientated towards charismatic large predators and assemblage‐level patterns remain largely unexplored. We conducted an extensive camera‐trap survey in a natural quasi‐experimental setting in KwaZulu‐Natal, South Africa. We sampled across a protection gradient characterized by a provincial protected area (highest and formal protection status), a private ecotourism reserve, game ranches and traditional communal areas (lowest protected status). We evaluated assemblage‐level and species‐specific responses of free‐ranging carnivores to the varying management contexts and associated environmental gradients. Despite similar assemblage composition between management contexts, site‐scale carnivore richness and occupancy rates were greater in the formal protected area than adjacent private reserve and game ranches. Carnivore occupancy was more similar between these private wildlife areas, although putative problem species were more common in the private reserve, and contrasted with depauperate assemblages in least protected communal lands. Variation in carnivore occupancy probabilities was largely driven by land use contexts, that is, the level and nature of protection, relative to underlying fine‐scale landscape attributes (e.g. distance to conservation fences) or apex predator populations. Synthesis and applications. Our findings provide convincing empirical support for the added value of multi‐tenure conservation estates augmenting and connecting South Africa's protected areas. However, our emphasis on free‐ranging carnivores exemplifies the importance of maintaining areas under long‐term formal protection and the risks with viewing lucrative wildlife business as a conservation panacea. We suggest that unmanaged carnivore species be the formal components of carnivore reintroduction and recovery programmes to better gauge the complementary conservation role of South Africa's private land. Our findings provide convincing empirical support for the added value of multi‐tenure conservation estates augmenting and connecting South Africa's protected areas. However, our emphasis on free‐ranging carnivores exemplifies the importance of maintaining areas under long‐term formal protection and the risks with viewing lucrative wildlife business as a conservation panacea. We suggest that unmanaged carnivore species be the formal components of carnivore reintroduction and recovery programmes to better gauge the complementary conservation role of South Africa's private land.