Ecological networks act as extensions of protected areas for arthropod biodiversity conservation

Authors


Correspondence author. E-mail: jpryke@sun.ac.za

Summary

1. Landscape-scale ecological networks (ENs) are composed of linear corridors and are widely used to mitigate the adverse effects of intensive land use. One drawback with ENs used for conservation is that being small or linear they result in more edge relative to interior than would be the case naturally. Furthermore, there is little evidence to date that ENs do conserve indigenous biodiversity.

2. Here, we use five arthropod taxa at many sites over two geographical areas within South Africa with different elevations and grassland types to test the conservation value of remnant grassland ENs in a plantation forestry context. In particular, the relative value of arthropod biodiversity in exotic plantation blocks, their edges and the interiors of ENs among the plantations were compared with those in neighbouring protected areas (PAs). We use the effects that the plantation blocks have on the adjacent PAs as a reference for comparing the ENs among the plantations. Arthropods were selected to represent biodiversity, as they are small, diverse, habitat sensitive, resource dependent, ecologically important and can be sampled in large numbers.

3. In total, 10 422 individuals from 244 species were sampled. Importantly, there were no significant differences in species richness, abundance or assemblage composition between EN interior zones and PA interior zones in both geographical areas.

4. Using earlier established edge zones of 32 m, we found that plantation blocks had the lowest species richness and abundance compared with either grassland edge zones (<32 m from the edge) or grassland interior zones (≥ 32 m from the edge).

5.Synthesis and applications. Ecological networks are established to conserve biodiversity in areas of intensive land use. Provided that ecological networks are wide enough (i.e. >64 m) to overcome edge effects, they can support similar levels and quality of arthropod biodiversity as protected areas. Remnant grassland ecological networks in agroforestry can provide natural finger-like extensions from neighbouring protected areas and therefore have conservation value.

Introduction

Ecological networks (ENs) are strips of remnant habitat designed to connect protected areas and other areas of high natural value across transformed landscapes (Jongman 1995; Samways, Bazelet & Pryke 2010). They aim to alleviate the effects of fragmentation of remnant natural areas, particularly in managed landscapes such as agriculture and plantation forestry (Jongman 1995). ENs consist mostly of linear corridors, often connected to protected areas (PAs) and ideally are sufficiently large to ensure connectivity between habitat patches for organism dispersal on evolutionary as well as on ecological time-scales (Beier & Noss 1998). However, ENs as a mitigation measure lead to more ‘edge’ than would occur naturally, as they are essentially long and narrow (Samways, Bazelet & Pryke 2010).

When putting ENs into practice, we first need to understand how biodiversity responds to edges between the transformed and natural areas. Understanding the edge effects between transformed landscapes and conservation areas is critical to conservation planning. This is particularly true in transformed landscapes and in areas under pressure from agriculture, where remaining natural areas are often fragmented (Sala et al. 2000).

Edge effects are caused by structural changes along the transformed natural edge (Cadenasso et al. 2003; Hamberg, Lehvävirta & Kotze 2009), as well as through changes in soil moisture and nutrients (Li et al. 2007). Over time, secondary effects such as roads and invasion by alien species can alter the micro-habitat, which then leads to further deterioration of habitat along the edge. The response of biodiversity to these changes along edges is often seen as a two-zone effect, with an edge zone that is influenced by the interface between a transformed area and a natural one, and an interior zone where species richness, abundance and assemblage composition are no longer influenced by the distance to the edge (Ries et al. 2004). Ideally, this natural interior zone should represent the landscape as if it was untransformed. However, there is often much variability in the width of the edge zone, and the situation can be very complex (Alignier & Deconchat 2011; Pryke & Samways 2012). Nevertheless, simple, robust edge zone sizes are required for practical conservation design and implementation.

Corridors within ENs also need to overcome this edge zone to be effective in their design. However, corridors are often simply defined as movement corridors for focal species (Hilty, Lidicker & Merenlender 2006). Yet, the aim of ENs is to conserve biodiversity, so they also need to include the inherent biological complexity of the whole ecosystem (Samways, Bazelet & Pryke 2010). To achieve this, a significant proportion of the interior zone of the corridor in the EN needs to be conserved. Furthermore, the conceptual framework for corridors dictates that they need to promote the attributes of being a conduit, habitat and source, but not a filter, barrier or sink (Hess & Fischer 2001). Following on from this, many theoretical frameworks and models based on landscape ecology and EN principles have been established (Linehan, Gross & Finn 1995; Gurrutxaga, Lozano & Barrio 2010; Hepcan & Özkan 2011). Yet only a few ENs have actually been implemented, most notably the Pan European Ecological Network (Jongman et al. 2011), the greenways in China (Yu, Li & Li 2006) and within the exotic timber plantations of South Africa (Samways, Bazelet & Pryke 2010).

Boitani et al. (2007) questioned the conservation value of ENs on the grounds that despite being used across many systems, there is as yet few data supporting their value to conservation. To successfully understand the connectivity value of ENs, we need to determine the relative biodiversity value of edges and EN corridor interiors relative to the PA edges and interiors. Here, we use five arthropod taxa in two different geographical areas (with different elevations and grassland types) to address the relative arthropod biodiversity values of plantation blocks, as well as the edges and interiors of ENs compared with those of PAs. The plantation blocks were exotic timber plantations adjacent to PAs in South Africa. With these results, we are then able to determine the effectiveness of EN corridors in increasing the spatial extent of PAs, through having add-on linear areas of quality interior habitat.

Materials and methods

Study area and design

We used the 500 000 ha of ENs set aside by the South African timber industry, of which 80% is remnant grassland (Kirkman & Pott 2002; Samways 2007). Plantation forestry using alien trees is a serious risk to local biodiversity, as blocks of exotic trees contain little indigenous biodiversity (Samways & Moore 1991; Pryke & Samways 2009; Bremer & Farley 2010). In response, ENs aim to mitigate the adverse effects of these plantation forestry blocks through improving connectivity between extensive natural habitats (Beier & Noss 1998; Samways, Bazelet & Pryke 2010). On average, one-third of any plantation is set aside for conservation purposes (mostly as ENs) to maintain natural hydrology and natural levels of biodiversity (Kirkman & Pott 2002). However, plantation blocks and adjacent grasslands have sharp edges, with edge effects, which present particular challenges for designing ENs in association with PAs.

We studied five different commercial plantations, in two different geographical areas. Three of these plantations were in the KwaZulu-Natal Midlands (termed here high-elevation sites; HES) and two plantations were in Northern Zululand (termed here low-elevation sites; LES), South Africa. The HES were dominated by threatened Midlands Mistbelt Grassland and Drakensberg Foothill Moist Grassland and were generally hilly and dominated by clay soils with rocky outcrops. These HES were burned annually and had few natural grazers still in the system. The LES were dominated by Maputaland Wooded Grassland (Mucina & Rutherford 2006) and were very flat compared with the HES and dominated by sandy soils. The LES were burned infrequently, and because of the lack of a fence between them and the neighbouring iSimangaliso Wetland Park (a World Heritage Site), there were many large animals (including the African elephant Loxodonta africana, white rhino Ceratotherium simum and African buffalo Syncerus caffer) that used these ENs. Exotic timber plantations are the most abundant form of land transformation in both these areas (Kirkman & Pott 2002). The three HES – Gilboa (29°16 S; 30°18 E; along with the neighbouring Karkloof Nature Reserve), Good Hope (29°40 S; 29°58 E) and Maybole (29°44 S; 30°15 E) – had similar elevations c. 1000–1700 m asl. The two LES – Nyalazi (28°39 S; 32°60 E) and DukuDuku (28°59 S; 32°42 E) – had elevations of c. 20–90 m asl and bordered iSimangaliso Wetland Park. Gilboa lies ±50 km from both Good Hope and Maybole, which were ±30 km from each other, while Nyalazi and DukuDuku were adjacent to each other, although had a combined length of c. 40 km. DukuDuku and Gilboa, the nearest HES and LES, were ±210 km from each other.

Transects were set out across two types of landscape mosaic: from plantation block into a grassland corridor of the EN, and from a plantation block into the adjoining PA, with the PA acting as the reference site. Each transect began inside the plantation block and ran across a grassland corridor within an EN. For the reference sites, the transect began from inside the plantation block and ran out across the grassland of a PA. Nine sampling stations were arranged along each transect, set out on a log2 scale. Three sampling stations were in the plantation block (32, 16 and 8 m from the edge), three in the edge zone (0, 8 and 16 m from the plantation block edge) and three in interior zones (32, 63 and 128 m from the plantation block edge) for both areas per focal arthropod group (Fig. 1). The zoning of the transect was based on previous work in the same area, which showed that even the most sensitive species showed no response to the edge at a distance >32 m (Pryke & Samways 2012). All plantation blocks were mature pine trees Pinus patula at the HES, and mature eucalypts Eucalyptus grandis at LES, representing the most common timber species used in these areas (Fig. S1). Eight replicated transects were set out for corridors, and eight for PAs, for both the HES and the LES, giving a total of 32 transects, 96 zones and 288 sampling stations. All field work was intensive to avoid seasonal effects until >10 000 individuals were sampled, and the focal species accumulation curve began to flatten out. Field work was carried out by three workers, with two working at any one time, during late summer between February and April 2009.

Figure 1.

 Conceptual diagram of the sampling transects used in this study showing each of the nine stations. The stations were grouped into three zones: plantation block, edge zone and interior zone. The maximum distance from the plantation block after which the most sensitive taxa showed no response to the plantation was established as 32 m (Pryke & Samways 2012), and this is the border between the edge and interior zone used here. Four sampling techniques were used at each station on the transect, namely: a 200-m diurnal search, a 100-m nocturnal search, two pitfall traps and 100 sweeps of a sweep net. The searches and sweep netting were conducted parallel to the plantations block-corridor/protected area boundary.

Arthropod sampling

At all nine stations along the 32 transects, we sampled arthropods using four sampling techniques: 200-m diurnal searches, 100-m nocturnal searches, two pitfall traps and 100 sweeps of a sweep net (Fig. 1). Target taxa were the Formicidae, Araneae, Orthoptera, Lepidoptera and Scarabaeidae. These taxa were chosen as they were the most abundant groups and represent various functional feeding groups. The Formicidae and Araneae are predominantly predatory groups (with some omnivorous and herbivorous species within the Formicidae), and their presence is an indication of other prey species being present. The Orthoptera are mostly herbivorous and Lepidoptera are mostly herbivorous and nectarivorous, and represent the plant species present. Scarabaeidae (predominately dung beetles) are detritivorous and specifically respond to mammal dung and fungi. These scarabaeid species can be seen as indicators of the presence of mammals and fungi. There is a range of mobility between these taxa, with the Lepidoptera, Orthoptera and Scarabaeidae being the strongest fliers, although some species within these taxa have small, localized ranges. The Formicidae and Araneae have smaller ranges than the other taxa, although the presence of all these species would indicate a response to resources present at the point locality.

Diurnal searches at all the stations targeted flying arthropods, between 10.00 and 15.00 h, on sunny, windless days. Nocturnal searches were carried out with search lights after 20.00 h, only on clear, windless nights. Both diurnal and nocturnal searches were conducted by one observer (J.S.P.) to reduce the observer bias. The observer walked 200 m for diurnal and 100 m for nocturnal sampling, parallel to the plantation edge, recording all focal arthropods. If a specimen was not identifiable in the field, it was captured and preserved for later identification. The two pitfall traps were one metre apart and each trap was 70 mm in diameter, which effectively captures many rare species of ants (Abensperg-Traun & Steven 1995) and spiders (Brennan, Majer & Moir 2005). Traps were half-filled with a 50% ethylene glycol solution (Woodcock 2005) and left open for 3 days at a time. Arthropods on vegetation at all stations were sampled using 100 sweeps of a 40-cm sweep net. One sweep was a single back and forth movement through the grass, rotated between three field workers. The sweep netting was also conducted parallel to the forest edge at each station of the transect. All individuals sampled were identified to species or morphospecies, and all spider vouchers are housed in the South African National Collection of Arachnida, ARC, Pretoria, South Africa, while the other vouchers are housed in the Stellenbosch University Entomological Collection, South Africa.

Statistical analyses

Mean species richness and abundance were calculated for the PA and corridor transects per area and for all arthropods together, and also for ants, spiders, grasshoppers, butterflies and scarab beetles separately. Further analyses determined similarities between the three zones along the transect (plantation block, edge zone, interior zone; Fig. 1). Generalized linear mixed models (GLMMs) were used to analyse species richness and abundance data (O’Hara 2009; Zuur, Elena & Elphick 2010) in SAS 9.2 (SAS Institute Inc., Cary, North Carolina, USA) with variables incorporated into the model: type of zone (plantation, edge or interior), geographical area and whether the transect was in a corridor or in a PA. Poisson distributions (with a log link function, as these are count data) were used for all data, as means were >5, and the minimum number of successes and failures were <5 (Bolker et al. 2009). As these analyses showed no overdispersion of variances compared with the models, Wald chi-square (Z), statistics were calculated using the penalized quasi-likelihood technique (Bolker et al. 2009).

Permutational multivariate analyses of variance (permanova) (Anderson 2001) in primer 6 (PRIMER-E 2008) were performed to determine t- and P-values using 9999 permutations to assess changes in assemblage composition between transects into corridors and into the PA per area for the overall focal arthropod assemblage and for the five target taxonomic groups independently. Analyses were also conducted for the three zones (plantation block, edge or interior) per transect type (EN corridor or PA) per area (HES or LES). Analyses were performed using Bray–Curtis similarity measures, with these data fourth-root transformed to reduce the weight of common species (PRIMER-E 2008).

Results

Comparisons of arthropod biodiversity in protected areas and ecological network corridors

When all sampled taxa were combined, there were no significant differences in species richness, abundance or assemblage composition between PA and EN corridor interior zones for either geographical area (Tables 1, 2, S1 and S2). This was by far the most significant result.

Table 1.   Total species richness and abundance values (in brackets) of overall and various taxa sampled at various locations. Abundance values are given in parentheses
LocationAll taxaAntGrasshSpiderButterScara
  1. Grassh, grasshopper; Butter, butterfly; Scara, scarab beetle; HES, high-elevation sites, LES, low-elevation sites; Pl, plantation blocks; Ed, edge zone (<32 m from the plantation edge); In, interior zone (≥32 m from the plantation edge).

All sites
 All244 (10 422)79 (7918)55 (521)50 (1185)40 (609)20 (189)
 Pl86 (1402)35 (1216)2 (2)25 (131)15 (37)9 (16)
 Ed209 (4260)63 (3001)48 (308)42 (538)38 (297)18 (116)
 In170 (4760)58 (3701)32 (211)36 (516)33 (275)11 (57)
Protected areas
 All185 (5112)64 (3633)34 (194)38 (763)33 (393)16 (129)
 Pl60 (378)24 (285)1 (1)16 (60)14 (24)5 (8)
 Ed149 (2179)44 (1414)31 (133)28 (363)32 (186)14 (83)
 In128 (2555)45 (1934)20 (60)29 (340)25 (183)9 (38)
Corridors
 All192 (5310)63 (4285)44 (327)40 (422)34 (216)11 (60)
 Pl54 (1024)27 (931)1 (1)15 (71)6 (13)5 (8)
 Ed154 (2081)49 (1587)36 (175)32 (175)28 (111)9 (33)
 In122 (2205)43 (1767)22 (151)26 (176)25 (92)6 (19)
HES
 All159 (5418)50 (4000)43 (420)37 (749)17 (197)12 (52)
 Pl45 (304)19 (233)1 (1)15 (65)5 (7)5 (8)
 Ed128 (2095)37 (1435)39 (243)29 (280)13 (107)10 (30)
 In104 (3019)37 (2342)24 (176)24 (404)13 (83)6 (14)
LES
 All141 (5004)51 (3918)17 (101)30 (436)34 (412)9 (137)
 Pl54 (1098)25 (993)1 (1)13 (66)11 (30)4 (8)
 Ed119 (2165)41 (1566)14 (65)25 (258)31 (190)8 (86)
 In91 (1741)29 (1359)10 (35)20 (112)27 (192)5 (43)
Table 2.   Permutational multivariate analysis of variance (permanova) calculating the differences in assemblage composition for all arthropods and the five focal arthropod taxa. Comparisons were between the stations in plantation blocks (Pl), those in edge zones (<32 m from plantation edge; Ed) and stations in the interior zone (≥32 m from plantation edge; In)
LocationAll taxaAntGrasshSpiderButterScara
  1. ≠ designates significantly different assemblage; = designates non-significant difference.

  2. Grassh, grasshoppers; Butter, butterfly; Scara, scarab beetle; PA, protected area; Co, corridor; H, high-elevation sited; L, low-elevation sites.

All sites
 PA vs. CoPA≠CoPA=CoPA≠CoPA≠CoPA≠CoPA=Co
  PlPA=CoPA=CoPA=CoPA=CoPA=CoPA=Co
  EdPA=CoPA=CoPA=CoPA≠CoPA=CoPA=Co
  InPA=CoPA=CoPA=CoPA=CoPA=CoPA=Co
 Pl vs. EdPl≠EdPl≠EdPl=EdPl≠EdPl≠EdPl=Ed
  PAPl≠EdPl≠EdPl=EdPl≠EdPl=EdPl=Ed
  CoPl≠EdPl≠EdPl=EdPl≠EdPl=EdPl=Ed
 Pl vs. InPl≠InPl≠InPl=InPl≠InPl≠InPl=In
  PAPl≠InPl≠InPl=InPl≠InPl=InPl=In
  CoPl≠InPl=InPl=InPl≠InPl=InPl=In
 Ed vs. InEd=InEd=InEd=InEd=InEd=InEd=In
  PAEd=InEd=InEd=InEd=InEd=InEd=In
  CoEd=InEd=InEd=InEd=InEd=InEd=In
High- vs. low-elevation sitesH≠LH≠LH≠LH≠LH≠LH≠L
High-elevation sites (HES)
 PA vs. CoPA≠CoPA=CoPA≠CoPA≠CoPA=CoPA=Co
  PlPA=CoPA=CoPA=CoPA=CoPA=CoPA=Co
  EdPA=CoPA=CoPA=CoPA≠CoPA=CoPA=Co
  InPA=CoPA≠CoPA=CoPA=CoPA=CoPA=Co
 Pl vs. EdPl≠EdPl≠EdPl=EdPl≠EdPl=EdPl=Ed
  PAPl≠EdPl=EdPl=EdPl=EdPl=EdPl=Ed
  CoPl≠EdPl≠EdPl=EdPl=EdPl=EdPl=Ed
 Pl vs. InPl≠InPl≠InPl=InPl≠InPl=InPl=In
  PAPl≠InPl≠InPl=InPl≠InPl=InPl=In
  CoPl≠InPl=InPl=InPl=InPl=InPl=In
 Ed vs. InEd=InEd=InEd=InEd=InEd=InEd=In
  PAEd=InEd=InEd=InEd=InEd=InEd=In
  CoEd=InEd=InEd=InEd=InEd=InEd=In
Low-elevation sites (LES)
 PA vs. CoPA≠CoPA=CoPA=CoPA=CoPA≠CoPA=Co
  PlPA=CoPA=CoPA=CoPA=CoPA=CoPA=Co
  EdPA≠CoPA=CoPA=CoPA=CoPA=CoPA=Co
  InPA=CoPA=CoPA=CoPA=CoPA=CoPA=Co
 Pl vs. EdPl≠EdPl≠EdPl=EdPl≠EdPl≠EdPl=Ed
  PAPl≠EdPl≠EdPl=EdPl≠EdPl=EdPl=Ed
  CoPl≠EdPl≠EdPl=EdPl≠EdPl=EdPl=Ed
 Pl vs. InPl≠InPl≠InPl=InPl≠InPl≠InPl=In
  PAPl≠InPl≠InPl=InPl=InPl=InPl=In
  CoPl≠InPl=InPl=InPl=InPl=InPl=In
 Ed vs. InEd=InEd=InEd=InEd=InEd=InEd=In
  PAEd=InEd=InEd=InEd=InEd=InEd=In
  CoEd=InEd=InEd=InEd=InEd=InEd=In

Looking at these results in more detail, corridor transects had a significantly higher abundance compared with those of PAs, although this was attributed to significantly more individuals in the plantation blocks of corridor transects (Fig. 2; Tables 1 and S2). There were also significant differences in overall species assemblage composition for the whole transect between corridor and PA, which was attributable to edge zone compositional differences in the LES (Table 2).

Figure 2.

 Summary results of generalized linear mixed models (with Poisson distribution) on means (1 ± SE) for stations along a transect from within a plantation block into the interior of a corridor (open bar) or protected area (grey bar). These were analysed for species richness in the high-elevation sites (a) and low-elevation sites (b) and abundance in the high-elevation sites (c) and low-elevation sites (d). Vertical, dashed line represents plantation block boundary (edge). Different letters above bars represent significantly different means (5% level).

The most species-rich and abundant group was ants, followed by grasshoppers, spiders, butterflies and scarab beetles (Table 1). Ants were more species rich and abundant in the corridor plantation blocks compared with PA plantation blocks (Tables S1 and S2). Grasshoppers were significantly more species rich and abundant in the corridor interior zone compared with the PA interior zone (Tables S1 and S2). Spiders showed a few significant differences in species richness between corridors and PAs (Table S1), although they were significantly more abundant in the PA interior zone compared with the corridor interior zone (Table S2). Butterflies showed non-significant differences in species richness between corridors and PAs (Table S1) and a higher abundance in the PA interior zone (Table S2). Scarab beetles showed higher species richness in the PA edge zone compared with corridor edge zone (Table S1) and higher abundance in PA interior zone compared with corridor interior zone (Table S2).

Ants and scarab beetles showed no significant differences in assemblage composition between PA and corridors for the whole transect or for any of the three zones (Table 2). Grasshoppers and butterflies showed significant compositional differences between PAs and corridors only for whole transects and not per zone (Table 2). Spiders showed significant differences between PAs and corridors for the whole transect and the edge zones (Table 2).

Arthropod biodiversity between the three zones along the transect

Species richness and abundance were significantly lower in plantation blocks compared with edge zones or interior zones of the transect (Fig. 2; Tables 1, S1 and S2). The distinct edge effect of 32 m can be seen at the HES, with the edge having significantly higher species richness (Fig. 2, Table 1), while LES showed fewer significant differences, although species richness peaked between 8 and 16 m from the plantation edge (Fig. 2). Assemblage composition was similar in edge and interior zones, but both zones were dissimilar to the plantation blocks (Table 2).

All groups showed significantly higher species richness and abundance in both edge and interior zones compared with plantation blocks (Tables 1, S1 and S2). Ant, spider, butterfly and scarab beetle species richness showed no significant differences between edge and interior zones (Tables 1 and S1). The abundance of ants was significantly higher in the edge zone compared with the interior zone, while spider and butterfly abundances showed non-significant differences (Table S2). Grasshopper species richness and abundance and scarab beetle abundance were significantly higher in the edge zones of both ENs and PAs compared with those in the interior zones (Tables S1 and S2).

The assemblage composition of all focal taxa was similar, when edge zones of both ENs and PAs were compared with interior zones (Table 2). Ants and spiders showed assemblage compositional dissimilarity between plantation blocks and edge zones or interior zones (Table 2). Grasshopper and scarab beetle assemblages showed no significant differences in composition between plantation block, edge zones or interior zones (Table 2). The butterfly assemblages were significantly different between plantation block, edge zones or interior zones (Table 2).

Discussion

Effectiveness of corridors in extending protected area size

The most important finding was that when interior zones of PAs and EN corridors were compared, there were no significant differences in species richness, abundance or assemblage composition for the whole assemblage between PAs and EN corridor interiors, in both geographical areas. This means that corridors are sufficiently natural (after exclusion of the 32-m edge zone) to be effective in increasing the area of the PAs. This also means that ENs with wide enough corridors can be seen as extensions of PAs (Fig. 3). Traditionally, ENs were established for movement between isolated fragments within transformed landscapes (Jongman 1995, Jongman 2004). However, our results suggest that wide corridors within the ENs add additional quality habitat to PAs as well as being movement corridors (Fig. 3).

Figure 3.

 A conceptual diagram illustrating the various functional types of ecological network corridors as protected area (PA) extensions within the transformed landscape. The plain grey area represent the transformed landscape (i.e. plantation blocks), the hashed area (light grey) represents the edge zone, and the clear area the interior zone of both corridors and PAs. The dotted line is the theoretical edge zone within the plantation block (although this is not relevant for conservation planning and so is excluded in this study). The three different corridors represent three different functions that corridors could have. Corridor type 1 is a connection between PAs, allowing organisms to move through the transformed landscape, while also being a PA extension as it has an interior zone which also allows indigenous organisms to use it as habitat per se. Corridor type 2 is only a movement corridor (but may also be of value to early successional species) as there is no interior zone for arthropod biodiversity conservation. However, it may be suitable for a few highly habitat tolerant species. Corridor type 3 represents an unconnected corridor that has value as a protected area extension and also as a future potential connection to other PAs.

Grasshoppers were significantly more species rich in the interior zones of EN corridors compared with PAs, probably due to their preference for edges. Ants were the only group to show assemblage compositional differences in the interior zone of corridors vs. inside PAs, but only at one of the two geographical areas. Four taxa showed significant abundance results, although this was in both directions, with spiders, butterflies and scarab beetles significantly preferring interior zones of PAs, while grasshoppers significantly preferred interior zones of corridors. Despite some taxa responding differentially to PA vs. corridor, there were fewer differences than expected. This suggests that for many taxa, corridors are similar to PAs >32 m from plantation blocks.

Edge effect size and overall biodiversity between different zones

There was a distinct 32-m edge zone (seen in changes in arthropod species richness and abundance) at both HES and LES. This verifies other studies in the area (Pryke & Samways 2012) and so an edge zone of 32 m has wide geographical and elevational applicability in this region and is here defined as <32 m from the plantation block, while the interior zone is ≥32 m from the plantation block.

The transformed plantation blocks had the lowest species richness and abundance in comparison with those of either edge or interior zones. However, edge zones tended to have higher species richness compared with that of interior zones of both EN corridors and PAs, while interior zones had higher abundance compared with that of edge zones. There were also significant differences in assemblage composition in plantation blocks compared with those of edge or interior zones, yet both edge and interior zones were similar. This is probably due to anthropogenic disturbances such as roads (Samways, Bazelet & Pryke 2010) or alien plants (Holway 2005), which allow generalist opportunistic species to enter the system (Didham et al. 1996) although here not necessarily to the demise of specialist species.

The <32-m edge used here is similar to the >25 m that ant species showed to tarred road boundaries in the same geographical area (Samways, Osborn & Carliel 1997) and to mesophytic forest urban boundaries (Ivanov & Keiper 2010). Ants here showed similar species richness in interior and edge zones, while being far more abundant in the interior zones. Grasshoppers were the only group to have significantly higher species richness in edge zones compared with interior zones, apparently due to the strong response by generalist and opportunistic species which influenced the composition of the whole assemblage. So, despite the grasshoppers responding strongly to management (Bazelet & Samways 2011), the edge effect was highly significant, suggesting that design features relating to edges are nevertheless highly influential for this group.

The scarab beetle assemblage was similar for all zones, with a significantly higher abundance in the edge zone. These beetles appear to be edge specialists in this agroforestry system. This is different from the forest-cerrado edges in Brazil, where habitat type is more of a determinant than edge (Durães, Martins & Vaz-de-Mello 2005). Here, this edge specialization is attributed to high small mammal abundance at the edges of plantation blocks (Wilson et al. 2010), and large animals resting in the shade of these areas during the heat of the day, so more dung is available at the edges. In contrast, both spiders and butterflies showed few differences between the edge and interior zones.

The similarity of edge zones to interior zones, particularly in species assemblage composition, is most likely due to the conservative nature of the 32-m edge zone. In other words, many species or taxa stopped experiencing edge effects at shorter distances than 32 m to the plantation block, and interior species were recorded as edge zone species. Nevertheless, in terms of conservation and forestry management a 32-m edge zone seems appropriate.

The significance of ECOLOGICAL NETWORKS in a wider context

Although we sampled only arthropod taxa, the results can be viewed in a wider taxonomic context. ENs (at distances >32 m) were similar to PAs for all taxa sampled. This response is likely to be consistent for many other taxa not sampled here. For example, a suite of plant species would need to be present for the sampled butterflies and grasshoppers, or certain mammal species for the dung beetles, suggesting that these plants and mammals are using the ENs similarly to the butterflies, grasshoppers and dung beetles. Furthermore, these results may have a wider application than just for these remnant ENs. We have shown here that once the edge effect is overcome, these ENs effectively become PAs. However, the size of the edge zones will change depending on the type of transformation. For example, intensive agriculture may have wider edge zones owing to spray drift from pesticide applications. The type of natural habitat to be conserved would also influence the size of the edge zone, for example tropical rain forests are highly sensitive to edge effects and so it is likely they would need larger edge zones (Durães, Martins & Vaz-de-Mello 2005). This means that our results are, to some extent, applicable globally, provided that the local size of the edge zones is first established.

Protected areas play a significant role for maintaining biodiversity at current levels (Butchart et al. 2010; Hoffmann et al. 2010). In practice, PAs need to be as large as possible to (i) buffer adverse conditions, (ii) reduce ecological relaxation (Tilman 1997) and (iii) encompass as much spatial heterogeneity as possible. Yet PAs are usually surrounded by a matrix appropriated by humans for agricultural and/or urban use. Extending PAs can then be done in two ways, which are not mutually exclusive: land sparing and/or land sharing (Phalan et al. 2011). ENs are a land sparing that benefits arthropod biodiversity and ecosystems processes locally, but importantly, they also act as an extension of a PA when connected to that PA. Therefore, ENs have an add-on value that is enormous, as it enables considerable conservation activity without compromising production. Quality ENs are therefore a serious way forward for effective conservation.

Conclusions

The conservation value of ENs is justifiably being questioned owing to lack of supporting data (Boitani et al. 2007). However, we confirm here that provided corridor width is great enough, corridors making up ENs have considerable biodiversity value, at least for arthropods. The concept of corridors and ENs is based on connectivity to enable organisms to move through the fragmented landscape (Hilty, Lidicker & Merenlender 2006). For arthropods, these concepts need to be put into perspective, especially as dispersal in most species is strongly linked to resource-searching behaviours (foraging, mate or lek location, etc.) (Baguette & Van Dyck 2007). Here, both highly and less mobile species from various functional feeding groups used the interior zones of corridors (i.e. ≥32 m from the plantation block) as though they were a habitat source, as in the PAs.

Arguably, the PAs that were used here as reference sites may themselves have experienced a loss of biodiversity because of the surrounding habitat transformation (Mairorano, Falcucci & Boitani 2008). This is particularly relevant as these sites were on the PA – plantation boundary and so may also have been influenced by the plantations. Nevertheless, PAs need to be included in an area-wide conservation plan and should not remain islands in a transformed matrix (Mairorano, Falcucci & Boitani 2008). ENs help here by effectively increasing the size of the PAs, improving connectivity across the transformed landscape.

The interiors of the EN corridors had similar arthropod biodiversity levels and assemblages composition to PAs, suggesting that corridors with widths >64 m had a profile similar to that of nearby PAs. The 32-m edge zone in this study appears to be a conservative estimate of grassland edge effects around timber plantation blocks. Thus, corridors of <64 m will be mainly edge and have conservation value only for certain species and as solely movement or early successional corridors, as with the butterflies. The 250 m suggested by Pryke & Samways (2001) is still appropriate, as this incorporates interior space for the more sensitive species. Following on from this point, an ecological network of larger habitat corridors as suggested by Samways, Bazelet & Pryke (2010) will certainly reduce the total amount of edge across the entire network. Conservation plans will also need to consider the 32-m edge zone around plantation blocks as transitional areas from transformed to natural conditions, and this would need to be taken into account when designing agroforestry landscapes to include conservation of biodiversity.

We used a multi-taxa approach to minimize the influence of one group on the overall result, yet all the responses for the various taxa were similar, which contrast with some other multi-taxa studies (Kotze & Samways 1999; Tropek, Spitzer & Konvicka 2008; Pryke & Samways 2010). This illustrates the similar and very strong influence that plantation blocks have on a whole range of taxa, creating highly significant edge effects into the natural grassland. As all these groups used corridors of the ENs as if they were PAs, we can be optimistic that many other groups, not studied here, may well also benefit from these ENs. Furthermore, deployment of large-scale ENs in other parts of the world in a production context is likely to have enormous benefit for biodiversity over much larger areas than currently realized.

Acknowledgements

We thank G. Schreiner, P Grant and R. Gaigher for field assistance; C. Burchmore, D. Burden, P. Gardiner, D. van Zyl, L. Shaw of Mondi for providing maps, accommodation at field sites, technical assistance and local knowledge. We also thank Ezemvelo KZN Wildlife, iSimangaliso Wetland Authority, Mondi South Africa, Mondi-Shanduka and SiyaQhubeka Plantations for permitting sampling on their holdings. Funding was from the Mondi Ecological Network Programme (MENP) and co-funding from the National Research Foundation of South Africa (NRF).

Ancillary