An ecological network is as good as a major protected area for conserving dragonflies

Biological Conservation 191 (2015) 537–545

Contents lists available at ScienceDirect

Biological Conservation journal homepage: www.elsevier.com/locate/bioc

An ecological network is as good as a major protected area for conserving dragonflies James S. Pryke a,⁎, Michael J. Samways a, Kathleen De Saedeleer a,b a b

Department of Conservation Ecology and Entomology, Stellenbosch University, P/Bag X1, Matieland 7602, South Africa L'Ecole de Biologie, Université Catholique de Louvain, Louvain-la-Neuve, Belgium

a r t i c l e

i n f o

Article history: Received 2 March 2015 Received in revised form 23 July 2015 Accepted 28 July 2015 Available online xxxx Keywords: Agro-forestry Insect conservation Marshes Odonata Ponds Pools Protected areas Wallows

a b s t r a c t Freshwaters are highly threatened ecosystems, with agro-forestry being a major threat to sub-tropical wetlands. In the Maputaland–Pondoland–Albany global biodiversity hotspot of South Africa, large-scale ecological networks (ENs) of remnant vegetation have been set aside with the aim of mitigating the adverse effects of plantation forestry. However, the effectiveness of these ENs for maintaining freshwater biodiversity, especially that of still waters, is poorly known. In response, we compare mud wallows of large mammals, ponds and small marshes in an EN with those in an adjacent World Heritage Site protected area (PA) as reference. For this comparison we used dragonfly adults in view of their effectiveness as bioindicators. A total of 47 species was recorded at 105 sites. The EN shared 74% of its species with the PA. However, equal numbers of range restricted species were recorded from the EN and the PA. Five species were recorded as particular to the EN and seven to the PA, probably due to habitat heterogeneity across this type of landscape. Pond size, habitat heterogeneity, elevation and dissolved oxygen were important determinants for species richness and diversity. Proximity of plantation trees had only a minor effect, and then only on species composition. Mud wallows were the poorest habitat in terms of dragonfly diversity, owing to the intense disturbance. Wallows, ponds and marshes were largely complementary in their species composition. Overall, the freshwater system in the EN was a good surrogate for that in the PA, indicating the effectiveness of these ENs for maintaining the dragonfly assemblage. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Freshwaters have seen high loss in biodiversity, mostly through intensive human activity (Geist, 2011; Holland et al., 2012; Strayer and Dudgeon, 2010). Threats to freshwater biodiversity include loss and degradation of habitat, pollution, overexploitation, invasion by alien species, water extraction, flow regulation and global warming (Amis et al., 2009; Brinson and Malvarez, 2002; Ott, 2010; Samways and Taylor, 2004). In sub-tropical wetlands, habitat modification, storage and abstraction of water for agriculture and forestry are major threats to freshwater integrity (Cubbage et al., 2010; Dye, 2012; Junk, 2002). In the relatively wet, eastern region of South Africa is the globally important Maputaland–Pondoland–Albany Hotspot (Mittermeier et al., 2004), which is rich in biodiversity yet coincides with one of the main timber production areas of the country. Protected areas (PAs) are important for conserving the endemic and threatened biota of the region, as well as for protection of ecosystem services such as the supply and improved quality of freshwater. On the other hand, the demand for wood continues to rise globally, with plantation forestry responding to ⁎ Corresponding author. E-mail address: [email protected] (J.S. Pryke).

http://dx.doi.org/10.1016/j.biocon.2015.07.036 0006-3207/© 2015 Elsevier B.V. All rights reserved.

that demand (Cubbage et al., 2010). This means that a balance between timber production on the one hand, and preservation of wetlands on the other, needs to be found (Brinson and Malvarez, 2002; Vörösmarty et al., 2010). A solution to realizing this balance has been the development of large-scale ecological networks (ENs) (Samways et al., 2010). These are interconnected set-aside corridors of remnant indigenous vegetation that connect high value conservation areas, and have become increasingly important in biodiversity conservation and ecosystem management (Jongman et al., 2004). ENs are a major component of the Maputaland–Pondoland–Albany hotspot timber areas, amounting to roughly a third of the overall landscape, with the aim of mitigating the adverse effects of plantation compartments on local biodiversity and ecosystem processes. ENs are designed and managed to supply suitable habitats for organisms and to encourage their movement within the production landscape (Bazelet and Samways, 2011) based on principles of maintaining habitat heterogeneity and rare species (Pryke and Samways, 2015), as well as for maintaining ecological integrity and resilience. ENs also improve connectivity between PAs (Opdam et al., 2006) and extend the effectiveness of PAs (Pryke and Samways, 2012a, 2012b). However, when considering ENs as a tool for conservation, we need evidence of their efficacy and efficiency for maintaining biodiversity (Boitani et al., 2007).

538

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545

Since it is not possible to monitor the whole ecological community, adequate ecological indicators, representative of the structure, function, and composition of ecological systems are needed. These are required even if the relationship between potential indicator species and total biodiversity is still not well established owing to the complexity of ecological systems (Dale and Beyeler, 2001). Dragonflies are good ecological indicators for wetland and river quality assessment, as they are well known taxonomically and easier to identify than many other taxa, especially in the case of the adults (Chovanec et al., 2014; Kutcher and Bried, 2014; Oertli et al., 2005; Simaika and Samways, 2010). Although the sampling of exuviae is the only way to determine that the water body has allowed the breeding success of a dragonfly species (Raebel et al., 2010), in an African context, sampling of dragonfly adults has many advantages over the sampling of the earlier life stages (Bried et al., 2012a). Dragonfly species assemblages are also large enough for complex compositional assessments, and often comprise generalist as well as specialist species, the latter being important indicators of change (Simaika and Samways, 2009a). Here we use “dragonfly” for the whole order Odonata, unless otherwise specifically stated. They respond strongly to changes in water body conditions, especially vegetation structure (de Paiva Silva et al., 2010; Samways and Sharratt, 2010). Amis et al. (2009) commented that biodiversity conservation measures are most efficient where both the freshwater and terrestrial priority zones overlap. Therefore, the whole dragonfly assemblage is useful for providing an indication of the quality of freshwaters and their adjacent terrestrial habitats (Declerck et al., 2006) and they are also useful for selecting new areas for protection (Simaika and Samways, 2009b) and identifying important landscape characteristics (Raebel et al., 2012a). Although ENs have been implemented to offset afforestation, there is little information on whether the freshwater fauna is benefitting from this large-scale intervention. Kietzka et al. (2015) showed that conserving a variety of ecological gradients, i.e. high habitat heterogeneity, leads to increased species richness. Yet we still do not know how well ENs conserve dragonfly diversity compared to the natural standard of a PA. ENs are also a way to address the challenge articulated by Biggs et al. (2005) that we need to find ways to establish long-lasting improvements for establishing networks of conservation ponds. As dragonflies are good indicators of freshwater condition, this study hypothesizes that dragonfly diversity and assemblage structure are the same in the EN (within the agro-forestry mosaic) as in an adjacent reference area, a major PA (iSimangaliso Wetland Park, Zululand, South Africa). We focus on ponds and small marshes, as they were commonly present in both the PA reference area and in the EN, and are important water bodies contributing to local and regional biodiversity (Briers and Biggs, 2005; Davies et al., 2008a,2008b; Kadoya et al., 2004; Raebel et al., 2012a).

by wooded vegetation (Hart et al., 2014). Freshwater areas studied here were 1) mud wallows created by the large mammals (particularly African elephant, white rhinoceros and African buffalo) (Fig. 2(i), (ii)), 2) ponds (Fig. 2 (iii)), and 3) marshes (Fig. 2 (iv)). However, in reality, these three divisions were a spectrum with ‘mud wallows’ being ponds with little vegetation, ‘ponds’ being those water bodies with a distinct open water zone (Oertli et al., 2005) and well-vegetated margin, and ‘small marshes’ being water bodies with a complete cover of emergent macrophytes/floating vegetation. All water bodies that could be located were sampled, with the greatest distance between any two sites was 27 km and the closest distance was ca. 30 m. 2.2. Dragonfly sampling EN sampling days were alternated with PA sampling days to avoid temporal bias. Adult dragonfly sampling took place during the rainy season, when daily temperatures and water availability were at their highest. However, dragonflies were sampled only on sunny days between 14–26 February 2013, from 07 h30 to 15 h30. Unaided visual sampling is 100% accurate for Anisoptera and 80% accurate for Zygoptera when sampling species presence and abundance (Moore, 1991). It is even more accurate for Zygoptera when using close focus binoculars (Samways and Sharratt, 2010), as we did here. This was a species presence/absence study comparing dragonfly species composition of EN and PAs at the height of the unimodal subtropical flight season. To achieve this EN/PA comparison expediently without invoking changing conditions, the aim was to cover as many sites as possible at the height of the flying season i.e. an emphasis on spatial replication. This means that we only sampled each site once. This was done by two observers recording all adult dragonfly individuals encountered along a N20-m linear transect (each observer out and back = 80 m in total) around the margin of a wallow or a pond, or across a marsh. Species of doubtful identity were captured for later identification. Sampling was considered complete once a 10-min period had elapsed when either observer had not encountered any new species (Moore, 1991). The total average time spent recording per site was 47 min, which has previously been shown to be adequate for recording species presence in this region (Kietzka et al., 2015) and more than the 20–40 min recommended by Bried et al. (2012b). At ponds in South Africa, dragonfly species do not appear to have assembly rules i.e. species assemble based on habitat characteristics rather than which species are already present (Osborn and Samways, 1996). Sampling continued until a near asymptote of total species richness was reached at 105 sampling sites. A total of 47 species were recorded with Chao2 estimate of 56.7 (+/− 10.27) species and Jackknife2 estimate of 58.86 species. 2.3. Environmental variables

2. Sites and methods 2.1. Sites This study was conducted in the KwaZulu-Natal Province, South Africa (28°18′S, 32°24′E) (Fig. 1). A total of 105 lentic freshwater sites were sampled, of which 47 were in the EN (part of SiyaQhubeka New Generation Plantation) between Eucalyptus spp. plantation compartments (Fig. 1). A further 58 sites were sampled inside iSimangaliso Wetland Park, a major PA and a World Heritage Site. The iSimangaliso park is a 332 000 ha park that stretches from the town of St Lucia to the Mozambique border (Hart et al., 2014). Sampling here was restricted to the Western Shores section of the park in a ca. 30 km belt that runs between a commercial forestry areas and the protected area (Fig. 1). There was no fence between the PA and EN, which allowed indigenous large mammals to roam between the two. This area is a flat coastal landscape with many ponds and wetlands but few streams. The few lotic areas in this landscape are in the protected area and are dominated

For each site, the following continuous environmental variables (EVs) were measured: water temperature, dissolved oxygen and conductivity with an YSI 556 MPS instrument. Cloud percentage coverage was estimated, time of day recorded, and the distance of each site from the nearest plantation compartment edge was measured using QGIS 2.0 (QGIS Development Team, 2015). Categorical EVs were: whether the water body was in an EN or PA, the water body type (pond (a discrete water body with open water in the centre and a fringe of emergent vegetation) vs marsh (a wetland of continuous emergent vegetation) vs mud wallow (a muddy pool where large mammals had recently wallowed)), and water body size (N20 m diameter was considered large, b20 m diameter was considered small). Sala et al. (2004) and Carchini et al. (2007) were used as guidelines for measuring the vegetation component of the sites. However, as the various water bodies were not necessarily distinct from each other, and as the area was prone to grazing, trampling and wallowing by large mammals, the categories had to be adapted to local conditions

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545

539

Fig. 1. Map of the study area, showing the extent of the sampling both in the ecological networks and the protected area (PA). The grey shapes are the timber production compartments and the PA is to the east of these blocks. Points represent water bodies that were ponds (circles), marshes (squares) or megaherbivore wallowing holes (triangles), in the ecological networks (grey) and protected areas (black), as well as large (N / =20 m) (solid fill) and small (b20 m) (open) water bodies.

(Fig. 2(i)). Firstly, and for strictly comparative purposes, no naturally forested habitats were included in this study. All the water bodies here were composed only of reeds, grasses, sedges and water lilies. However, in places these were disturbed by large mammal activity, resulting in muddy wallows. The vegetation of the relatively undisturbed ponds and marshes was classified into percentages of 1) reeds, 2) grasses, 3) sedges, and 4) water lilies, to the nearest 10%. As these figures were dependent on amount of open water and of crushed vegetation (mud wallows), each site was also classified into the percentage of vegetation cover. 2.4. Statistical analyses Analyses of species richness were carried out using the lme4 package (Bates and Sarkar, 2007), within R software (R Development Core Team, 2013). Aikake Information Criterion (AIC) analyses were conducted to determine the best fit general linear mixed models (GLMMs) for the

species richness data. The GLMMs included the random effect of sampling day, time of sampling and cloud cover, as well as longitude and latitude to build spatial features into the model, while all other EVs were considered fixed effects. These data fitted a Poisson curve when a Likelihood Ratio Test was performed, thus a GLMM fit by a Laplace approximation and with a Poisson distribution was used (Bolker et al., 2009). Further analyses on Post-hoc analyses were performed on the water body type using a Tukey post-hoc test in the R package multcomp (Hothorn et al., 2008). These analyses were also conducted to see how the three water body types (ponds, marshland and wallows) responded to the variables independently, and how Anisoptera (dragonflies) and Zygoptera (damselflies) responded separately. To determine the differences in dragonfly assemblage composition response to categorical EVs, we used Permutational multivariate analyses of variance (PERMANOVA) in PRIMER 6 (PRIMER-E, 2008). These were performed to determine F- and p-values, as well as pairwise difference within tests, using 9999 permutations to assess changes in response to

540

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545

Fig. 2. (i) A white rhinoceros creating a wallow, (ii) a typical mud wallow in the ecological networks, (iii) a pond in the ecological networks (with the eucalyptus compartment margin in the background) and (iv) marsh in the protected area.

each of the EVs. Analyses were performed using Bray–Curtis similarity measures assessing the specific species using presence/absence data (Anderson, 2001). Elevation, the most influential spatial factor was included as a random effect. As the coefficient of variation among species was high (N 100%) and location and dispersion effects could be influencing these data (Warton et al., 2012), thus a second set of PERMANOVAs were produced from the 13 most common species (the maximum number of species to have a coefficient of variation below 50%). Dragonfly compositional responses to continuous EVs were determined by Canonical Correspondence Analysis (CCA), using CANOCO version 5 software (ter Braak and Šmilauer, 2012). CCA is considered a robust method, and accommodates skewed species distributions, interrelated EVs and incomplete environmental measurements (Palmer, 1993). Forward selection was used to rank EVs. Monte Carlo permutation tests, using 499 unrestricted random permutations, were performed to test the significance of the EVs on species distribution patterns. These tests were also performed on the three water body types independently. The non-parametric species estimators of Chao2 and Jackknife2 were calculated using PRIMER 6 (PRIMER-E, 2008). To test for spatial autocorrelation between dragonfly species richness we used both a Moran's I test using the ape package (Paradis et al., 2004) and a Mantel using the ade4 package, both in R (Dray and Dufour, 2007). Both these tests are appropriate for testing autocorrelation within datasets (Rossi, 1996). 3. Results 3.1. Species richness and sampling adequacy A total of 47 species was recorded (34 Anisoptera and 13 Zygoptera), representing 29% of all 162 species known nationally, and 69% of those known previously from the whole of this large PA. Eight pond or marsh specialist species previously known from the region (Samways, 2008) were not sampled here, five of these have yet to be recorded on the Western Shores (Hart et al., 2014). So while three species from the local species pool were not recorded, we added a further 16 species never previously recorded on the Western Shores of iSimanagliso,

with two of these not previously sampled even in the park as a whole (Hart et al., 2014; Appendix 1). This indicates good sampling adequacy. Furthermore, species accumulation curves flattened with species estimates for ponds and marshes nearing the observed species richness (pond: S = 37, Chao2 = 38.6 ± 1.9, Jackknife2 = 39.3; marsh: S = 40, Chao2 = 46 ± 5.9, Jackknife2 = 51.8). The maximum number of observed species at one site was 17, while the minimum was 1.

Table 1 Comparisons of species richness in response to continuous and categorical environmental variables. Species richness figures represent Wald-χ2 results from a generalized linear mixed model (GLMM with Poison distribution). Test

All sites

Ponds

Marsh

Wallows

Continuous variables Size of water body Distance to plantation Elevation Water temperature Conductivity Dissolved oxygen pH Water lily % Reed % Sedge % Grass % Vegetation cover

12.78⁎⁎⁎ 1.96 7.08⁎⁎ 2.99 3.70 5.67⁎ 0.06 0.03 1.76 0.68 0.29 3.62

3.23 0.92 4.55⁎ 1.04 2.58 0.17 0.77 1.53 1.89 0.89 1.18 2.31

10.31⁎⁎ 5.22⁎ 2.61 1.74 2.39 2.77 0.51 4.28⁎ 0.68 5.09⁎ 0.37 9.80⁎

6.86⁎ 3.46 4.50 3.92 3.39 5.89 3.41 –a 5.19 3.38 3.46 5.29

Categorical variables Water body type (WBT) EN vs PA Size of water bodyb (SWB) WBT + EN vs PA WBT + SWB EN vs PA + SWB WBT+ EN vs PA + SWB

17.14⁎⁎⁎ 0.31 11.83⁎⁎⁎ 34.02⁎⁎⁎ 49.38⁎⁎⁎ 14.53⁎⁎ 67.52⁎⁎⁎

– 1.74 2.54 – – 3.38 –

– 0.40 13.44⁎⁎⁎ – – 14.06⁎⁎ –

– 9.53⁎⁎ –c – – –c –

a

No water lilies were found in wallows, thus they were excluded from the analyses. Categorical “Size of water body” refers to small water bodies (b20 m in diameter) compared to large water bodies (N/=20 m). c All wallows were less than 20 m in diameter. ⁎ = p b 0.05. ⁎⁎ = p b 0.01. ⁎⁎⁎ = p b 0.001. b

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545

The continuous EV which influenced species richness the most was water body size, which showed a strong positive correlation (Table 1; Appendix 2). Species richness also had a significantly positive correlation with dissolved oxygen and elevation (Table 1; Appendix 3). When the dragonfly species richness of water body types was analyzed independently, ponds, marshes and wallows showed a positively significant relationship to water body size (Table 1; Appendix 2). Species richness of marshes was also positively correlated with distance to plantation, percentage water lily cover, percentage sedge cover and the overall vegetation cover (Table 1; Appendix 3). Species richness also showed significant responses to the categorical EVs of water body type and water body size, as well as responses to the combinations of all the EVs (Table 1). When water bodies were evaluated independently, marshes showed a positive response to water body size, while ponds did not (Table 1). Generally, larger ponds and marshes had higher species richness compared to the smaller ponds, with the mud wallows having the lowest species richness (Figs. 3, 4). The only categorical EV which did not show a significant response to species richness was whether the water body was inside the EN or PA (Table 1, Fig. 3). When the Anisoptera and Zygoptera were analysed separately, the Anisoptera showed similar results to the overall assemblage with higher species richness in the ponds and larger water bodies (Table 4; Fig. 3). The Zygoptera (damselflies) only showed a species richness response to water body type, with higher species richness in the ponds and marshes compared to the wallows (Table 4; Fig. 3). Neither the Moran I test (observed coefficient = 0.029; p = 0.205) or the Mantel test (observed coefficient = 0. 076; p = 0. 059) were significant, thus species richness of dragonflies was not autocorrelated with distance between sites.

541

3.2. Species composition About 15% of all species (7/47) were observed only in the PA, and 11% (5/47) only in the EN, while 74% (35/47) of species were sampled in both. Of the EN or PA specialists, three were sampled in marshlands, one in a pond, five in ponds and marshes, while another one was sampled only in mud wallows (Appendix 1). Seven species were ‘species of special concern’ (SOSC) as they were either geographical range-restricted and/or threatened and/or endemic species. These species were Zosteraeschna minuscula, Agriocnemis gratiosa, Agriocnemis ruberrima, Gynacantha usambarica, Lestes uncifer, Orthetrum robustum and Urothemis luciana, L. uncifer and A. ruberrima were sampled in sites that had N 7 species, while the other rare species were in sites with N10 species. All these SOSCs, except U. luciana (only in the PA) and G. usambarica (only in the EN), were recorded from both EN and PA. The most important continuous EVs for determining species composition were distance to plantation compartments, elevation, percentage reed cover and percentage vegetation cover (Table 2). When water body types were analyzed separately, ponds were significantly influenced in species composition by elevation, while marshes were significantly influenced by distance to plantation, percentage reed cover, and percentage grass cover. In turn, wallows were significantly influenced by pH and percentage reed cover (Table 2). The categorical EVs and all the combinations of these EVs showed significant differences in species composition, even when separated into different water body types (Table 2; Fig. 5). Smaller water bodies showed particularly variable species composition (Fig. 5). Generally, mud wallows were different to marshes and ponds, while all water body types showed significant differences between locations in the EN or PA (Table 3). When Anisoptera and Zygoptera were compared separately, they both showed compositional differences between water body type and size, as well as whether the water body was in the EN or PA. 4. Discussion

Fig. 3. Pairwise comparisons of species richness between: water body type; ecological network (EN) and protected area (PA); water body size (small b 20 m; large = / N20 m) for all species combined (a) and dragonflies (open) and damselflies separately (grey) (b). Mean (±1 SE); different letters represent significant differences.

The high number of species recorded here (47, even without any forest and stream specialists) equates to nearly 0.81% of all globally known dragonfly species (Kalkman et al., 2008) and over a quarter of those known from South Africa, highlighting the region's status as a global biodiversity hotspot (Mittermeier et al., 2004). All species recorded here are known to be resident in the area (Samways, 2008). As in other studies, our results emphasize the high value of ponds as contributors to regional biodiversity (Briers and Biggs, 2005; Davies et al., 2008a; Kietzka et al., 2015; Indermuehle et al., 2004; Raebel et al., 2012b; Williams et al., 2003). Furthermore, and importantly, our results showed the equivalence of the protected area (PA) and the agroforestry mosaic, with its ecological network (EN), in terms of species richness (with ¾ of all species shared between them). EN vs PA was also the only categorical variable that showed no significant changes in species richness. The importance of the EN was underscored by one species of special concern (G. usambarica) being recorded in the EN but not in the PA. Interestingly, G. usambarica was recorded in an EN mud wallow, yet it is a forest specialist that oviposits in mud and may have been positively phototaxic to shading caused by the nearby plantation trees (Fig. 2 (iii)) and the muddiness of the wallow. The remaining one quarter of species not shared by the EN and PA can be accounted for by habitat heterogeneity across the overall landscape, which also has a strong influence on terrestrial insects in the same area (Pryke and Samways, 2015). Subtle changes in habitat type, especially in vegetation, are known to influence dragonfly species composition both in South Africa (Smith et al., 2007) and elsewhere (Declerck et al., 2006; Kadoya et al., 2004). Furthermore, seven species were recorded only in the PA, and five only in EN, which supports the landscape heterogeneity perspective. Nevertheless, there may have been a slight impoverishing effect from the plantation compartments (with seven species only found in the PA), which was a significant

542

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545

Fig. 4. Pairwise comparisons of species richness between: (i) water body type and ecological network (EN) and protected area (PA); (ii) water body type and water body size (small b 20 m; large = / N20 m); (iii) EN vs PA and water body size, (iv) water body type, ENs vs PA and water body size. Mean (±1 SE); different letters above bars represent significantly different means (5% level).

environmental variable for species composition. Overall however, even though there was some effect from the plantation trees, the EN in this agro-forestry mosaic supported a sufficient network of ponds to provide enough habitat to support a similar fauna to that in the PA, unlike the a situation in the United Kingdom were ponds in agriculture tend to be degraded (Biggs et al., 2005). Tropical Brazilian ponds and freshwater nodes have been shown to be of critical conservation value even though most remain in unprotected areas (de Marco et al., 2014). This emphasizes that in the tropics, the function of the freshwater systems is not necessarily compromised from being in a production or semi transformed landscape, although these systems still require careful management to prevent degradation (Maltchik et al., 2012). The important point here is that the EN is an effective mitigation measure for plantation forestry, and indeed is excellent for some of the rarest lotic species. The effect of habitat heterogeneity was seen in differences in species richness and composition of ponds and marshes, and in the different sizes of these two habitat types. Pond size is known to affect dragonfly species number in Japan (Kadoya et al., 2004), as well as in Italy, although this was an interrelated variable with pond depth and how long water has been present in the pond (Carchini et al., 2007). Interestingly, there was also an effect of inter-habitat heterogeneity, with ponds and marshes showing responses to a variety of environmental variables when analyzed separately. This suggests that it is not sufficient to simply conserve a few ponds and marshes and expect the regional species pool to be conserved, but rather to conserve a range of ponds and marshes to conserve all species in this landscape. It is important to note that the dragonflies and damselflies did not respond differently in this study, suggesting that both benefited from habitat heterogeneity, despite having different ecological niche preferences. We only focused on one

spatial scale, and this strong association with habitat heterogeneity may differ at different scales (Kadoya et al., 2008). Nevertheless, this importance of landscape-scale heterogeneity has been observed in other taxa, for example moths have higher species richness in more complex woodlands (Merckx et al., 2012), or grassland butterfly diversity is improved by the presence of rocky outcrops (Crous et al., 2013) and grasshopper composition changes across elevation gradients (Crous et al., 2014). In fact, natural landscape heterogeneity is considered the most important variable when planning ENs for terrestrial biodiversity (Pryke and Samways, 2015). The additional habitat heterogeneity provided by the EN, raises the question on how best to manage and design ponds and wetlands for better dragonfly conservation. Conserving the all water bodies no matter of the type or size would be the first conservation priority. Furthermore, large mammals (particularly hippos and elephants) are engineering these water bodies, creating channels and removing vegetation from the edge, provided their populations do not get too high, will they will continue to positively influence the dragonfly diversity (Samways and Grant, 2008). Finally, there is the option to create artificial ponds (Declerck et al., 2006; Raebel et al., 2012a) to reduce distance to water bodies, which has an influence on dragonfly communities (Raebel et al., 2012b). This will allow less mobile species to disperse more easily across a greater spatial scale. This would need to be done with a great deal of consideration and as additional water bodies can alter distribution of mammal herbivores (Smit et al., 2007). The landscape studied here is topographically flat and dominated by lentic habitats, particularly ponds and marshes. We found most of the species we would expect to find in ponds and marshes in the area, yet there are a further 22 species previously recorded from this general

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545 Table 2 Differences in species composition in relation to environmental variables. Values represent a Pseudo-F value calculated using a CCA for the continuous variables and a PERMANOVA for the categorical variables. Test Continuous variables Size of water body Distance to plantation Elevation Water temperature Conductivity Dissolved oxygen pH Water lily % Sedge % Reed % Grass % Vegetation cover

All sites

Pond

Marsh

Wallow

1.9 2.5⁎⁎ 1.7⁎⁎ 1.4 1.6 1.4 0.9 1.3 1.3 3.5⁎⁎ 1.2 2.4⁎⁎

1.3 1.3 1.5⁎ 1.4 0.8 1.1 0.4 0.9 0.8 1.3 1.1 0.9

1.4 2.6⁎⁎⁎ 1.4 1.1 2.7 1.2 1.3 0.6 0.8 3.6⁎⁎ 2.1⁎ 1.3

0.9 1.6 0.9 1.2 0.7 0.8 1.7⁎ –a 1.0 3.2⁎ 0.9 1.2

Categorical variables — all species Water body type (WBT) 7.22⁎⁎⁎ EN vs PA 6.38⁎⁎⁎ 13.40⁎⁎⁎ Size of water bodyb (SWB) WBT + EN vs PA 4.69⁎⁎⁎ WBT + SWB 5.68⁎⁎⁎ EN vs PA + SWB 7.60⁎⁎⁎ WBT+ EN vs PA + SWB 3.80⁎⁎⁎

– 2.00⁎ 5.89⁎⁎⁎ – – 2.85⁎⁎⁎ –

– 3.36⁎⁎ 2.19⁎ – – 2.11⁎⁎ –

– 2.49⁎ –c – – –c –

Categorical variables — 13 most common species Water body type (WBT) 9.45⁎⁎⁎ EN vs PA 10.88⁎⁎⁎ b 15.27⁎⁎⁎ Size of water body (SWB) WBT + EN vs PA 5.77⁎⁎⁎ WBT + SWB 5.62⁎⁎⁎ EN vs PA + SWB 12.39⁎⁎⁎ WBT+ EN vs PA + SWB 3.33⁎⁎⁎

– 2.60⁎ 8.59⁎⁎⁎ – – 3.73⁎⁎⁎ –

– 3.98⁎⁎ 2.50⁎ – – 2.13 –

– 3.01⁎ –c – – –c –

a

No water lilies were found in wallows, thus they were excluded from the analyses. Categorical “Size of water body” refers to small water bodies (b20 m in diameter) compared to large water bodies (N/=20 m). c All wallows were less than 20 m in diameter. ⁎ = p b 0.05. ⁎⁎ = p b 0.01. ⁎⁎⁎ = p b 0.001. b

region which are not captured in the dataset here (Hart et al., 2014). Most of these species occur only in lotic habitats, especially wooded streams (Samways, 2008; Hart et al., 2014). Wooded streams are restricted to the PA–EN boundary and the PA, with very few streams in the EN. This emphasizes the importance of streams as well as lentic water bodies in iSimangaliso Wetland Park for conserving Odonata diversity in this region. The main reason for the lower species richness found in response to water body type was as a result of mud wallows always being poorer in species than marshes and ponds. This is probably due to the highly disturbed nature of these wallows from constant trampling and mudbathing by large mammals (Fig. 2(i), (ii)), and such impoverishment has been recorded in Europe as a result of cattle trampling (Declerck et al., 2006, but see Raebel et al., 2012b) and in Canada as a result of cattle grazing (Foote and Hornung, 2005). Furthermore, these wallows were small in size and had almost no vegetation. This is consistent with the results of Kadoya et al. (2004), Smith et al. (2007) and Raebel et al. (2012b), who showed that freshwater quality and thus dragonfly species diversity, results, among other variables, from the richness and structural complexity of the vegetation which increases the microhabitat diversity. In addition, the oftentimes temporal character of wallows only permits rapidly growing (as larvae) and readily dispersing species to complete a full life cycle e.g. Pantala flavescens. Hydroperiod has been shown to have a strong effect on ponds in Italy and the United Kingdom (Bella Della et al., 2005; Raebel et al., 2012b). Characteristically, the wallows occurred more frequently inside the EN than in the PA, suggesting that large animal disturbance was overall more intense in the EN than PA, which may have also contributed to some

543

Table 3 Results of pairwise PERMANOVAs using a combination of water body type, water body size and whether the water body was in ecological networks (EN) or protected area (PA). Test

t

Type Marsh; pool Marsh; wallow Pool; wallow

EN vs PA + water body size 1.01 EN-Large; EN-Small 3.59⁎⁎⁎ PA-Large; PA-Small 3.08⁎⁎⁎ EN-Large; PA-Large EN-Small; PA-Small

Test

Water body type + EN vs PA Pond-EN; Pond-PA Marsh-EN; Marsh-PA Wallow-EN; Wallow-PA Pond-EN; Marsh-EN Wallow-EN; Pond-EN Wallow-EN; Marsh-EN

1.48⁎ 1.66⁎ 1.80⁎⁎ 0.85 3.36⁎⁎⁎ 2.85⁎⁎⁎

Pond-PA; Marsh-PA Wallow-PA; Pond-PA Wallow-PA; Marsh-PA

0.91 1.32 0.79

t 3.32⁎⁎⁎ 1.88⁎⁎ 2.12⁎⁎⁎ 2.14⁎⁎⁎

Water body type + EN vs PA+ water body size Pond-EN-Large; Pond-PA-Large 1.66⁎⁎ Pond-EN-Small; Pond-PA-Small 0.88 Marsh-EN-Large; Marsh-PA-Large 1.52⁎ Marsh-EN-Small; Marsh-PA-Small 1.25 Wallow-EN-Small; 1.80⁎⁎ Wallow-PA-Small

Pond-EN-Large; Marsh-EN-Large Pond-EN-Small; Marsh-EN-Small Wallow-EN-Small; Pond-EN-Small Water body type + water body size Wallow-EN-Small; Marsh-EN-Small Pond-Large; Pond-Small 2.31⁎⁎⁎ Pond-PA-Large; Marsh-PA-Large Marsh-Large; Marsh-Small 1.79⁎⁎ Pond-PA-Small; Marsh-PA-Small Pond-Large; Marsh-Large 1.07 Wallow-PA-Small; Pond-PA-Small Pond-Small; Marsh-Small 1.31 Wallow-PA-Small; Marsh-PA-Small Wallow-Small; 1.99⁎⁎ Marsh-Small Wallow-Small; 1.99⁎⁎⁎ Pond-EN-Large; Pond-EN-Small Pond-Small Pond-PA-Large; Pond-PA-Small Marsh-EN-Large; Marsh-EN-Small Marsh-PA-Large; Marsh-PA-Small

1.10 0.92 2.31⁎⁎⁎ 1.40 0.77 1.08 0.92 0.77

1.73⁎⁎ 1.85⁎⁎ 1.56⁎ 1.36

⁎ = p b 0.05. ⁎⁎ = p b 0.01. ⁎⁎⁎ = p b 0.001.

lowering of dragonfly species richness in the EN compared to PA. Elephants have a particularly strong effect, with the elephant population in the iSimangaliso Wetland Park having a preference for the

Table 4 Comparisons of species richness and differences in species composition in relation to water body type and size and whether it was in a protected areaor the ecological network for the Anisoptera and Zygotera separately. Species richness figures represent Wald-χ2 results from a generalised linear mixed model (GLMM with Poisson distribution), while species composition represents a Pseudo-F value calculated using PERMANOVA. Test Species richness Anisoptera (dragonflies) Water body type EN vs PA Size of water body Zygoptera (damselflies) Water body type EN vs PA Size of water body Species composition Anisoptera (dragonflies) Water body type EN vs PA Size of water body Zygoptera (damselflies) Water body type EN vs PA Size of water body

All sites

Pond

Marsh

Wallow

6.99⁎ 0.01 10.82⁎⁎

– 2.11 3.03

– 0.28 10.60⁎⁎

– 7.50⁎⁎ –

20.89⁎⁎⁎ 1.09 2.78

– 0.01 0.01

– 0.49 4.64

– 5.13 –

6.52⁎⁎⁎ 8.40⁎⁎⁎ 14.94⁎⁎⁎

– 2.61⁎ 8.55⁎⁎⁎

– 3.72⁎⁎ 2.70⁎

– 3.25⁎⁎ –

10.90⁎⁎⁎ 5.91⁎⁎ 13.05⁎⁎⁎

– 0.68 0.99

– 2.40 3.54⁎

– 4.01⁎ –

“Size of water body” refers to small water bodies (b20 m in diameter) compared to large water bodies (N/=20 m). All wallows were less than 20 m in diameter. ⁎ = p b 0.05. ⁎⁎ = p b 0.01. ⁎⁎⁎ = p b 0.001.

544

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545

stimulate the timber production land stewards to continue to instigate ENs. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.biocon.2015.07.036. Acknowledgements We thank Mondi Global and the Global Change Grand Challenge funded by the National Research Foundation, South Africa (grant number: 81719) for financial support. We also thank Ezemvelo KZN Wildlife, iSimangaliso Wetland Authority, Mondi South Africa and SiyaQhubeka Plantations for permitting sampling on their holdings. KDS was supported by Programme de mobilité international Mercator de l'Université Catholique de Louvain funded by la Fédération Wallonie-Bruxelles via le Fonds d'aide à la mobilité étudiante. References

Fig. 5. Canonical analysis of principal coordinates ordination for the i) all sites and ii) large water bodies only, showing water bodies that were ponds (circles), marshes (squares) or megaherbivore wallowing holes (triangles), with those in the ecological network (grey) and protected areas (black), as well as large (N / =20 m) (solid fill) and small (b20 m) (open) in diameter.

plantation landscape as opposed to the PA (Jachowski et al., 2012). These wallows could be seen as ecological traps, with adults laying eggs in them and then the wallows drying up before the next generation can emerge (Raebel et al., 2010). However, these dragonfly species have reached a near evolutionary equilibrium in the presence of these megaherbivores. Evidence suggests that there is a suite of species that is able to sustain their populations in the face of such intense disturbance, possibly in part by conditions being unsuitable for their predators (including other dragonflies) (Samways and Grant, 2008).

5. Conclusions Ways are being sought to offset the adverse effects of agricultural production activities. One approach is to implement ENs as a land sparing approach which not only recognizes biodiversity conservation but also ecological processes such as hydrological cycles. Dragonflies embrace both these, being affiliated with water. Yet these insects are not the only organisms on the landscape that we invested here. Large mammals, as well as other forms of biodiversity, are also a priority, with dragonflies having to fit in with the natural impacts that some of the megaherbivores in particular have on the water systems. Added to this is the fact that there is remarkable heterogeneity among species of various taxa, including that of the dragonflies. Our results suggest that ENs are a highly effective means of dragonfly conservation, as well as that of various other, terrestrial fauna and also flora (Joubert and Samways, 2014). This is very encouraging, especially as the offset land given over to conservation in this agro-forestry mosaic must be demonstrated to be effective for biodiversity conservation so as to

Amis, M.A., Rouget, M., Lotter, M., Day, J., 2009. Integrating freshwater and terrestrial priorities in conservation planning. Biol. Conserv. 142, 2217–2226. Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecol. 26, 32–46. Bates, D.M., Sarkar, D., 2007. lme4: Linear mixed-effects models using S4 classes. R package version 0.99875-6. Bazelet, C.S., Samways, M.J., 2011. Identifying grasshopper bioindicators for habitat quality assessment of ecological networks. Ecol. Indic. 11, 1259–1269. Bella Della, V., Bazzanti, M., Chiarotti, F., 2005. Macroinvertebrate diversity and conservation status of Mediterranean ponds in Italy: water permanence and mesohabitat influence. Aquat. Conserv. 15, 583–600. Biggs, J.T., Williams, P., Whitfield, M., Nicolet, P., Weatherby, A., 2005. 15 years of pond assessment in Britain: results and lessons learned from the work of Pond Conservation. Aquat. Conserv. 15, 693–714. Boitani, L., Falcucci, A., Maiorano, L., Rondinini, C., 2007. Ecological networks as conceptual frameworks or operational tools in conservation. Conserv. Biol. 21, 1414–1422. Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Stevens, M.H.H., White, J.S.S., 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends Ecol. Evol. 24, 127–135. Bried, J.T., D'Amico, F., Samways, M.J., 2012a. A critique of the dragonfly delusion hypothesis: why sampling exuviae does not avoid bias. Insect Conserv. Divers. 5, 398–402. Bried, J.T., Hager, B.J., Hunt, P.D., Fox, J.N., Jensen, H.J., Vowels, K.M., 2012b. Bias of reduced-effort community surveys for adult Odonata of lentic waters. Insect Conserv. Divers. 5, 213–222. Briers, R.A., Biggs, J.T., 2005. Spatial patterns in pond invertebrate communities: separating environmental and distance effects. Aquat. Conserv. 15, 549–557. Brinson, M.M., Malvarez, A.I., 2002. Temperate freshwater wetlands: types, status, and threats. Environ. Conserv. 29, 115–133. Carchini, G., Della Bella, V., Solimini, A.G., Bazzanti, M., 2007. Relationships between the presence of odonate species and environmental characteristics in lowland ponds of central Italy. Int. J. Limnol. 43, 81–87. Chovanec, A., Schindler, M., Waringer, J., Wimmer, R., 2014. The dragonfly association index (Insecta: Odonata) — a tool for the type-specific assessment of lowland river. River Res. Appl. http://dx.doi.org/10.1002/rra.2760. Crous, C.J., Samways, M.J., Pryke, J.S., 2013. Rockiness determines meso-scale conservation of butterflies in Afro-montane grassland. J. Insect Conserv. 18, 77–83. Crous, C.J., Samways, M.J., Pryke, J.S., 2014. Grasshopper assemblage response to surface rockiness in Afro-montane grasslands. Insect Conserv. Divers. 7, 185–194. Cubbage, F., Koesbandana, S., Mac Donagh, P., Rubilar, R., Balmelli, G., Morales Olmos, V., De La Torre, R., Murara, M., Hoeflich, V.A., Kotze, H., Gonzales, R., Carrero, O., Frey, G., Adams, T., Turner, J., Lord, R., Huang, J., MacIntyre, C., McGinley, K., Abt, R., Phillips, R., 2010. Global timber investments, wood costs, regulation, and risk. Biomass Bioenergy 34, 1667–1678. Dale, V.H., Beyeler, S.C., 2001. Challenges in the development and use of ecological indicators. Ecol. Indic. 1, 3–10. Davies, B.R., Biggs, J., Williams, P., Whitfield, M., Nicolet, P., Sear, D., Bray, S., Maund, S., 2008a. Comparative biodiversity of aquatic habitats in the European agricultural landscape. Agric. Ecosyst. Environ. 125, 1–8. Davies, B.R., Biggs, J., Williams, P., Lee, J.T., Thompson, S., 2008b. A comparison of the catchment sizes of rivers, streams, ponds, ditches and lakes: implications for protecting aquatic biodiversity in an agricultural landscape. Hydrobiologia 597, 7–17. de Marco, P., Nogueira, D.S., Correa, C.C., Vieira, T.B., Silva, K.D., Pinto, N.S., Bichsel, D., Hirota, A.S.V., Vieira, R.R.S., Carneiro, F.M., de Oliveira, A.A.B., Carvalho, P., Bastos, R.P., Ilg, C., Oertli, B., 2014. Patterns in the organization of Cerrado pond biodiversity in Brazilian pasture landscapes. Hydrobiologia 723, 87–101. de Paiva Silva, D., De Marco, P., Chaves Resende, D., 2010. Adult odonate abundance and community assemblage measures as indicators of stream ecological integrity: A case study. Ecol. Indic. 10, 744–752. Declerck, S., De Bie, T., Ercken, D., Hampel, H., Schrijvers, S., van Wichelen, J., Gillard, V., Mandiki, R., Losson, B., Bauwens, D., Keijers, S., Vyverman, W., Goddeeris, B., Meester, De, Brendonck, L., Martens, K., 2006. Ecological characteristics of small farmland ponds: associations with land use practices at multiple spatial scales. Biol. Conserv. 131, 523–532.

J.S. Pryke et al. / Biological Conservation 191 (2015) 537–545 Dray, S., Dufour, A.B., 2007. The ade4 package: implementing the duality diagram for ecologists. J. Stat. Softw. 22, 1–20. Dye, P., 2012. A review of changing perspectives on Eucalyptus water-use in South Africa. For. Ecol. Manag. 301, 51–57. Foote, A.L., Hornung, C.L.R., 2005. Odonates as biological indicators of grazing effects on Canadian prairie wetlands. Ecol. Entomol. 30, 273–283. Geist, J., 2011. Integrative freshwater ecology and biodiversity conservation. Ecol. Indic. 11, 1507–1516. Hart, L.A., Bowker, M., Tarboton, W., Downs, C.T., 2014. Species composition, distribution and habitat types of Odonata in the iSimangaliso Wetland Park, KwaZulu-Natal, South Africa and the associated conservation implications. PLoS ONE 9, e92588. Holland, R.A., Darwall, W.R.T., Smith, K.G., 2012. Conservation priorities for freshwater biodiversity: the key biodiversity area approach refined and tested for continental Africa. Biol. Conserv. 148, 167–179. Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous inference in general parametric models. Biom. J. 50, 346–363. Indermuehle, N., Oertli, B., Menetrey, N., Sager, L., 2004. An overview of methods potentially suitable for pond biodiversity assessment. Arch. Sci. 57, 121–130. Jachowski, D.S., Slotow, R., Millspaugh, J.J., 2012. Physiological stress and refuge behavior by African elephants. PLoS One 7, e31818. Jongman, R.H.G., Külvik, M., Kristiansen, I., 2004. European ecological networks and greenways. Landsc. Urban Plan. 68, 305–319. Joubert, L., Samways, M.J., 2014. Equivalence of grasslands in an ecological network and a World Heritage Site. Biodivers. Conserv. 23, 2415–2426. Junk, W.J., 2002. Long term environmental trends in tropical wetlands. Environ. Conserv. 29, 414–435. Kadoya, T., Suda, S.-I., Washitani, I., 2004. Dragonfly species richness on man-made ponds: effects of pond size and pond age on newly established assemblages. Ecol. Res. 19, 461–467. Kadoya, T., Suda, S.-I., Tsubaki, Y., Washitani, I., 2008. The sensitivity of dragonflies to landscape structure differs between life-history groups. Landsc. Ecol. 23, 149–158. Kalkman, V.J., Clausnitzer, V., Dijkstra, K.-D.B., Orr, A.G., Paulson, D.R., van Tol, J., 2008. Global diversity of dragonflies (Odonata) in freshwater. Hydrobiologia 595, 351–363. Kietzka, G.J., Pryke, J.S., Samways, M.J., 2015. Landscape ecological networks are successful in supporting a diverse dragonfly assemblage. Insect Conserv. Divers. 8, 229–237. Kutcher, T.E., Bried, J.T., 2014. Adult Odonata conservatism as an indicator of freshwater wetland condition. Ecol. Indic. 38, 31–39. Maltchik, L., Dalzochio, M.S., Stenert, C., Rolon, A.S., 2012. Diversity and distribution of aquatic insects in Southern Brazil wetlands: implications for biodiversity conservation in a Neotropical region. Rev. Biol. Trop. 60, 273–289. Merckx, T., Feber, R.E., Hoare, D.J., Parsons, M.S., Kelly, N.A., Macdonald, D.W., 2012. Conserving threatened Lepidoptera: towards an effective woodland management policy in landscapes under intense human land-use. Biol. Conserv. 149, 32–39. Mittermeier, R.A., Gil, P.R., Hoffman, M., Pilgrim, J., Brooks, T., Mittermeier, C.G., Lamoreaux, J., da Fonseca, G.A.B., 2004. Hotspots Revisited. Cemex, Mexico City, Mexico. Moore, N.W., 1991. The development of dragonfly communities and consequences of territorial behavior: a 27-year study on small ponds at Woodwalton Fen, Cambridgeshire, United Kingdom. Odonatologica 20, 203–231. Oertli, B., Biggs, J., Céréghino, R., Grillas, P., Joly, P., Lachavanne, J.-B., 2005. Conservation and monitoring of pond diversity: introduction. Aquat. Conserv. 15, 535–540. Opdam, P., Steingröver, E., van Rooij, S., 2006. Ecological networks: A spatial concept for multi-actor planning of sustainable landscapes. Landsc. Urban Plan. 75, 322–332. Osborn, R., Samways, M.J., 1996. Determinants of adult dragonfly assemblage patterns at new ponds in South Africa. Odonatologica 25, 49–58. Ott, J. (Ed.), 2010. Monitoring Climate Change with Dragonflies. Pensoft, Sophia, Bulgaria. Palmer, M.W., 1993. Putting things in even better order—the advantages of canonical correspondence-analysis. Ecology 74, 2215–2230. Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20, 289–290. PRIMER-E, 2008. PERMANOVA and PRIMER 6. PRIMER-E, Lutton, Ivybridge.

545

Pryke, J.S., Samways, M.J., 2012a. Ecological networks act as extensions of protected areas for arthropod diversity conservation. J. Appl. Ecol. 49, 591–600. Pryke, J.S., Samways, M.J., 2012b. Conservation management of complex natural forest and plantation edge effects. Landsc. Ecol. 27, 73–85. Pryke, J.S., Samways, M.J., 2015. Landscape heterogeneity crucial for rare species in ecological networks. Landsc. Ecol. 30, 597–607. QGIS Development Team, 2015. QGIS Geographic Information System. Open Source Geospatial Foundation (URL http://qgis.osgeo.org). R Development Core Team, 2013. R 3.0.1. R Development Core Team, Vienna, Austria. ISBN 3-900051-07-0. Raebel, E.M., Merckx, T., Feber, R.E., Riordan, P., Macdonald, D.W., Thompson, D.J., 2010. The dragonfly delusion: why it is essential to sample exuviae to avoid biased surveys. J. Insect Conserv. 14, 523–533. Raebel, E.M., Merckx, T., Feber, R.E., Riordan, P., Thompson, D.J., Macdonald, D.W., 2012a. Multi-scale effects of farmland management on dragonfly and damselfly asssemblages of farmland ponds. Agric. Ecosyst. Environ. 161, 80–87. Raebel, E.M., Merckx, T., Feber, R.E., Riordan, P., Macdonald, D.W., Thompson, D.J., 2012b. Identifying high-quality pond habitats for Odonata in lowland England: implications for agri-environment schemes. Insect Conserv. Divers. 5, 422–432. Rossi, J.-P., 1996. Statistical tool for soil biology. XI. Autocorrelogram and Mantel test. Eur. J. Soil Biol. 32, 195–203. Sala, J., Gascón, S., Boix, D., Gesti, J., Quintana, X.D., 2004. Proposal of a rapid methodology to assess the conservation status of Mediterranean wetlands and its application in Catalunya (NE Iberian Peninsula). Arch. Sci. 57, 141–152. Samways, M.J., 2008. Dragonflies and damselflies of South Africa. Pensoft, Sofia, Bulgaria. Samways, M.J., Grant, P.B.C., 2008. Elephant impact on dragonflies. J. Insect Conserv. 12, 493–498. Samways, M.J., Sharratt, N.J., 2010. Recovery of endemic dragonflies after removal of invasive alien trees. Conserv. Biol. 24, 267–277. Samways, M.J., Taylor, S., 2004. Impact of invasive alien plants on Red-Listed South African dragonflies (Odonata). S. Afr. J. Sci. 100, 78–80. Samways, M.J., Bazelet, C.S., Pryke, J.S., 2010. Provision of ecosystem services by largescale ecological networks. Biodivers. Conserv. 19, 2949–2962. Simaika, J.P., Samways, M.J., 2009a. An easy-to-use index of ecological integrity for prioritizing freshwater sites and for assessing habitat quality. Biodivers. Conserv. 18, 1171–1185. Simaika, J.P., Samways, M.J., 2009b. Reserve selection using Red Listed taxa in three global biodiversity hotspots: dragonflies in South Africa. Biol. Conserv. 142, 638–651. Simaika, J.P., Samways, M.J., 2010. Comparative assessment of indices of freshwater habitat conditions using different invertebrate taxon sets. Ecol. Indic. 676, 1–9. Smit, I.P.J., Grant, C.C., Devereux, B.J., 2007. Do artificial waterholes influence the way herbivores use the landscape? Herbivore distribution patterns around rivers and artificial surface water sources in a large African savanna park. Biol. Conserv. 136, 85–99. Smith, J., Samways, M.J., Taylor, S., 2007. Assessing riparian quality using two complementary sets of bioindicators. Biodivers. Conserv. 16, 2695–2713. Strayer, D.L., Dudgeon, D., 2010. Freshwater biodiversity conservation: recent progress and future challenges. J. N. Am. Benthol. Soc. 29, 344–358. ter Braak, C.J.F., Šmilauer, P., 2012. Canoco reference manual and user's guide: software for Ordination (version 5.0). Microcomputer Power, Ithaca, USA. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S.E., Sullivan, C.A., Reidy Liermann, C., Davies, P.M., 2010. Global threats to human water security and river biodiversity. Nature 467, 555–561. Warton, D.I., Wright, S.T., Wang, Y., 2012. Distance-based multivariate analyses confound location and dispersion effects. Methods Ecol. Evol. 3, 89–101. Williams, P., Whitfield, M., Biggs, J., Bray, S., Fox, G., Nicolet, P., Sear, D., 2003. Comparative biodiversity of rivers, streams, ditches and ponds in an agricultural landscape in Southern England. Biol. Conserv. 115, 329–341.