Landscape-scale drivers of mammalian species richness and functional diversity in forest patches within a mixed land-use mosaic

Ecological Indicators 113 (2020) 106176

Contents lists available at ScienceDirect

Ecological Indicators journal homepage: www.elsevier.com/locate/ecolind

Original Articles

Landscape-scale drivers of mammalian species richness and functional diversity in forest patches within a mixed land-use mosaic

T

Yvette C. Ehlers Smitha, David A. Ehlers Smitha, Tharmalingam Ramesha,b, Colleen T. Downsa,⁎ a b

Centre for Functional Biodiversity, School of Life Sciences, University of KwaZulu-Natal, Private Bag X01, Scottsville, Pietermaritzburg 3209, South Africa Sálim Ali Centre for Ornithology and Natural History, Anaikatti, Coimbatore, Tamil Nadu 641108, India

ARTICLE INFO

ABSTRACT

Keywords: Camera-trapping Community assemblage Forest fragmentation Landscape scale Mammal diversity

We estimated the influence of habitat fragmentation characteristics (patch size, isolation and number of neighbouring patches), habitat and land management types at the community level, specifically, on species richness, functional- α (alpha), guild- and β (beta) diversity of forest and dense bush habitat patches in southern KwaZulu-Natal Province, South Africa. We used mammal camera-trap data from 245 survey points across 157 habitat patches. We partitioned β diversity-change (βtotal) into its nestedness (βnestedness) and turnover (βturnover) components. Species turnover was the dominant component of species β diversity, while nestedness was the dominant component of functional β diversity. Carnivore and insectivore diversity significantly decreased with increasing isolation from large patches. Patch size did not influence species-, functional- or guild diversity. We found that the number of neighbouring patches nested within the mixed land-use matrix were important drivers of functional- and species α diversity, carnivore- and herbivore diversity. Fragmentation characteristics alone did not explain the patterns of diversity tested, but habitat and land management type were important factors for all diversity indices. We recommend incentivised protection for privately owned forest patches, as they act as stepping stones for movement of forest-associated mammals throughout the landscape. Furthermore, we emphasise the importance of habitat quality and of effective management of forest in mixed land-use mosaic landscapes to prevent further degradation.

1. Introduction Biodiversity loss is driven by land-use change, habitat loss and fragmentation as a result of human population increase (Tscharntke et al., 2005; Pereira et al., 2012). Fragmentation and habitat loss imply the increase in number of smaller patches, a decrease in patch size and an increase in isolation distances (Fahrig, 2003), which exerts various pressures on biodiversity at both community levels and on functional traits (Fahrig, 2003; Pardini et al., 2005; Magrach et al., 2014), and most specifically impacting habitat specialists (Ries et al., 2004). The result is a modified or compromised ecosystem (Hector et al., 2001) with reduced ecosystem services (Allan et al., 2015). MacArthur and Wilson’s (1967) Island Biogeography Theory (IBT) is based on the principle that large, connected islands support higher species diversity in comparison to small, isolated islands. Fragmented forest patches are often described as habitat islands in a sea of modified landscapes (Broadbent et al., 2008; Laurance et al., 2009; Gibson et al., 2013). However, the number of species that will eventually disappear from fragmented forest patches (considered relaxation, or owed as extinction

⁎

debt; Tilman et al., 1994) will vary based on the patch size, the surrounding habitat matrix, the dispersal ability and mobility of a given species, and the isolation distance from a potential source population (Prugh et al., 2008). Under the relaxation process (as per Tilman et al., 1994), extinction is the dominant phenomenon rather than colonisation. Indicator species, communities or traits, are used to assess the status and trends of biodiversity (Feld et al. 2009). The current literature is increasingly shifting focus from species diversity to functional diversity when assessing the influence of fragmentation (Hevia et al., 2016; Si et al., 2016). Species diversity alone does not describe ecosystem processes and community structure (Safi et al., 2011), as it considers all species being equally different from each other and does not consider different ecological functions amongst species (Si et al., 2015, 2016). By modifying the landscape, we may select for different functional traits. Therefore, the knowledge of functional diversity enables an assessment of how anthropogenic habitat changes influence species assemblages (Ehlers Smith et al., 2015, 2018a,b). Ehlers Smith et al. (2018c) showed that decreasing patch size causes

Corresponding author. E-mail addresses: [email protected] (Y.C. Ehlers Smith), [email protected] (D.A. Ehlers Smith), [email protected] (C.T. Downs).

https://doi.org/10.1016/j.ecolind.2020.106176 Received 6 March 2018; Received in revised form 5 December 2019; Accepted 31 January 2020 1470-160X/ © 2020 Elsevier Ltd. All rights reserved.

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

Fig. 1. Map of the study region within the UGU district municipality of southeast KwaZulu-Natal (KZN) Province, South Africa, indicating the habitat patches surveyed.

a homogenisation of habitat structures and negatively impacts habitat complexity, which limits the number of niches and resources available for exploitation (Bonthoux et al., 2012) and ultimately dictates the range of functional traits that a habitat patch may provision for. Homogenised habitats often do not support a diverse range of functional traits (Carmona et al., 2012). Where habitat and niche availability are reduced, species diversity may remain constant while functional diversity decreases, as specialist species with diverse traits are replaced by generalist species (Safi et al., 2011), thus habitat loss drives functional trait loss, causing relaxation or extinction of species which possess pressured traits. It is critical to understand the influences of landscape-scale habitat fragmentation on different diversity indices, especially as species and functional diversity address ecological questions and conservation challenges (Mason and de Bello, 2013). Mammals possess a wide array of ecological and social strategies that equate to a broad suite of functional traits (de Bello et al., 2010; Ahumada et al., 2011). Forest mammals consist of a variety of functional guilds including frugivores, insectivores, browsing herbivores and carnivores, or have solitary, pair or group living social strategies. Each guild fulfils crucial ecological functions such as nutrient cycling, and are predators or prey items (Boshoff et al., 1994; Bowkett et al., 2008; Emerson and Brown, 2013; Lacher et al., 2019; Pellerin et al., 2010; Seufert et al., 2010). Where some functional groups are absent or removed, a system generally becomes imbalanced (Lacher et al., 2019). By calculating both 1) species and 2) functional diversity at the level of: i) alpha (α) diversity (i.e. diversity indices measured at the local site/patch level); ii) gamma (γ) diversity (i.e. the regional pool of diversity indices measured, Whittaker, 1960, 1972), and iii) beta (β) diversity (the change or dissimilarities between site-specific α and regional γ diversity) one can measure changes in diversity across spatial scales (Anderson et al., 2011; Mason and de Bello, 2013; Socolar et al., 2015). The patterns of change in β diversity (βtotal) can be partitioned into two antithetic mechanisms: a) nestedness (βnestedness), dissimilarity through loss (or gain) of diversity indices measured resulting in nested subsets of the γ diversity at the α level, and b) turnover (βturnover), the

dissimilarity as a result of replacement of any diversity measure from the γ diversity pool to the α diversity level (Baselga, 2010; Baselga and Leprieur, 2015; Soininen et al., 2018). In this study, we applied MacArthur and Wilson’s (1967) IBT to fragmented landscapes (i.e. habitat islands in a sea of modified landscapes; Broadbent et al., 2008; Laurance et al., 2009; Gibson et al., 2013). We expanded on the IBT by including patch interconnectivity, habitat- and land management types in addition to patch size and isolation distances at the landscape scale. These fragmentation characteristics were used to elucidate patterns in forest-associated mammal diversity, including species- and functional α and β diversity, and feeding guild diversity within the IOCB forest patches in southern KwaZulu-Natal (KZN) Province, South Africa. The units in which diversity are measured respond differently to fragmentation pressures. To this end we specifically hypothesised the following: H1. Suggest not confusing a hypothesis with a prediction immediately therein: i) decreasing patch size and ii) increasing isolation from mainland habitat patches would have a negative effect on diversity indices being tested (i.e. species richness and functional α diversity and individual guild diversity); H2. A higher clustering of habitat patches (patch connectivity) was predicted to have a positive influence on the diversity indices in H1, resulting in improved connectivity in the form of stepping-stones; H3. As above fragmentation characteristics such as patch size, isolation and patch connectivity would not act alone in their influence on the aforementioned diversity indices. The habitat patch type and the surrounding land-use would influence diversity indices as specific habitat types are preferred by individual species various landuse types exert different disturbance levels H4. Turnover would be the dominant process of β diversity change as forest specialist species are replaced by disturbance-tolerant generalists as the fragmentation gradient became more severe.

2

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

2. Methods

2.3. Mammalian functional and biological traits

2.1. Study region

From the camera-trap photographs, we created a mammal species presence/absence matrix for each surveyed forest patch and calculated a matrix of functional traits present within the survey region. The trait matrix consisted of functional traits reflecting species’ habitat and resource use, including: activity patterns (diurnal vs nocturnal); body mass (kg); feeding guild (omnivore; insectivore; carnivore; herbivore browser; herbivore generalist; frugivore); habitat preference (forest specialist vs generalist) and social organisation (solitary; pair bonded; small family group; large group living), derived from the known literature (Kingdon, 1997; Skinner and Chimimba, 2005). Additionally, we defined feeding guild dominance by calculating the average guild percentage value within the survey area. Each feeding guild was represented by the number of species constituting a guild within each forest patch. For each guild we calculated its percentage representation across all forest patches surveyed (de Bello et al., 2010).

Our research was conducted within the IOCB forest patches of Ugu district municipality of KZN Province, South Africa (Ehlers Smith et al., 2017a,b, 2018a,b). The southern limit of our research area was the Umtamvuna River (31°04′46.69″ S, 30°11′39.87″ E; Fig. 1), near Port Edward, and the Umkomazi River, near Umkomaas (30°12′1″ S 30°48′4″ E) was the northern boundary. The climate is as subtropical, with a range in mean annual rainfall of 440–1400 mm (Mucina and Rutherford, 2011). The area contains naturally fragmented patches of indigenous forest within a mixed land-use habitat matrix (Olivier et al., 2013; GeoTerra Image, 2014), but anthropogenic actions have further fragmented the forest habitat. The IOCB contains various forest subclasses; the dominant forest types within the study region are scarp and lowland coastal forest. Scarp Forest is the most ancient of the two forest types and it was a refugium during the Quaternary climatic events (Cooper, 1985; Lawes, 1990; Eeley et al., 1999, 2001; Lawes et al., 2000a, 2007). Subsequently, because of paleo-climatic change and biogeographic influences, this forest type has been naturally fragmented since the last glacial maximum (ca.18,000 years Before Present; Moll and White, 1978; White, 1978; Cooper, 1985). Conversely, lowland coastal forests were established after the glacial maximum (ca. 8000 years ago; White, 1978; Lawes, 1990; Eeley et al., 1999). Various studies surmise the extent of historic lowland coastal forest loss (Cooper, 1985; Lawes, 2002; Berliner, 2009; Olivier et al., 2013), and the consensus is that anthropogenic activities within the region date back to the late Iron Age (1300 s). With the early peoples came the gradual fragmentation and reduction in the Coastal Forest habitat. Additionally, extensive patches of Coastal Thicket /Dense Bush (hereafter dense bush) existed (Eeley et al., 1999; Mucina and Rutherford, 2011; GeoTerra Image, 2014), which also supported forest-dependent mammals, and based on structural and plant species composition can be considered regenerating forests (Ehlers Smith et al., 2017a,c). Extant forest patches are smaller and more closely situated than modelled, historic forest patches (Olivier et al., 2013), indicating continued forest clearance and anthropogenic fragmentation. Modern development has caused extensive transformation of natural habitats for trade and industry development (Geldenhuys and MacDevette, 1989; Midgley et al., 1997), resulting in a matrix of fragmented forest patches nested in a mixed land-use matrix.

2.4. Quantifying fragmentation characteristics We utilised the latest land-cover layers based on 30 m resolution Landsat 8 imagery (GeoTerra Image, 2014) in ArcGIS v10.2 (ESRI, 2011) for area calculations, distance mapping and connectivity of forest patches. The fragment size was adjusted by considering splitting effect of roads within ArcGIS. We used the Ugu District Municipality, Road Network 2015 data layer, downloadable from gis.kzntransport.gov.za/ maprequests_generic_dm.aspx. These were roads that were either tarred or gravel roads, utilisable by vehicles that were maintained by the district municipality. We calculated forest patch connectivity as the number of surrounding individual patches within an 800 m buffer of each surveyed patch. This buffer size was based on the estimated maximum dispersal distance for the most specialised, but least vagile forest terrestrial mammal within the study region, blue duiker Philantomba monticola (Lawes et al., 2000b). Additionally, as ‘landscape isolation measures’, isolation distances were calculated as the straight-line nearest neighbourhood distance from large forest patches, those patches larger than 90 ha. This distinction was made based on the largest coastal forest patch within the study region. Each forest patch was also classified according to the land-use/ management category that it fell within, i.e. nature reserve, farmland or residential (See Ehlers Smith et al., 2017a,b; 2018a,b for classification method). As a proxy for land-use intensity or as indication of the human infrastructure present, we used the number of roads within a 1 km buffer of each selected site (as per Ehlers Smith et al., 2019). Nature reserves can be considered as the least disturbed with the lowest landuse intensity (0.06 km of road per km2) and residential areas the most disturbed and highest intensity (65.71 km of road per km2). Farmland was considered an intermediate between the two land-use categories (26.71 km of road per km2; Ehlers Smith et al., 2019).

2.2. Camera-trap survey design and implementation Full methods on survey design and camera-trap station selection are described in Ehlers Smith et al. (2017a). Site selection was based on the 2014 land-cover geographic information system (GIS) map (GeoTerra Image, 2014), where habitat patches were either categorised as coastal forest or dense bush. We overlaid a 400 m × 400 m grid over each identified patch in ArcGIS (ESRI, 2011) to assign camera-trap survey stations. We deployed infrared motion detection camera-traps (Moultrie® M880, EBSCO Industries, Inc., USA), to survey the presence of mammals at each survey point within forest and dense bush habitat patches, hereafter referred to as forest patches. Camera-traps were set at a height of 20–30 cm above ground, attached to a robust tree on a game trail or within an open glade allowing the camera sensor optimum range. Vegetation within the view range of camera stations were cleared to avoid misfires. The camera-traps were installed for 24 h a day with a 30 s motion triggered delay setting at each predesignated survey station for a minimum of 21 days, to reduce the probability of change in species’ occupancy. The first survey cycle was conducted between June 2014 and May 2015, and the follow-up surveys were conducted between June 2015 and May 2016, resulting in one full survey for each camera-trap station per season (Ehlers Smith et al., 2017a).

2.5. Analyses Species richness (patch level) was calculated as the total number of native species (excluding domestic and introduced) recorded from each camera-trap station within that patch. We adopted the concept described by Villéger et al. (2008) to describe functional diversity for a community within a patch (Functional α diversity), where a species is distributed in a multidimensional functional space and functional richness (FRic) is described as the volume of the functional (niche) space occupied by the species present within the community (Hevia et al., 2016; Si et al., 2016; Ehlers Smith et al., 2018c). To calculate FRic, we incorporated the species-trait matrix (as described above), utilising the Distance-Based Functional Diversity Indices analysis tools within the “FD” package (Laliberté and Legendre, 2010) in R v3.3.1 (R Core Team, 2015). Quantitative traits (e.g. mass) 3

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

Table 1 Mammalian community and associated biological and functional traits recorded from study sites in South Africa’s Indian Ocean Coastal Belt in the present study. (Ψ = Naïve station occupancy across the study region; Body mass data from Smithers and Chimimba, 2005). Species

Latin

Activity pattern

Body mass (kg)

Diet

Habitat

Grouping

Naïve Ψ

Chacma baboon Banded mongoose Black-backed jackal Blue duiker Bushbuck Bushpig Cane rat Cape porcupine Caracal Grey duiker Large-grey mongoose Large-spotted genet Marsh mongoose Red duiker Rock hyrax Samango monkey Scrub hare Slender mongoose Vervet monkey White-tailed mongoose

Papio ursinus Mungos mungo Canis mesomelas Philantomba monticola Tragelaphus scriptus Potamochoerus larvatus Thryonomys swinderianus Hystrix africaeaustralis Caracal caracal Sylvicapra grimmia Herpestes ichneumon Genetta tigrina Atilax paludinosus Cephalophus natalensis Procavia capensis Cercopithecus mitis labiatus Lepus saxatilis Galerella sanguinea Cercopithecus pygerythrus Ichneumia albicauda

Diurnal Diurnal Nocturnal Diurnal Diurnal Nocturnal Nocturnal Nocturnal Nocturnal Diurnal Diurnal Nocturnal Nocturnal Diurnal Diurnal Diurnal Nocturnal Diurnal Diurnal Nocturnal

28 1.87 10 4.35 51.25 71.75 10.1 17 12.96 17.7 3.2 1.9 3.05 11.8 3.35 5.5 3 0.447 5.1 3.6

Omnivore Insectivore Carnivore Herbivore browser Herbivore browser Omnivore Herbivore Herbivore Carnivore Herbivore browser Carnivore Carnivore Carnivore Herbivore browser Herbivore Frugivore Herbivore Insectivore Omnivore Insectivore

Generalist Generalist Generalist Specialist Specialist Generalist Generalist Generalist Generalist Generalist Generalist Generalist Generalist Specialist Generalist Specialist Generalist Generalist Generalist Generalist

Large group Large group Pair Pair Small group Small group Single Pair Solitary Pair Single Single Single Pair Small group Large group Single Single Large group Single

0.07 0.03 0.10 0.90 0.92 0.59 0.02 0.62 0.17 0.37 0.05 0.88 0.41 0.15 0.19 0.18 0.00 0.12 0.54 0.01

were log-transformed to obtain normal distribution of trait values. We used Gower distance (Gower, 1966) to calculate pairwise distance scores of species’ traits present in the community, considering multiple traits using the “FD” package. Then principal coordinate analysis (PCoA) was performed, using the resultant functional distance matrix of species’ traits. The trait space is a convex hull volume (calculated via the Quickhull algorithm) and is defined by linked extreme values of species traits, resulting in the FRic score (as described above; Villeger et al., 2008). Five of the patches from our overall survey region contained too few species to constitute a functional community to calculate a functional diversity score and were subsequently excluded from the analysis. The first three PCoA axes were retained to describe functional α diversity, which explained 91% of total inertia (Eigenvalues of principal coordinate analysis of axes 1 = 0.67; axes 2 = 0.45; axes 3 = 0.32, See Supporting Information Appendix 1). Finally, we calculated a functional diversity score per patch using the FRic index of the “FD” package. We used the “betapart” package (Baselga and Orme, 2012) in R v3.3.1 (R Core Team, 2015) to partition species- and functional β diversities into their spatial turnover (βsim as per Baselga, 2010) and nestedness (βnes) resultant components. We applied the Sørensen dissimilarity index to create a matrix of pairwise species traits that describes overall species- and functional β diversities. The Simpson’s dissimilarity index describes the effect of turnover (βsim), and their difference to describe nestedness (βnes, Baselga, 2010, 2012). We calculated the proportion of the nestedness component to overall β diversity. Where the calculated β diversity ratio exceeded 50%, the dominant process in β diversity change was because of the nestedness component, and vice versa for spatial turnover (Dobrovolski et al., 2012; Si et al., 2016). We used Generalised Linear Models (GLMs) within the package “lme4” (Bates et al., 2015) to test the significant influence of the continuous explanatory variables: forest patch size; isolation distance from nearest mainland patch (hereafter isolation) and the influence of number of neighbouring patches (as interconnectivity) and habitat and land-use classification (categorical variables) on species-, functionaland individual feeding guild α diversity. For each of our explanatory variables we ranked different models that have each of our covariates in different combination. We included a ‘full model’ (which included each response variable) and also a model for each response variable independently, among others. Each covariate had the same likelihood of explaining the patterns in the data

irrespective of whether it was in combination with other variables in the model. Thus, we used an information theoretic (IT) analysis approach to select the best approximating/parsimonious model based on Akaike’s Information Criterion (AIC) as described by Burnham and Anderson (2003). Although all models with Δ AIC < 2 can be considered equal in their ability to predict, we present beta coefficients based on the best approximating model, with the lowest Δ AIC (0) (Burnham and Anderson, 2003) There was no multicollinearity between forest patch size and isolation distance (Spearman rank correlation: N = 123; r = 0.110, P = 0.113), therefore we retained both as explanatory variables of landscape fragmentation. Each feeding guild was represented by the number of each species constituting a guild within each forest patch. We used the log-link function and different error structures were selected based on the distribution of the data, to correct for heteroscedasticity and to deal with overdispersion (Cameron and Trivedi, 2013). Using the “ecodist” package (Goslee and Urban, 2007) in R (R Core Team, 2015) we incorporated the resultant matrices created as above (Sørensen and Simpson) into multiple regression of distance matrices (MRM), which estimated the effect of forest patch size and isolation on mammalian species- and functional β diversities and the resultant components, spatial turnover and nestedness (Lichstein, 2007). The variables within the MRM are distance matrices and values in a distance matrix are not independent, but are related to one another, therefore, Pvalues were estimated by permutation test (9,999 runs) (Legendre et al., 1994; Si et al., 2016). 3. Results Throughout the survey region we recorded 20 mammalian species (γ diversity; Table 1), across 245 camera-trap survey sites within 157 distinct forest patches within two survey cycles. We excluded from the analyses species that were deliberately or accidentally introduced into the region (e.g. non-native and domestic species such as Bos Taurus, domestic cow; Canis familiaris, domestic dog; Connochaetes taurinus, blue wildebeest and Equus quagga, plains zebra). The mean forest patch species α diversity equated to 6.8 species ± SD 1.9 (range 4–14), whereas the mean camera-trap survey site species α diversity equaled 6.41 species ± SD 1.6 (range 3–13; Tables 1 and 2). For each diversity measure tested, the set of alternative best approximating models (Δ AIC < 2) highlighted the best predictive ability 4

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

the influence of the number of neighbouring patches on carnivores and insectivore diversity, was dependent on the interaction with the habitat classification types and included the ‘isolation from mainland’ variable. Where the neighbouring patches were classified as dense bush, it had a negative influence (Fig. 3a). Isolation from mainland patches negatively influenced carnivores and insectivore diversity (Table 4; Fig. 3b). Herbivore diversity was the lowest within protected areas (Fig. 3c) and was positively influenced by number of neighbouring patches (Table 4; Fig. 3d). Only one model with Δ AIC ≤ 2 was strongly associated for each of the carnivores, insectivore and herbivore guild diversity measures. The eight models Δ AIC ≤ 2 indicated, with the exception of patch size, that each of the individual covariates influenced omnivore and frugivore diversity. Though the best approximating model with the lowest Δ AIC (0) showed that omnivore and frugivore diversity was the most strongly influenced by isolation from mainland patches, which also had the highest Akaike weight (Table 3; Fig. 3e). There were highly significant correlations between mammalian species and functional diversity at the α (r = 0.8, P < 0.0001; Fig. 4a), overall β (r = 0.5, P < 0.001; Fig. 4b), β turnover (r = 0.6, P < 0.001; Fig. 4c) and β nestedness (r = 0.5, P < 0.001; Fig. 4d) levels. Spatial turnover was the dominant component of β species richness, as the nestedness component accounted for 35% (±0.26), but 59% (±0.35) of mammalian β functional diversity (Table 5). None of the fragmentation characteristics (forest patch size, no of neighbouring patches and isolation distance) within the MRM had a significant influence on mammalian species- nor functional β diversity and their respective components, nestedness and turnover (P > 0.05; Table 6).

Table 2 The characteristics of the 157 habitat patches surveyed within the Indian Ocean Coastal Belt of KwaZulu-Natal, South Africa in the present study.

Mean Std. Dev Min. Max.

Patch size (ha)

Distance to mainland (m)

No. of clusters

Species richness

Functional diversity

27.59 74.45 0.06 773.99

2571.29 2640.01 0.00 8688.44

4.89 3.39 0.00 18.00

6.76 1.95 4.00 14.00

4.44 3.02 0.04 11.27

and represented the variation within the data (Tables 3 and 4; Burnham and Anderson, 2003). The models Δ AIC ≤ 2 indicated that patch size and habitat type did not influence species richness. The covariates representing the number of neighbouring patches and land-use classification were present within both best approximating models (with the lowest Δ AIC (0)) for both species’ richness and functional α diversity, respectively (Tables 3 and 4). Both species and functional diversity increased with an increase in the number of surrounding patches (Fig. 2b, e). Protected areas overall had lower species α diversity, whereas residential areas showed lower functional (FRic) α diversity (Table 4; Fig. 2a, d). Isolation from mainland patches had a significant negative influence (P < 0.05) on species α diversity (Table 4; Fig. 2c) and the covariate ‘patch size’ was not associated with species richness. Patch size was only present within one model (Δ AIC = 1.8) associated with FRic α diversity. (See Supplementary Information Appendix 2 for individual surveyed patch characteristics). We found variation in the representation of mammalian functional guilds within our survey region. Herbivore browsers accounted for the largest proportion (% of total community composition) of the functional guild across the survey region (35%, represented by four species), followed by carnivores (31%, five species), omnivores (18%, three species), generalist herbivores (12%, four species), frugivores (3%, one species) and insectivores (3%, three species). As overall within-guilddiversity was low, we grouped all herbivores (browsers and generalist herbivores), carnivores with insectivores and omnivores with frugivores for further analyses. None of our feeding guilds were influenced by patch size. Results from the best approximating models (Δ AIC = 0; Table 3) indicated that

4. Discussion Ecosystem functioning is reliant on multiple abiotic and biotic processes across various groups of organisms (de Bello et al., 2010; Lundin et al., 2012). Miller-Rushing et al. (2019) highlighted the lack of studies that test fragmentation and interactions with other effects across scales and which species benefit and which are harmed by fragmentation at landscape scales. We showed that for this study system, MacArthur and Wilson’s (1967) IBT alone did not describe sufficiently the influence of habitat

Table 3 GLMs (Δ AIC < 2), based on AIC rankings, for testing fragmentation measures on species richness and functional α diversity and feeding guild diversity across the survey region as calculated within each survey patch. Response

Model

Models (Δ AIC < 2)

Resid. Dev

Resid. Df

Disp.

AICc

Δ AIC

ωi

1 2

Land use + Isolation distance (logged) + Neighbouring patches Isolation distance (logged) + Neighbouring patches

51.16 57.06

152.00 154.00

0.34 0.37

648.15 649.81

0.00 1.66

0.23 0.10

Functional diversity2

1 2 3 4 5

Land use + Neighbouring patches Land use + Isolation distance (logged) + Neighbouring patches Habitat + Land use + Neighbouring patches Habitat + Land use + [Habitat*Neighbouring patches] Land use + Patch size (logged) + Neighbouring patches

118.00 99.79 99.46 96.94 100.82

156.00 152.00 151.00 149.00 152.00

0.76 0.66 0.66 0.65 0.66

752.70 752.72 754.33 754.37 754.51

0.00 0.02 1.63 1.67 1.80

0.31 0.30 0.14 0.13 0.12

Carnivore & Insectivore diversity1

1

Habitat + Isolation distance (logged) + [Habitat*Neighbouring patches]

68.81

150.00

0.46

471.48

0.00

0.42

1

Land use + Neighbouring patches

46.23

153.00

0.30

538.36

0.00

0.30

1 2 3 4 5 6 7 8

Isolation distance (logged) Habitat (Null) Land use Neighbouring patches Isolation distance (logged) + Neighbouring patches Habitat + Neighbouring patches Land use + Neighbouring patches

53.70 51.91 56.74 53.18 55.38 53.33 51.27 51.44

155.00 154.00 156.00 154.00 155.00 154.00 153.00 153.00

0.35 0.34 0.36 0.35 0.36 0.35 0.34 0.34

531.90 532.19 532.88 533.46 533.58 533.61 533.65 533.82

0.00 0.29 0.99 1.56 1.68 1.71 1.76 1.92

0.12 0.11 0.08 0.06 0.05 0.05 0.05 0.05

Species richness

1

Herbivore diversity

1

Omnivore & Frugivore diversity3

Resid. Dev = Residual deviance; Resid. Df = Residual degrees of freedom; Disp. = Dispersion parameter; AICc = Akaike’s Information Criterion; Δ AIC = Delta AIC; ωi = summed Akaike weights. [*] – indicates inclusion of explanatory variables and interaction terms (:) [e.g. Variable A* Variable B, expands to A + B + A:B]; + indicates the inclusion of an explanatory variable in the model. 1 Poisson distribution with log-link function; 2Gamma distribution with log-link function, 3Negative binomial distribution with log-link function. 5

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

One of the limitations of this study is that it does not include the total available amount of habitat within the study area. Carnivore-, insectivore-, frugivore- and omnivore diversity were influenced by isolation. An increase in isolation reduced diversity of these individual guilds. Reduced movement of seed dispersers between degraded forests may affect plant community composition (Magrach et al., 2014). Mammalian seed dispersal functions within our study region were performed by omnivorous species (accounting for 18% of the metacommunity), in addition to the obligate frugivores (3% of the metacommunity). For example, the samango monkey (C. mitis labiatus), a specialised frugivorous, group living, cheek pouch monkey (Enstam and Isbell, 2011) predominantly existed in the larger mature, scarp forest reserves within our study area, that harbour a larger proportion of climax forest tree species (Ehlers Smith et al., 2017a, 2017b) and therefore a large isolation distance would particularly influence the colonisation ability of samango monkeys and the seeds that they disperse. Conversely, Prugh et al. (2008) found that specialist mammalian species (by nature more restricted by specific habitat characteristics than generalists) were not sensitive to patch size or isolation, contrasting the concept that the “habitat island” paradigm fits terrestrial systems (viz. MacArthur and Wilson, 1967). Similar to Prugh et al. (2008), Fahrig (2017) highlighted that fragmentation effects influenced threatened or specialist species positively. Lawes et al. (2000b) found that samango monkeys within southern mistbelt forests were not influenced by the isolation distances between patches. Our result, particularly in relation to specialist frugivores such as samango monkeys contrasts the findings of these studies. Caracals (Caracal caracal) and black-backed jackals (Canis mesomelas), although classed as meso-carnivores, act as the top predators within our study system with the absence of leopards (Panthera pardus) (Ehlers Smith et al., 2019). Caracals and black-backed jackals have relatively large home ranges (caracals > 4000 ha, Ramesh et al., 2017; black-backed jackals 11.4–18,200.00 ha Rowe-Rowe, 1982; Humphries et al., 2016) and are theoretically capable of covering larger distances, yet they were negatively influenced by the isolation effect. These species were until recent years absent from the region and are persecuted throughout agricultural landscapes that make up most of the study area (Ehlers Smith et al., 2019), and the isolation of patches within the mixed land-use mosaic might hinder further colonisation. Isolation is relevant to metapopulation dynamics if it represents the proximity of potential colonisation possibilities, rather than just the presence of a habitat patch (Prugh et al., 2008). If immigration is responsible for rescuing populations from extinction, more isolated patches are more likely to undergo local extinctions and delayed recolonisation (repopulating empty patches that were once occupied) in comparison with less isolated patches (Clinchy et al., 2002). Prugh et al. (2008) showed that isolation and patch size were poor predictors of occupancy, as often the land-use matrix and habitat quality were not considered. Similarly, Schooley and Branch (2009) in their study of the area-isolation paradigm found that patch colonisation depended on patch area, patch quality, spatial connectivity to potential population source patches and the land-use types within the matrix also influenced metapopulation dynamics. Our results (as the case potentially with caracal and black-backed jackal within our study region) also show that fragmentation pressures alone did not adequately explain the patterns in mammalian diversity recorded, because of the strong effects of the surrounding matrix (i.e. land-use) and the quality of the habitats (H3). An increase in the number of neighbouring patches (H2) had a significant positive influence on various guild-, species-, and functional α diversity. Our results suggest that for those diversity indices uninfluenced by isolation, an increase in the number of neighbouring patches may reduce the influence of isolation. Isolation (H1) had a negative influence on carnivore, and insectivore diversity, but depending on the habitat type (H3), isolation might be buffered by the number of surrounding patches. Within coastal, and scarp forest patches an increase

Table 4 Breakdown of model-averaged coefficients based on most parsimonious GLM models (Δ AIC < 2) to test the influence of patch size, isolation distance from mainland patches, connectivity, habitat and land-use classifications on species richness-, functional (FRic)- and individual guild diversities calculated within each survey patch. Response

Coefficients

Estimate

Std. Error

Z-value

P-value

Species richness1

(Intercept) Protected Area Residential Area Isolation Neighbouring Patches

1.952 −0.218 −0.100 −0.062 0.038

0.133 0.093 0.075 0.032 0.009

14.578 2.327 1.317 1.932 4.026

0.020 0.188 0.053 0.000

Functional diversity (FRic)2

(Intercept) Protected Area Residential Area Neighbouring Patches Isolation Dense bush Scarp forest Neighbouring Patches: Dense bush Neighbouring Patches: Scarp forest Patch size

1.403 −0.246 −0.451 0.080

0.206 0.162 0.127 0.021

6.791 1.504 3.538 3.820

0.133 0.000 0.000

−0.095 −0.041 0.075 −0.077

0.055 0.254 0.253 0.035

1.697 0.163 0.295 2.167

0.090 0.871 0.768 0.030

0.002

0.040

0.058

0.953

0.047

0.068

0.684

0.494

(Intercept) Dense bush Scarp forest Isolation Neighbouring Patches Neighbouring Patches: Dense bush Neighbouring Patches: Scarp forest

0.831 0.374 −0.147 −0.166 0.085

0.243 0.222 0.314 0.058 0.021

3.413 1.683 −0.468 −2.857 4.088

0.092 0.640 0.004 0.000

−0.102

0.035

−2.928

0.003

−0.033

0.035

−0.946

0.344

Herbivore diversity1

(Intercept) Protected Area Residential Area Neighbouring Patches

1.225 −0.410 −0.110 0.030

0.087 0.129 0.097 0.013

14.100 −3.173 −1.129 2.357

0.002 0.259 0.018

Omnivore and Frugivore diversity3

(Intercept) Isolation Dense bush Scarp forest Protected Area Residential Area Neighbouring Patches

0.520 −0.146 −0.264 0.321 0.234 −0.266 0.029

0.224 0.043 0.113 0.151 0.132 0.118 0.016

2.314 3.374 2.313 2.108 1.754 2.243 1.773

0.001 0.021 0.035 0.079 0.025 0.076

Carnivore and Insectivore diversity1

1 2 3

Poisson distribution with log-link function. Gamma distribution with log-link function. Negative binomial distribution with log-link function.

fragmentation on mammalian diversity. Our results also only partly supported our predictions. Diversity indices responded differently to landscape-scale pressures, thus emphasising the importance of quantifying variation within the community using a range of diversity indices, in addition to species richness (Safi et al., 2011). Our first prediction (H1) applied the concepts of IBT, where decreasing patch size and increasing isolation distance would negatively affect diversity indices. Patch size, however, did not influence either species richness or functional diversity. Though, isolation distance influenced species richness negatively, but did not influence functional diversity. The more isolated from a mainland (source) patch, the lower the number of species within a patch. Fahrig (2017) highlighted the importance of habitat amount within the landscape as patch size alone does not describe patterns of diversity within fragmented landscapes. 6

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

Fig. 2. Significant results from Generalised Linear Modelling for Species richness (α) by a) Land-use classification; b) Number of neighbouring patches, and by c) Isolation from mainland patches (km). Functional α diversity by d) Land-use classification; e) Number of neighbouring patches and f) Number of neighbouring patches within specific habitat sub-classes within surveyed patches of the Indian Ocean Coastal Belt, South Africa. (Habitat classifications: CF = Coastal Forest; DB = Dense bush; SF = Scarp forest, Land-use classifications: FL = Farmland; PA = Protected Area; RA = Residential Area). Predictions are based on modelaveraged coefficients from most parsimonious GLM models (Δ AIC < 2).

in the number of neighbouring adjacent patches had a positive influence on carnivore, and insectivore diversity (H2). However, where surrounding patches where classified as dense bush habitat, an increase in the number of surrounding patches had a negative effect on carnivore, and insectivore diversity. This may reflect a preference for mature forest stands (e.g. coastal, and scarp forests, Ehlers Smith et al., 2017a). Thus, for carnivore, and insectivore diversity within our study region, the habitat type itself (considering the carrying capacity of a forest patch and the availability of food items), plays an important role on diversity compared to just fragmentation characteristics alone (H3). Fahrig et al. (2019) highlighted the scenario where researchers have shown that species richness is lower in a small patch compared to a large patch, and subsequently conservation agencies interpret the result

to mean that smaller fragments (broken apart) of habitat has much lower conservation value. Although coastal and scarp forest distribution within our study area are limited, even small forest patches contained a relatively high mammalian diversity, highlighting the importance of these patches in maintaining species richness, providing “steppingstone” connectivity throughout the landscape and supporting previous research showing that small forest patches play crucial conservation roles in enhancing landscape connectivity (Turner, 1996; Turner and Corlett, 1996). Remaining small habitat patches may buffer the effects of historic forest loss and fragmentation, and act as sources of ecosystem services (Gagic et al., 2015). However, within our survey region small patches were the most threatened. We were not able to resurvey many patches as they had been cleared for agriculture or development 7

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

(caption on next page)

8

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

Fig. 3. Significant results from Generalised Linear Modelling for individual feeding guilds sampled within habitat patches of the Indian Ocean Coastal Belt, South Africa: a) carnivore and insectivore diversity and influence of number of neighbouring patches by habitat type; b) carnivores and insectivore diversity and influence of isolation from mainland patches; c) herbivore diversity and land-use classification; d) herbivore diversity and influence of number of neighbouring patches; e) omnivore and frugivore diversity and isolation from mainland patches; f) omnivore and frugivore diversity and influence of land-use classifications, and g) omnivore and frugivore diversity and influence of habitat classifications. (Habitat classifications: CF = Coastal Forest; DB = Dense bush; SF = Scarp forest, Land-use classifications: FL = Farmland; PA = Protected Area; RA = Residential Area). Predictions are based on model-averaged coefficients from most parsimonious GLM models (Δ AIC < 2). P-values signify values from Tukey post-hoc testing.

Fig. 4. Correlations between mammalian a) α species and α functional diversities, b) overall species and functional β diversities, c) species and functional turnover and d) species and functional nestedness components in the mammalian communities of habitat patches in the Indian Ocean Coastal Belt, South Africa.

within residential areas, which highlights a gap within the implementation of the National Forests Act 1998 and relevant environmental impact assessment procedures. Patches within residential areas had lower functional diversity, but high species richness. This is likely due to the persistence of generalist species tolerant to disturbance at the expense of specialists, as highlighted by the higher nestedness proportion of functional β diversity change. Changes in functional and species richness often display very different response trajectories along land-use intensity gradients, and intensification processes tend to reduce functional diversity (Hevia et al., 2016). Herbivore diversity was the lowest within protected areas.

Table 5 Mean (±SD, range) values of pairwise mammalian species richness and functional β diversities, turnover and nestedness of mammalian communities from 157 habitat patches in South Africa’s Indian Ocean Coastal Belt in the present study as calculated within each survey patch.

β diversity Turnover (βsim) Nestedness (βnes) Nestedness proportion

Species richness

Functional diversity

0.36 0.23 0.13 0.35

0.56 0.24 0.34 0.59

± ± ± ±

0.12 0.15 0.09 0.26

(0–0.86) (0–0.86) (0–0.56) (0–0.86)

± ± ± ±

0.23 (0–1.00) 0.25 (0–1.00) 0.26 (0–1.00) 0.345 (0–1.00)

Table 6 Multiple regressions on the distance matrices (MRM) of the fragmentation effects (forest patch size, connectivity and isolation distance) on mammalian taxonomic and functional β diversities and their components (turnover and nestedness) from data collected in South Africa’s Indian Ocean Coastal Belt Forests in the present study.

Taxonomic β diversity Taxonomic turnover Taxonomic nestedness Functional β diversity Functional turnover Functional nestedness

Patch size

Isolation

−0.0001 −3.3E−05 −3.5E−05 7.67E−05 0.0001 −0.0001

E−06

2.78 2.33E−06 4.48E−07 −5.3E−07 1.18E−06 −1.7E−06

Connectivity

Intercept

R2

F

P-value

−0.0004 −0.0007 0.0003 0.0053 0.0064 −0.0011

0.35 0.23 0.13 0.54 0.21 0.35

0.005 0.002 0.001 0.005 0.008 0.001

21.31 7.16 4.0 22.37 32.45 4.11

P P P P P P

9

> > > > > >

0.05 0.05 0.05 0.05 0.05 0.05

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al.

Ehlers Smith et al. (2018a, 2019) showed that individual herbivore species had specific habitat associations and tolerances to disturbance. The protected areas surveyed within our study region were predominantly scarp and coastal forest (as opposed to dense bush habitats) and had the lowest land-use intensity but they had a high number of visitors that used the hiking trails (Ehlers Smith et al., 2019). Taylor and Knight (2003) showed that recreational use of nature reserves can influence the spatio-temporal behaviour patterns of ungulates. Therefore, a combination of human presence and habitat associations may influence the number of herbivore species present within protected areas. Specialist species are replaced by disturbance-tolerant generalists as the fragmentation gradient became more severe (Krauss et al., 2003). Our results supported the fourth predication (H4) and the findings of Beca et al. (2017) in that species turnover was the dominant driver of species β diversity change in forest-anthropogenic land-use matrices studied. This holds true for thicket/dense bush patches (the dominant habitat class, considered to be regenerating forest) and low-lying coastal forest patches that are well connected and host generalist species such as the common duiker (Sylvicapra grimmia) (Ehlers Smith et al., 2019). However, the large scarp forest patches acted as paleorefugia in which specialist species persisted through the last glacial maximum (Cooper, 1985; Lawes, 1990; Eeley et al., 1999, 2001; Lawes et al., 2000a, 2007), which would suggest that nestedness as a result of historic isolation of biotas in different refugia would be responsible for the β diversity change in scarp forest patches (Baselga, 2010). As a consequence of species extinction filters, (especially climatic or ecological challenges (Balmford, 1996)), species that persist may be considered relics of a greater species pool (Lawes et al., 2007). Although the concept of “community saturation” because of species interactions has largely been superseded (Loreau, 2000), others have found that South African forest mammal assemblages are unsaturated, with limited regional enrichment (Lawes et al., 2000a). Lawes et al. (2000a) suggested that the forest communities are highly resistant and resilient, but unsaturated forest mammalian assemblages exist within small forest patches of less than 1000 ha (i.e. most of our sampled patches) as a result of regional scale and historical influences. However, our results showed that the process responsible for diversity change may differ depending on the history of patches and that species recruitment through forest regeneration is possible.

quantity, for mammal richness, particularly in mixed-land-use mosaics is an aspect that should be incorporated into future research. Author contributions YCES conceived paper with DAES, TR and CTD. CTD and YCES sourced funding. YCES and DAES collected the data. YCES analysed the data and wrote the paper. DAES, TR and CTD contributed valuable comments to the manuscript. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements The authors wish to thank the anonymous reviewers for their constructive comments on this manuscript, and Ezemvelo KZN Wildlife for granting permission to conduct research within their protected area network. We thank all private landowners for allowing us access, the assistance of the Ezemvelo KZN honorary officers as well as the support from local conservancies. The University of KwaZulu-Natal (ZA), the Gay Langmuir Trust (ZA), the Hans Hoheisen Trust (ZA), the Whitley Wildlife Conservation Trust (UK), the National Research Foundation (ZA) and the Claude Leon Foundation (ZA) kindly provided financial support. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ecolind.2020.106176. References Ahumada, J.A., Silva, C.E., Gajapersad, K., Hallam, C., Hurtado, J., et al., 2011. Community structure and diversity of tropical forest mammals: data from a global camera trap network. Phil. Trans. R. Soc. Lond. B Biol. Sci. 366, 2703–2711. Allan, E., Manning, P., Alt, F., Binkenstein, J., Blaser, S., et al., 2015. Land use intensification alters ecosystem multi-functionality via loss of biodiversity and changes to functional composition. Ecol. Lett. 18, 834–843. Anderson, M.J., Crist, T.O., Chase, J.M., Vellend, M., Inouye, B.D., et al., 2011. Navigating the multiple meanings of β diversity: a roadmap for the practicing ecologist. Ecol. Lett. 14, 19–28. Balmford, A., 1996. Extinction filters and current resilience: the significance of past selection pressures for conservation biology. Trends Ecol. Evol. 11, 193–196. Baselga, A., 2010. Partitioning the turnover and nestedness components of beta diversity. Glob. Ecol. Biogeogr. 19, 134–143. Baselga, A., Gómez-Rodríguez, C., Lobo, J.M., 2012. Historical legacies in world amphibian diversity revealed by the turnover and nestedness components of beta diversity. PLoS ONE 7, e32341. Baselga, A., Leprieur, F., 2015. Comparing methods to separate components of beta diversity. Methods Ecol. Evol. 6, 1069–1079. Baselga, A., Orme, C.D.L., 2012. Betapart: an R package for the study of beta diversity. Methods Ecol. Evol. 3, 808–812. Bates, D., Mächler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1406–5823. Beca, G., Vancine, M.H., Carvalho, C.S., Pedrosa, F., Alves, R.S.C., Buscariol, D., Peres, C.A., Ribeiro, M.C., Galetti, M., 2017. High mammal species turnover in forest patches immersed in biofuel plantations. Biol. Cons. 210, 352–359. Berliner, D., 2009. Systematic Conservation Planning for South Africa’s Forest Biome: An Assessment of the Conservation Status of South Africa’s Forests and Recommendations for Their Conservation. PhD Thesis. University of Cape Town, Cape Town. Bonthoux, S., Barnagaud, J.-Y., Goulard, M., Balent, G., 2012. Contrasting spatial and temporal responses of bird communities to landscape changes. Oecologia 172, 563–574. https://doi.org/10.1007/s00442-012-2498-2. Boshoff, A.F., Palmer, N.G., Vernon, C.J., Avery, G., 1994. Comparison of the diet of crowned eagles in the savanna and forest biomes of south-eastern South Africa. S. Afr. J. Wildl. Res. 24, 26–31. Bowkett, A.E., Rovero, F., Marshall, A.R., 2008. The use of camera-trap data to model habitat use by antelope species in the Udzungwa Mountain forests, Tanzania. Afr. J. Ecol. 46, 479–487. Broadbent, E., Asner, G., Keller, M., Knapp, D., Oliveira, P., Silva, J., 2008. Forest

5. Conclusions Describing the relationship between mammalian diversity and landscape-scale fragmentation characteristics, whilst accounting for land-use within the matrix, emphasises the importance of conserving remnant habitat patches. Metrics such as patch size, connectivity and isolation independently proved poor ecological indicators of mammal diversity within our study region. Ultimately ecosystem services provided by forest communities are essential considerations within regions with a viable agricultural economy, and where public health and wellbeing in urban settings are to be considered. Our results showed the limitations of the established forest protected areas in sustaining forest mammalian assemblages, but highlighted mammalian species resilience and persistence within the unprotected habitat patches of the mixed land-use mosaic. We emphasise the importance of effective management of forest landscapes to prevent further degradation. We implore the conservation authorities to enact stricter regulations on the clearance of smaller forested habitats to secure the future of a threatened and restricted habitat. The question ‘Is fragmentation good or bad for diversity?’ is driven by the debate on ‘effects of fragmentation per se’ (fragmentation independent of habitat amount, Fahrig, 2017; Fletcher et al., 2018; Fahrig et al., 2019; Miller-Rushing et al., 2019). The ‘answer’ depends on how diversity is being measured, the spatial scale used, and interactions effects being tested. Quantifying habitat quality and assessing quality vs 10

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al. fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon. Biol. Conserv. 14, 1745–1757. Burnham, K.P., Anderson, D.R., 2003. Model Selection and Multimodel Inference: A Practical Information-theoretic Approach. Springer Science & Business Media. Cameron, A., Trivedi, P., 2013. Regression Analysis of Count Data Vol. 53 Cambridge University Press, Cambridge. Carmona, C.P., Azcárate, F.M., de Bello, F., Ollero, H.S., Lepš, J., Peco, B., 2012. Taxonomical and functional diversity turnover in Mediterranean grasslands: interactions between grazing, habitat type and rainfall. J. Appl. Ecol. 49, 1084–1093. Clinchy, M., Haydon, D.T., Smith, A.T., 2002. Pattern does not equal process: What does patch occupancy really tell us about metapopulation dynamics? Am. Nat. 159, 351–362. Cooper, K.H., 1985. The Conservation Status of Indigenous Forests in Transvaal, Natal and Orange Free State, South Africa. Wildlife Society of South Africa, Durban. de Bello, F., Lavorel, S., Díaz, S., Harrington, R., Cornelissen, J.H.C., et al., 2010. Towards an assessment of multiple ecosystem processes and services via functional traits. Biodivers. Conserv. 19, 2873–2893. Dobrovolski, R., Melo, A.S., Cassemiro, F.A.S., Diniz-Filho, J.A.F., 2012. Climatic history and dispersal ability explain the relative importance of turnover and nestedness components of beta diversity. Glob. Ecol. Biogeogr. 21, 191–197. Eeley, H.A.C., Lawes, M.J., Piper, S.E., 1999. The influence of climate change on the distribution of indigenous forest in KwaZulu-Natal, South Africa. J. Biogeogr. 26, 595–617. Eeley, H.A.C., Lawes, M.J., Reyers, B., 2001. Priority areas for the conservation of subtropical indigenous forest in southern Africa: a case study from KwaZulu-Natal. Biodivers. Conserv. 10, 1221–1246. Ehlers Smith, D.A., Ehlers Smith, Y.C., Downs, C.T., 2017a. Indian Ocean Coastal Thicket is of high conservation value for species and functional diversity of forest-dependent bird communities in a landscape of restricted forest availability. Forest Ecol. Manage. 390, 157–165. Ehlers Smith, Y.C., Ehlers Smith, D.A., Ramesh, T., Downs, C.T., 2017b. The importance of microhabitat structure in maintaining forest mammal diversity in a mixed land-use mosaic. Biodivers. Conserv. 26, 2361–2382. Ehlers Smith, Y.C., Ehlers Smith, D.A., Ramesh, T., Downs, C.T., 2018a. Forest habitats in a mixed urban-agriculture mosaic landscape: patterns of mammal occupancy. Landscape Ecol. 33, 59. https://doi.org/10.1007/s10980-017-0580-1. Ehlers Smith, Y.C., Ehlers Smith, D.A., Ramesh, T., Downs, C.T., 2019. Novel predators and anthropogenic disturbance influence spatio-temporal distribution of forest antelope species. Behav. Process. 159, 9–22. Ehlers Smith, Y.C., Ehlers Smith, D.A., Seymour, C.L., Thébault, E., van Veen, F.J.F., 2015. Response of avian diversity to habitat modification can be predicted from life history traits and ecological attributes. Landsc. Ecol. 30, 1225–1239. Ehlers Smith, D.A., Si, X., EhlersSmith, Y.C., Downs, C.T., 2018b. Seasonal variation in avian diversity and tolerance by migratory forest specialists of the patch-isolation gradient across a fragmented forest system. Biodivers. Conserv. 27, 3707–3727. Ehlers Smith, D.A., Si, X., Ehlers Smith, Y.C., Kalle, R., Ramesh, T., Downs, C.T., 2018c. Patterns of avian diversity across a decreasing patch-size gradient in a critically endangered sub-tropical forest system. J. Biogeog. 45, 2118–2132. Emerson, S.E., Brown, J.S., 2013. Identifying preferred habitats of samango monkeys (Cercopithecus (nictitans) mitis erythrarchus) through patch use. Behav. Processes 100, 214–221. Enstam, K.L., Isbell, L.A., 2011. The guenons (Genus Cercopithecus) and their allies. Behavioral Ecology of poly-specific associations. In: Campbell, S.K., Fuentes, C.J., MacKinnon, A., Panger, M., Bearder, S. (Eds.), Primates in Perspective, 2nd ed. Oxford University Press, New York, pp. 252–274. Environmental Systems Research Institute [ESRI]. 2011. ArcGIS Desktop v10.2. Environmental Systems Research Institute, Redlands, CA. Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Ann. Rev. Ecol. Environ. Syst. 34, 487–515. Fahrig, L., 2017. Ecological responses to habitat fragmentation per se. Ann. Rev. Ecol. Environ. Syst. 48, 1–23. Fahrig, L., Arroyo-Rodríguez, V., Bennett, J.R., Boucher-Lalonde, V., Cazetta, E., et al., 2019. Is habitat fragmentation bad for biodiversity? Biol. Cons. 230, 179–186. Feld, C.K., da Silva, P.M., Sousa, J.P., de Bello, F., Bugter, F., et al., 2009. Indicators of biodiversity and ecosystem services: a synthesis across ecosystems and spatial scales. Oikos 118, 1862–1871. Fletcher Jr., R.J., Didham, R.K., Banks-Leite, C., Barlow, J., Ewers, R.M., et al., 2018. Is habitat fragmentation good for biodiversity? Biol. Cons. 226, 9–15. Gagic, V., Bartomeus, I., Jonsson, T., Taylor, A., Winqvist, C., et al., 2015. Functional identity and diversity of animals predict ecosystem functioning better than species based indices. Proc. R. Soc. B. Biol. Sci. 282, 2–8. Geldenhuys, C., MacDevette, D., 1989. Conservation status of coastal and montane evergreen forest. In: Huntley, B.J. (Ed.), Biotic Diversity in Southern Africa. Oxford University Press, Oxford, pp. 224–238. GeoTerraImage, 2014. The 2013-14 South African National Land-cover dataset. Data layer for download from: https://egis.environment.gov.za/national_land_cover_ data_sa. Gibson, L., Lynam, A.J., Bradshaw, C.J.A., He, F., Bickford, D.P., et al., 2013. Near complete extinction of native small mammal fauna 25 years after forest fragmentation. Science 341, 1508–1510. Goslee, S.C., Urban, D.L., 2007. The ecodist package for dissimilarity-based analysis of ecological data. J. Stat. Softw. 22, 1–19. Gower, J.C., 1966. Some distance properties of latent root and vector methods used in multivariate analysis. Biometrika 53, 325–338. Hector, A., Joshi, J., Lawler, S., Spehn, E., Wilby, A., 2001. Conservation implications of the link between biodiversity and ecosystem functioning. Oecologia 129, 624–628.

Hevia, V., Carmona, C.P., Azcárate, F.M., Torralba, M., Alcorlo, P., et al., 2016. Effects of land use on species and functional diversity: a cross taxon analysis in a Mediterranean landscape. Oecologia 181, 959–970. Humphries, B., Ramesh, T., Hill, T., Downs, C.T., 2016. Habitat use and home range of black-backed jackals (Canis mesomelas) on farmlands in the Midlands of KwaZuluNatal, South Africa. Afr. Zool. 51, 37–45. Kingdon, J., 1997. The Kingdon Field Guide to African Mammals. Academic Press, London. Krauss, J., Steffan-Dewenter, I., Tscharntke, T., 2003. Local species immigration, extinction, and turnover of butterflies in relation to habitat area and habitat isolation. Oecologia 137, 591–602. Laliberté, E., Legendre, P., 2010. A distance based framework for measuring functional diversity from multiple traits. Ecology 91, 299–305. Laurance, W.F., Goosem, M., Laurance, S.G.W., 2009. Impacts of roads and linear clearings on tropical forests. Trends Ecol. Evol. 24, 659–669. Lacher Jr, T.E., Davidson, A.D., Fleming, T.H., Gómez-Ruiz, E.P., McCracken, G.F., OwenSmith, N., Peres, C.A., Vander Wall, S.B., 2019. The functional roles of mammals in ecosystems. J. Mamm. 100, 942–964. Lawes, M.J., 1990. The distribution of the samango monkey Cercopithecus mitis erythrarcus Peters, 1852 and Cercopithecus mitis labiatus I Geoffroy, 1843) and forest history in southern Africa. J. Biogeog. 17, 669–680. Lawes, M.J., 2002. The forest ecoregion. In: Le Roux, J. (Ed.), The Biodiversity of South Africa: Indicators, Trends and Human Impacts. Struik Publishers, Cape Town, pp. 8–10. Lawes, M.J., Eeley, H.A.C., Piper, S.E., 2000a. The relationship between local and regional diversity of indigenous forest fauna in KwaZulu-Natal Province, South Africa. Biodivers. Conserv. 9, 683–705. Lawes, M.J., Mealin, P.E., Piper, S.E., 2000b. Patch occupancy and potential metapopulation dynamics of three forest mammals in fragmented Afromontane forest in South Africa. Conserv. Biol. 14, 1088–1098. Lawes, M.J., Eeley, H.A.C., Findlay, N.J., Forbes, D., 2007. Resilient forest faunal communities in South Africa: a legacy of palaeoclimatic change and extinction. J. Biogeog. 34, 1246–1264. Legendre, P., Lapointe, F.-J., Casgrain, P., 1994. Modeling brain evolution from behavior: a permutational regression approach. Evolution 48, 1487–1499. Lichstein, J.W., 2007. Multiple regression on distance matrices: a multivariate spatial analysis tool. Plant. Ecol. 188, 117–131. Loreau, M., 2000. Are communities saturated? On the relationship between alpha, beta and gamma diversity. Ecol. Lett. 3, 73–76. Lundin, O., Smith, H.G., Rundlof, M., Bommarco, R., 2012. When ecosystem services interact: crop pollination benefits depend on the level of pest control. Proc. R. Soc. B. Biol. Sci. 280, 2012–2243. MacArthur, R.H., Wilson, E.O., 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, NJ. Magrach, A., Laurance, W.F., Larrinaga, A.R., Santamaria, L., 2014. Meta-analysis of the effects of forest fragmentation on interspecific interactions. Conserv. Biol. 28, 1342–1348. Mason, N.W.H., de Bello, F., 2013. Functional diversity: a tool for answering challenging ecological questions. J. Veg. Sci. 24, 777–780. Midgley, J.J., Cowling, R.M., Seydack, A.H.W., van Wyk, G.F., 1997. Forest. In: Cowling, R.M., Richardson, D.M., Pierce, S.M. (Eds.), Vegetation of Southern Africa. Cambridge University Press, Cambridge, pp. 278–299. Miller-Rushing, A.J., Primack, R.B., Devictor, V., Corlett, R.T., Cumming, G.S., et al., 2019. How does habitat fragmentation affect biodiversity? A controversial question at the core of conservation biology. Biol. Cons. 232, 271–273. Moll, E.J., White, F. 1978. The Indian Ocean Coastal Belt. In: Werger, M.J.A. (Ed.), Biogeography and Ecology of Southern Africa. Junk, The Hague, pp. 561–598. Mucina, L., Rutherford, M.C., 2011. The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19. South African National Biodiversity Institute, Pretoria. Olivier, P.I., van Aarde, R.J., Lombard, A.T., 2013. The use of habitat suitability models and species-area relationships to predict extinction debts in coastal forests, South Africa. Divers. Distrib. 19, 1353–1365. Pardini, R., de Souza, S.M., Braga-Neto, R., Metzger, J.P., 2005. The role of forest structure, fragment size and corridors in maintaining small mammal abundance and diversity in an Atlantic forest landscape. Biol. Conserv. 124, 253–266. Pellerin, M., Saïd, S., Richard, E., Hamann, J.L., Dubois-Coli, C., Hum, P., 2010. Impact of deer on temperate forest vegetation and woody debris as protection of forest regeneration against browsing. For. Ecol. Manage. 260, 429–437. Pereira, H.M., Navarro, L.M., Martins, I.S., 2012. Global biodiversity change: the bad, the good, and the unknown. Ann. Rev. Environ. Resour. 37, 25–50. Prugh, L.R., Hodges, K.E., Sinclair, A.R.E., Brashares, J.S., 2008. Effect of habitat area and isolation on fragmented animal populations. Proc. Natl. Acad. Sci. U.S.A. 105, 20770–20775. Ramesh, T., Kalle, R., Downs, C.T., 2017. Space use in a South African agriculture landscape by the caracal (Caracal caracal). Eur. J. Wildl. Res. 63, 11. Ries, L., Fletcher, R.J., Battin, J., Sisk, T.D., 2004. Ecological responses to habitat edges: Mechanisms, models, and variability explained. Ann. Rev. Ecol. Evol. Syst. 35, 491–522. Rowe-Rowe, D.T., 1982. Home range and movement of black-backed jackals in an African Montane region. S. Afr. J. Wildl. Res. 12, 79–84. Safi, K., Cianciaruso, M.V., Loyola, R.D., Brito, D., Armour-Marshall, K., Diniz-Filho, J.A.F., 2011. Understanding global patterns of mammalian functional and phylogenetic diversity. Phil. Trans. R. Soc. Lond., B, Biol. Sci. 366, 2536–2544. Seufert, V., Linden née Heikamp, B., Fischer, F., 2010. Revealing secondary seed removers: Results from camera trapping. Afr. J. Ecol. 48, 914–922. Schooley, R.L., Branch, L.C., 2009. Enhancing the area–isolation paradigm: Habitat

11

Ecological Indicators 113 (2020) 106176

Y.C. Ehlers Smith, et al. heterogeneity and metapopulation dynamics of a rare wetland mammal. Ecol. Appl. 19, 1708–1722. Si, X., Baselga, A., Ding, P., 2015. Revealing beta-diversity patterns of breeding bird and lizard communities on inundated land-bridge islands by separating the turnover and nestedness components. PLoS ONE 10, e0127692. Si, X., Baselga, A., Leprieur, F., Song, X., Ding, P., 2016. Selective extinction drives species and functional alpha and beta diversities in island bird assemblages. J. Anim. Ecol. 85, 409–418. Skinner, J.D.J., Chimimba, C.T.C., 2005. The Mammals of the Southern African Sub-region. Cambridge University Press, Cambridge. Socolar, J.B., Gilroy, J.J., Kunin, W.E., Edwards, D.P., 2015. How should beta-diversity inform biodiversity conservation? Trends Ecol. Evol. 31, 67–80. Soininen, J., Heino, J., Wang, J., 2018. A meta-analysis of nestedness and turnover components of beta diversity across organisms and ecosystems. Glob. Ecol. Biogeog. 27, 96–109. Taylor, A.R., Knight, R.L., 2003. Wildlife responses to recreation and associated visitor perceptions. Ecol. Appl. 13 (4), 951–963.

Tilman, D., May, R.M., Lehman, C.L., Nowak, M.A., 1994. Habitat destruction and the extinction debt. Nature 371, 65–66. Turner, I.M., 1996. Species loss in fragments of tropical rain forest: a review of the evidence. J. Appl. Ecol. 33, 200–209. Turner, I.M., Corlett, R.T., 1996. Are fragments worth conserving? Reply from Turner, I.M. and Corlett, R.T. Trends. Ecol. Evol. 11, 507–508. Tscharntke, T., Klein, A.M., Steffan-Dewenter, I., Thies, C., 2005. Landscape perspectives on agricultural intensification and biodiversity – ecosystem service management. Ecol. Lett. 8, 857–874. Villeger, S., Mason, N.W., Mouillot, D., 2008. New multidimensional functional diversity indices for a multifaceted framework in functional ecology. Ecology 89, 2290–2301. White, F., 1978. The Afromontane region. In: Werger, M.J.A. (Ed.), Biogeography and Ecology of Southern Africa. Junk, The Hague, pp. 465–513. Whittaker, R.H., 1960. Vegetation of the Siskiyou Mountains, Oregon and California. Ecol. Monogr. 30, 279–338. Whittaker, R.H., 1972. Evolution and measurement of species diversity. Taxon 21, 213–251.

12