Journal list menu
Plant, herbivore and parasitoid community composition in native Nothofagaceae forests vs. exotic pine plantations
Abstract
- Converting natural areas into land used for production causes dramatic changes in the configuration of landscapes. Both the loss and fragmentation of native habitats contribute to biodiversity loss world-wide and the consequent creation of artificial edges can have a significant influence on community assembly. The conservation value of plantation forests has been identified for specific species, but it is not clear whether exotic pine plantations can also be used for the preservation of native communities in general.
- We studied whether community composition of different trophic levels (plants, herbivorous caterpillars, parasitoids) changed across native Nothofagaceae forests to exotic pine plantations, and whether habitat edges affected communities differently depending on the forest type considered. To accomplish this, we sampled plants, herbivorous caterpillars and parasitoids in native Nothofagaceae and exotic pine plantation forests and compared community composition of each trophic level across habitats.
- We found that community composition of plants, herbivorous caterpillars and parasitoids differed significantly between native and exotic plantation forests, and that variation in the composition of the upper trophic levels was strongly dependent on variation in the composition of the lower trophic level. Moreover, differences in community composition were mostly the result of species turnover, suggesting that plantations are complementary habitats for some species, but cannot be a substitute habitat for all native forest species. Furthermore, edge effects had a strong impact on the composition of native communities, such that certain species were only present in the interior of the native habitat.
- Synthesis and applications. Large areas of native vegetation, where the interior remains intact, are essential to preserve species that are susceptible to edge effects and that cannot occupy other habitat types. Creating straight instead of winding edges could decrease the impact that plantations have on native forests. Furthermore, increasing the representativeness of native plant communities in exotic plantation forests would cascade up to higher consumer trophic levels, considerably increasing the conservation value of these commercial stands.
1 INTRODUCTION
Land-use conversion for human use alters the structure and complexity of landscapes (Baessler & Klotz, 2006), with the resulting habitat loss and fragmentation among the main causes of biodiversity loss and species extinction globally (Fahrig, 2003; Krauss et al., 2010). In New Zealand, conversion of native forest to agriculture, plantation forestry and urbanization has greatly changed the landscape, reducing the area of native habitat available for native species and increasing the extent of habitat edges (Brockerhoff, Jactel, Parrotta, Quine, & Sayer, 2008; Ewers et al., 2006). For instance, the Nelson and Marlborough area (South Island) was originally covered by native southern beech (Nothofagaceae) forest from the coast to the alpine tree-limit, whereas nowadays this native forest is highly fragmented, with only 26% of the native vegetation remaining in this area (Ewers et al., 2006).
Exotic plantations have been suggested to harbour a reasonably high proportion of the original native forest diversity, reducing the biodiversity loss typically associated with land-use change (Lindenmayer & Luck, 2005; Pardini et al., 2009). In New Zealand, exotic plantation forests (predominantly Pinus radiata forests) comprise 22% of the forested area (FAO, 2007), and have been identified as important substitute habitats for the preservation of specific beetle (Brockerhoff, Berndt, & Jactel, 2005; Pawson, Brockerhoff, Meenken, & Didham, 2008; Pawson, Ecroyd, Seaton, Shaw, & Brockerhoff, 2010), bird (Innes, Calder, & Williams, 1991; Kleinpaste, 1990), and bat species (Daniel, 1981). Plantation forests also harbour more of the native plant species frequently found in kanuka shrubland than many other production land uses, such as grazing pasture or arable croplands (Ecroyd & Brockerhoff, 2005). However, it is unknown whether the conservation value of exotic plantation forests (sensu Barlow et al., 2007; Farwig, Sajita, & Böhning-Gaese, 2008) applies more generally to broader groups of organisms, such as the plants, herbivores and parasitoids from native Nothofagaceae forests, as many exotic plantations in New Zealand used to be covered by this forest type. Furthermore, community composition differences between native and plantation forests could be amplified up the trophic chain, as higher trophic level species depend on the persistence of lower trophic level species (Roslin, Várkonyi, Koponen, Vikberg, & Nieminen, 2014; Steffan-Dewenter, 2003), which could be important for understanding the ecological consequences of forest conversion.
Beyond determining how similar the community composition of plantation forest is to that of native forest, it is important to understand the type of change in community composition across habitats. In particular, two processes can result in differences in community composition: losses or gains in the number of species (i.e. “nestedness” or “richness differences”) and species replacement (i.e. “species turnover”; Baselga, 2010; Baselga & Leprieur, 2015; Harrison, Ross, & Lawton, 1992). If plantations harbour a nested subset of the species that inhabit native forest, this would suggest that plantations could act as surrogates of native habitats. Conversely, if species turnover is the main community composition difference between habitat types, this would suggest that plantation forests are complementary to native forests in terms of providing habitat for species, and cannot act as substitute habitats. In that case, determining which species or sets of species with certain characteristics are more vulnerable to habitat changes will be an important focus of our conservation efforts. In this respect, a species’ degree of specialization or generalism (e.g. the number of prey species a predator can feed upon) can be a key determinant of a species capability of surviving in different habitats. For instance, specialist species have been found to be particularly sensitive to changes in habitat (Clavel, Julliard, & Devictor, 2011), because a species that depends on few resources (hosts) is more vulnerable to extinction (Aizen, Sabatino, & Tylianakis, 2012). Furthermore, increasing dominance of generalist over specialist species could increase biotic homogenization of communities (Olden, Poff, Douglas, Douglas, & Fausch, 2004) in response to land-use change.
Changes in the composition and spatial configuration of habitats in the landscape, such as the partial replacement of native habitats by commercial plantations, inherently create artificial edges (Sisk, Haddad, & Ehrlich, 1997), which are among the most pervasive outcomes of human land-use change. Edges tend to have different abiotic conditions than interiors (Ries, Fletcher, Battin, & Sisk, 2004), which can influence species survival (Ewers & Didham, 2008; Leidner, Haddad, & Lovejoy, 2010), suggesting that habitat edges can alter species diversity (Dyer & Landis, 1997) and hence the composition and dynamics of communities (Laurance, 2002). In the Nelson and Marlborough regions of New Zealand, the extensions of habitat edges formed by the juxtaposition of native and exotic plantation forests are large. Assessing the effects of edges on species composition in adjacent habitats represents an important step forward, as these effects are likely to be crucial for landscape-scale management of biodiversity in plantation-dominated landscapes.
In this study, we investigated how community composition of plant, herbivore and parasitoid species change between native Nothofagaceae forests and adjacent exotic pine plantations in New Zealand, and the extent to which plantation forests provide suitable habitats for these species. Specifically, we tested the following hypotheses: (1) community composition of different trophic levels (plants, herbivorous caterpillars, parasitoids) would differ between forest types (native Nothofagaceae forest vs. exotic pine plantations), as native forests are more structurally complex and diverse than the adjacent plantations (Stephens & Wagner, 2007). (2) Community composition of the lower trophic level would affect the composition of the trophic level immediately above, i.e. plant and herbivore composition would affect herbivore and parasitoid composition respectively, as consumers depend on their resources for survival. (3) Differences in community composition between forest types will be predominantly the result of species turnover, as the most specialized species (i.e. specialized in their feeding resources) would not be able to inhabit plantation forests. (4) Finally, because of the differences in structural complexity and diversity between native Nothofagaceae forest vs. exotic pine plantations, it could be expected that edge effects on community composition would differ among habitat types, and be stronger in native than plantation forests across all trophic levels.
2 MATERIALS AND METHODS
2.1 Study region and system
We selected eight sites in the Nelson and Marlborough region (172°47′E to 173°53′E and 41°12′S to 41°33′S) of South Island, New Zealand. Each site was formed by native Nothofagaceae forest adjacent to exotic closed-canopy P. radiata plantation forest (each forest type within a site was considered a plot), and we sampled at the edge and interior of each forest type, i.e. subplots (Figure 1). All species of the family Nothofagaceae in New Zealand were previously classified in the genus Nothofagus, but now belong to the genera Fuscospora and Lophozonia. We therefore refer to the mixed-species native forest by the collective term “Nothofagaceae forest” to reflect the dominant tree family. For more details on the study region, see Peralta, Frost, Rand, Didham, and Tylianakis (2014) and Frost, Didham, Rand, Peralta, and Tylianakis (2015).
To study changes in community composition across native Nothofagaceae forest vs. exotic pine plantations, we focused on organisms from three trophic levels: plants, herbivores (caterpillars) and parasitoids of caterpillars. In temperate forests, herbivorous caterpillars are considered to be amongst the most significant components of arthropod communities and the largest component of herbivores of canopy foliage (Schowalter, Hargrove, & Crossley, 1986). It has been suggested that community composition of these herbivore species may be more sensitive to environmental variation than changes in species richness per se (Summerville & Crist, 2003). In addition, moths and their parasitoids, have been identified as bioindicators for monitoring and assessing forest management practices (Maleque, Maeto, & Ishii, 2009).
2.2 Sampling
Sampling was conducted in a split-plot design at each of the eight sites, with the centre of the edge zone considered to be the last row of pine trees of the plantation forest (Figure 1). The pair of adjacent forest types (native Nothofagaceae forest vs. exotic pine plantation) represented the whole-plot scale in the design, and each whole-plot contained two plots (one native forest and one plantation forest) and four nested subplots: an edge in each forest type (10 m from the centre of the edge zone towards the forest interior) and an interior in each forest type (400–500 m from the centre of the edge zone and from any other edge of the forest patch), giving 32 subplots in total (four per site). The subplots were nested within plots and plots were nested within the whole-plot scale (i.e. site). At each site, we sampled all vascular plants and herbivorous caterpillars (which were reared to obtain parasitoids) along one 50 × 2 m transect (parallel to the edge) in each subplot. All the transects within a site were located within an elevational range of <100 m to avoid potential confounding effects caused by altitudinal differences within sites (De Sassi, Lewis, & Tylianakis, 2012), though elevation varied from 70 to 637 m across sites. Sampling was repeated once per month from December 2009 to February 2010, and from October 2010 to February 2011 (southern hemisphere summer).
Along each transect, we identified and beat all plants from the understorey vegetation up to a height of 2 m with a 1 m long PVC pipe to dislodge insect herbivores (Lepidoptera larvae), and counted the number of leaves beaten from each plant species. White collecting sheets (1 m²) were placed under the plants prior to beating, and dislodged caterpillars were collected from these. In addition, at 5 m intervals along each transect (i.e. at 10 sampling points) the canopy of the nearest accessible tree was sampled by using a 9 m pole pruner to cut subcanopy branches, which were then beaten over the sheets. For more details on the sampling procedure, see Peralta et al. (2014).
Herbivorous caterpillars were taken to the laboratory after field collection, where they were identified morphologically to species or morphospecies (hereafter “species”) and reared until they either developed into adults or parasitoids emerged. Herbivores were reared individually (in separate containers) under ambient conditions (16°C, relative humidity of 60% and a light rhythm of 16L:8D), and fed leaves of the plant species from which it had been collected and a supplemental general Lepidoptera growth medium (“beet army worm diet” from Bio-Serv Entomology Custom Research Diets and Environmental Enrichment Products, NJ, USA). Parasitoids (Hymenoptera and Diptera) were identified morphologically after their emergence. For male Hymenoptera parasitoids that could not be identified morphologically, we sequenced a region of the mitochondrial cytochrome C oxidase subunit I (COI) used in previous studies for parasitoid identification (Kaartinen, Stone, Hearn, Lohse, & Roslin, 2010), and related male sequences to those of female specimens that had been identified morphologically (see Peralta et al., 2014 for more details on species identification). A list of the plant, herbivore and parasitoid species involved in this study, and locations where voucher specimens have been deposited, can be found in Table S1 in Appendix S1).
To estimate the plant biomass sampled (used as a measure of the “relative abundance” of plant species along the transects), we multiplied the number of leaves from each plant species beaten on each transect by the average leaf mass per species. To calculate an average leaf mass (dry weight) per species, we weighed between 30 and 60 leaves of each plant species (depending on how variable they were in size). For 14 out of 99 species sampled, we estimated their weights based on the leaf weights of other species of similar leaf size (13 of them within the same genus), because of their scarce presence at the locations sampled. To obtain dry weight of leaves, foliage was dried in a drying oven at 60°C for 2 weeks.
2.3 Analyses
We pooled monthly sampling periods into a single dataset for each of the 32 subplots, because individual sampling dates were not independent replicates of forest types (native Nothofagaceae forest vs. exotic pine plantation) or locations (edge vs. interior), which were our main variables of interest. To determine whether plant community composition, herbivore community composition and parasitoid community composition changed between forest types and edge-interior locations (i.e. edge effects), we used split-plot permutational multivariate analyses of variance (PERMANOVA; Anderson, 2001; Anderson, Gorley, & Clarke, 2008), based on two community dissimilarity measures, as recommended by Anderson (2006). We used Jaccard and Hellinger dissimilarity measures because they differ in the emphasis they give to species composition vs. relative abundance (Jaccard incorporates only presence-absence of species, whereas Hellinger also incorporates relative abundances). The response variables (each measured with both dissimilarity measures) were therefore (1) dissimilarity in community composition of plants, (2) dissimilarity in community composition of herbivores and (3) dissimilarity in community composition of parasitoids. We fitted four-factor models, in which we entered site (random), forest type (fixed), plot (random), edge effect (edge vs. interior locations; fixed) and the interaction between forest type and edge effect as terms, with Type I (sequential) sums of squares (see Appendix S2 for more details on the PERMANOVA). We also made pair-wise comparisons between edge vs. interior locations within each forest type.
Next, we tested whether differences in herbivore and parasitoid community composition between forest types and edge-interior locations (i.e. edge effects) remained after taking into account differences in the composition of the next lowest trophic level, i.e. plants for herbivores and herbivores for parasitoids. This allowed us to determine whether there were factors, other than plant and herbivore composition, affecting the composition of the upper trophic levels (i.e. herbivores and parasitoids, respectively). We did this by including plant composition, the first axis of a principal coordinate analysis (PCO), as a covariate in the herbivore model and herbivore composition (the first axis of a PCO) as a covariate in the parasitoid model. By running each model twice, with the covariate (plant/herbivore composition) initially introduced as the last term in the model and then as the first term in the model (using Type I/sequential SS) we were able to determine how the significance of other variables in the models changed after accounting for variation explained by plant/herbivore composition. We used a Jaccard plant composition PCO axis for the herbivore PERMANOVA Jaccard models, and a Hellinger plant PCO axis for the models using Hellinger dissimilarity. Similarly, we used a Jaccard herbivore composition PCO axis for the parasitoid PERMANOVA Jaccard models, and a Hellinger herbivore PCO axis for the parasitoid models using Hellinger dissimilarity.
We tested the homogeneity of multivariate dispersions assumption of the PERMANOVA tests using a Permutational Analysis of Multivariate Dispersions (PERMDISP; Anderson, 2006) (see Appendix S2). We conducted all the permutation tests in the PRIMER v6/PERMANOVA+ environment (Anderson et al., 2008), with p values for the statistics (pseudo-F values) based on 9,999 permutations.
To determine the extent to which differences in community composition between native Nothofagaceae forests vs. exotic pine plantations resulted from nestedness or species turnover, we used a partitioning approach for decomposing Jaccard dissimilarity (Baselga, 2010). We used the “beta.multi” function of the betapart r package (Baselga & Orme, 2012) in the r environment (R Development Core Team, 2016) to estimate the proportion of Jaccard dissimilarity that is solely explained by species turnover (ßJTU) vs. the proportion of dissimilarity explained by nestedness, also known as richness differences in the losses or gains of species (ßJNE).
Finally, because specialist species tend to be more sensitive to habitat disturbances, we tested whether species that are only present in native forests tend to be more trophically specialized than those species that only inhabit, or can also inhabit, the plantation forest. To test this, we calculated the normalized degree of generalism (ND) of herbivore and parasitoid species, based on the entire dataset. Normalized degree represents the number of host species each species interact with (Dormann, 2011), normalized to account for differences in the number of partners available in the different habitats, and hence represents how generalist/specialist a species is relative to other species in the same trophic level. We used the “specieslevel” function of the bipartite r package (Dormann, Gruber, & Fründ, 2008). We then used two ANOVAs (one for herbivores and the other for parasitoids), and two Tukey tests, to determine whether the degree of generalism of herbivores and parasitoids (response variables in each of the models) differed between species that were only present in native forests, species that were also found in plantation forests, and species that were only present in plantation forests. We tested the corresponding normality and homoscedasticity assumptions and log-transformed normalized degree of both herbivores and parasitoids to meet the assumptions.
3 RESULTS
Overall, we sampled 99 different plant species (Table S1 in Appendix S1), 54% of which were found in both forest types, while 36% were present only in native forests and 10% were only present in plantation forests (see Table S2 in Appendix S1 for the shared number of species between each subplot type). From all the plants sampled, we collected and successfully reared 5,746 caterpillars from 90 species (Table S1 in Appendix S1). Of all these herbivore species, 55% were found in both forest types, 24% exclusively in native forests and 21% only in plantation forests. After herbivore rearing, 719 parasitoids from 60 species emerged (Table S1 in Appendix S1), 60% of which were found in both forest types, while 15% were only found in native forests and 25% exclusively in plantation forests. Of all the species found, only 11% of plant, 2% of herbivore and 6% of parasitoid species were found to be exotic. Most of the 11 exotic plant species were found in plantation forest interiors and in edges, on both the plantation and native forest sides. Only two herbivore species were exotic and they were only present in plantation forests, while the exotic parasitoid species were found both in plantation and native forests (Table S1 in Appendix S1).
We found significant differences in plant, herbivore and parasitoid composition among forest types (Figure 2, Figure S1 in Appendix S1), with metrics emphasizing both presence–absence alone (Jaccard) (Table S3 in Appendix S1) as well as the combined influence of differences in relative abundance and presence–absence (Hellinger) (Table 1). When plant or herbivore community composition was entered first in the herbivore or parasitoid models respectively (to determine whether observed differences in herbivore and parasitoid community composition could be accounted for solely by differences in composition of the hosts sampled), this became the only significant factor affecting community composition in terms of presence/absence (Table S4 in Appendix S1), suggesting that habitat effects on communities were largely mediated by host community changes.
| Community composition | Source | df | SS | MS | Pseudo-F | p (perm) |
|---|---|---|---|---|---|---|
| (a) Plant | Site | 7 | 4.901 | 0.700 | 1.816 | .005 |
| Forest type | 1 | 6.581 | 6.581 | 17.068 | <.001 | |
| Plot | 7 | 2.699 | 0.386 | 1.541 | .006 | |
| Edge effect | 1 | 0.521 | 0.521 | 2.086 | .035 | |
| Forest type × Edge effect | 1 | 0.590 | 0.590 | 2.358 | .026 | |
| Residuals | 14 | 3.502 | 0.250 | |||
| Total | 31 | 18.796 | ||||
| (b) Herbivore | Site | 7 | 2.991 | 0.427 | 2.017 | .002 |
| Forest type | 1 | 1.470 | 1.470 | 6.944 | <.001 | |
| Plot | 7 | 1.482 | 0.212 | 1.241 | .133 | |
| Edge effect | 1 | 0.332 | 0.332 | 1.944 | .063 | |
| Forest type × Edge effect | 1 | 0.558 | 0.558 | 3.271 | .002 | |
| Plant composition | 1 | 0.250 | 0.250 | 1.464 | .146 | |
| Residuals | 13 | 2.218 | 0.171 | |||
| Total | 31 | 9.301 | ||||
| (c) Parasitoid | Site | 7 | 5.772 | 0.825 | 1.659 | .004 |
| Forest type | 1 | 1.510 | 1.510 | 3.040 | .001 | |
| Plot | 7 | 3.478 | 0.497 | 1.115 | .223 | |
| Edge effect | 1 | 0.791 | 0.791 | 1.776 | .044 | |
| Forest type × Edge effect | 1 | 1.230 | 1.230 | 2.760 | .001 | |
| Herbivore composition | 1 | 1.074 | 1.074 | 2.411 | .006 | |
| Residuals | 13 | 5.792 | 0.445 | |||
| Total | 31 | 19.647 |
Differences in community composition between forest types were mostly due to species turnover (ßJTU), with over 50% of the Jaccard dissimilarity explained by this component across the three trophic levels (Figure 3). Furthermore, both herbivore and parasitoid species that were only present in native forests and only present in plantation forests tend to have a lower degree of generalism, i.e. were more specialized, than those species that were found in both forest types (F = 33.330, p < .001 and F = 35.573, p < .001 for herbivores and parasitoids respectively; Figure 4).
When assessing the differences in community composition between edge vs. interior habitats, we found that edges can have different effects on community composition of different trophic levels in different forest types (significant interaction effect between forest type and location, i.e. edge vs. interior). For instance, as predicted, in native Nothofagaceae forests plant community composition differed significantly between edge and interior locations (Table S5 in Appendix S1), whereas no differences were detected among locations in exotic pine plantation (Table S5 in Appendix S1). Conversely, herbivore composition differed significantly between edge and interior locations in both forest types (Table S5 in Appendix S1), while parasitoid composition only differed among locations in plantation forests when considering differences in the abundance of species (Hellinger dissimilarity; Table S5 in Appendix S1). This indicates that the parasitoid edge/interior habitat differences in plantation forests were driven more by changes in the relative abundances of parasitoid species, rather than their presence or absence. Significant differences in the relative abundances of herbivore species (using Hellinger distances) at edges vs. interiors persisted after controlling for plant composition (Table 2), suggesting some direct effect of interior vs. edge locations on their relative abundances.
| Community composition | Source | df | SS | MS | Pseudo-F | p (perm) |
|---|---|---|---|---|---|---|
| (a) Herbivore | Plant composition | 1 | 1.941 | 1.941 | 9.511 | <.001 |
| Site | 7 | 2.857 | 0.408 | 2.022 | .001 | |
| Forest type | 1 | 0.267 | 0.267 | 1.488 | .148 | |
| Plot | 7 | 1.396 | 0.199 | 1.169 | .202 | |
| Edge effect | 1 | 0.358 | 0.358 | 2.098 | .044 | |
| Forest type × Edge effect | 1 | 0.263 | 0.263 | 1.542 | .128 | |
| Residuals | 13 | 2.218 | 0.171 | |||
| Total | 31 | 9.301 | ||||
| (b) Parasitoid | Herbivore composition | 1 | 2.437 | 2.437 | 4.387 | <.001 |
| Site | 7 | 5.550 | 0.793 | 1.552 | .012 | |
| Forest type | 1 | 0.746 | 0.746 | 1.570 | .087 | |
| Plot | 7 | 3.557 | 0.508 | 1.141 | .181 | |
| Edge effect | 1 | 0.720 | 0.720 | 1.616 | .075 | |
| Forest type × Edge effect | 1 | 0.846 | 0.846 | 1.898 | .027 | |
| Residuals | 13 | 5.792 | 0.445 | |||
| Total | 31 | 19.647 |
4 DISCUSSION
Plantation forests have been identified as production areas with potential for the preservation of native species (Brockerhoff, Jactel, et al., 2008; Brockerhoff, Shaw, et al., 2008; Daniel, 1981; Kleinpaste, 1990; Pawson et al., 2008), particularly compared to other modified land-use types (De La Luz Avendaño-Yáñez, Sánchez-Velásquez, Meave, & Pineda-López, 2016; Ecroyd & Brockerhoff, 2005). In particular, other studies have found that exotic Pinus plantations can maintain large numbers of native forest species (Bonham, Mesibov, & Bashford, 2002; Pawson et al., 2010), but usually not all species found in nearby native forest (Neumann, 1979; Pharo, Lindenmayer, & Taws, 2004; Robson, Baker, & Murray, 2009). However, these studies have not considered community composition of adjacent trophic levels, nor community composition changes between plantation and natural forest interiors and edges. Given this, we examined how well exotic pine plantations maintain entire native Nothofagaceae forest communities across three interacting trophic levels. We found that despite the high proportion of species that inhabit both native and plantation forests (54%, 55% and 60% for plants, herbivores, and parasitoids respectively), the community composition of plants, herbivorous caterpillars and their parasitoids in fact differed significantly between native and commercial stands. Furthermore, we found that these changes are mainly due to species turnover across forest types, that upper trophic levels (herbivores and parasitoids) strongly depended on changes in the composition of host species trophic levels (plants and herbivores respectively), and that the most generalist species are the ones that can adapt to plantation forests. Although we focus on a specific case-study comparison of two forest types, and cannot directly attribute mechanistic causality to “naturalness” (native vs. exotic), “management” (unmanaged vs. managed forests) or “tree identity effects” (broadleaf vs. coniferous), we believe our findings will be broadly generalizable to many similar landscape contexts in which managed exotic plantations (of a variety of species) are grown next to semi-natural forest habitats (e.g. Bonham et al., 2002; Robson et al., 2009).
Across the three trophic levels, differences in community composition between native Nothofagaceae forests vs. exotic pine plantations were mostly a result of species turnover, which suggests that plantation forests are complementary habitats for some species, but cannot be a substitute habitat for all native forest species. Previous studies have shown that marked turnover of butterfly and termite species (Basset et al., 2017), as well as notable changes in ant species composition (Maeto & Sato, 2004), between native vs. plantation forests are not necessarily reflected in differences in species richness. This highlights the importance of incorporating community composition measures when comparing communities from different habitats, as it allows the assessment of the community as a whole and also seems to be more sensitive to environmental variation than more commonly used diversity metrics, such as species richness (Summerville & Crist, 2003).
Native herbivore and parasitoid species that inhabited both forest types were found to be the most generalist feeders. Their broader capacity to exploit different hosts appears to confer greater resilience to habitat changes (Gámez-Virués et al., 2015). Beyond dietary requirements, species can be specialized to different habitat types (or microhabitat characteristics), which can also influence their presence and relative abundance in plantation forests. For instance, it has been observed that habitat generalists can inhabit both old-growth forest and plantations, while woodland habitat specialists are only found in old-growth forests, and open-habitat specialists dominate in conifer plantations (Maeto & Sato, 2004). Despite the benefit that plantations provide for generalist species, our results show that plantations are not a substitute for native forests, and in order to conserve native forest specialists it is essential to preserve native forests. It has also been suggested that to increase the conservation of specialist species in plantation forests, a variety of management practices could be used to promote microhabitat structure that is beneficial to these species (Deal, Hennon, O'Hanlon, & D'Amore, 2014; Maleque et al., 2009). Besides, reducing biotic homogenization can increase the capacity of landscapes to provide multiple functions and increase their resilience to environmental disturbances (Olden et al., 2004).
Despite the common ability of herbivorous caterpillars to feed on different plant species, almost all the compositional variation in herbivore communities across forest types, and between edge and interior habitats, could be explained by variation in plant community composition, with only a small residual signal of other biotic and abiotic habitat-dependent factors. Similarly, an understanding of variation in herbivore community composition was sufficient to entirely explain variation in parasitoid community composition. This suggests that parasitoid survival might depend most sensitively on the presence of host species and other feeding resources, despite previous suggestions that higher trophic levels may be more susceptible to habitat fragmentation independent of its effects on lower trophic levels (Didham, Lawton, Hammond, & Eggleton, 1998; Komonen, Penttilä, Lindgren, & Hanski, 2000). Higher overlap of herbivore and parasitoid communities between forest types, as compared with the overlap observed for plant communities, could be driven by their high dispersal abilities across these habitats (Frost et al., 2015).
The conversion of native habitats to forestry and agricultural land not only causes habitat loss, but also results in fragmentation, increasing the extent of artificial habitat edges in the landscape. Therefore, when assessing the conservation value of production land, land managers not only need to account for the species that this land can host, but also the effects that habitat conversion has on the adjacent native remnant, such as edge effects (Ewers & Didham, 2008; Murcia, 1995). When assessing the effects of adjacent plantation land use on native forests, we found differences in plant and herbivore community composition between edge and interior habitats. These differences in plant community composition could be driven by sensitivities of more specialized species to differences in abiotic conditions (Murcia, 1995), as well as differences in seed dispersal (Fagan, Cantrell, & Cosner, 1999).
Unlike the strong edge responses observed in native forests, we did not find a comparable community-level response to edge effects in plantation forests. There was no significant difference in plant community composition between edge and interior habitats in plantation forests, potentially because the high density and canopy cover of pine trees are uniformly consistent across both edges and interiors of plantation forests. Moreover, pine trees create an above-ground layer of pine needles which acidifies the soil (Brand, Kehoe, & Connors, 1986) and may create suboptimal conditions for many native plant species.
In a conservation management sense, it is clear that plantation forests are not typically as diverse as native forests (Volpato, Prado, & dos Anjos, 2010), but they can nevertheless harbour substantial populations of species across many taxa (O'Hanlon & Harrington, 2011; Pawson et al., 2008, 2010; Quine & Humphrey, 2010), and hence their role in maintaining biodiversity should not be ignored (Lindenmayer & Hobbs, 2004). Overall, our findings suggest that land-use change and edge effects alter the composition of communities at different trophic levels. Furthermore, the effects of habitat edges on community composition differed depending on the forest type considered, which leads to different but not necessarily conflicting management actions. From a conservation point of view, it is important to consider that more than 43% of the herbivore and parasitoid species that were only present in native forests were solely found in native interiors, suggesting that preserving large areas of native forest, where the interior remains intact, will be important if we are to preserve these specialized species. Conversely, edge effects do not seem to affect the community composition of plantation forests (except for parasitoids). Nevertheless, reducing the extent of edges between pine plantations vs. native forests could decrease the impact that plantations have on native forests, and hence indirectly promote biodiversity conservation in the native forest. In many cases, this might be easily achievable by creating straight instead of winding edges.
Another plantation forest management strategy to broaden the suitability of plantations as habitat for a broader range of species could be to increase the representativeness of the native understorey plant community. This could be achieved by retaining native plant species during harvest and thinning operations, avoiding complete clear up of stands during harvesting and pre-planting (e.g. by leaving some vegetation, and/or not applying herbicides over the entire planting area; Hartley, 2002), and even considering the use of alternative planting systems, such as mixed forests (Hua et al., 2016) or agroforestry (Stamps & Linit, 1998). None of these practices are currently included in New Zealand legislation. Enriching the understorey plant community would favour upper consumer trophic levels as well, because bottom-up effects, such as the availability of host species across habitats, can have strong impacts on upper trophic levels. This could promote natural enemies of herbivore pests (Jactel, Brockerhoff, & Duelli, 2005; Jactel, Goulard, Menassieu, & Goujon, 2002; Quayle, Regniere, Cappuccino, & Dupont, 2003), which is one of the potential reasons why there tend to be lower population densities of damaging insects in mixed plantations than in monocultures (Guyot, Castagneyrol, Vialatte, Deconchat, & Jactel, 2016; Jactel & Brockerhoff, 2007). Besides these factors, the public largely prefers diverse stands over monocultures (Hartley, 2002).
In short, while efficiently producing fibre (Hartley, 2002), through the simple application of more suitable stand management practices (Jofré, Warn, & Reading, 2016; Lindenmayer, Franklin, & Fischer, 2006; Maleque et al., 2009; Norton, 1998), plantations have great potential to conserve biodiversity.
ACKNOWLEDGEMENTS
We thank The Department of Conservation, Nelson Forests Ltd, Hancock Timber Resource Group, Merrill and Ring, and D. Bryant for forest access. J. Dugdale, J. Berry, and R. Schnitzler provided taxonomic assistance. J., B., D., D. and S. Ladley, D. Conder, N. Etheridge, and D. Payton assisted with field and laboratory logistics. M. Barlet, Y. Brindle, D. Davies, C. Hohe, S. Hunt, A. Knight, T. Lambert, S. Litchwark, A. McLeod, V. Nguyen,, L. O'Brien, K. Trotter, T. Watson, L. Williamson and A. Young assisted with caterpillar collection and rearing. A. Varsani helped with molecular identifications. J. Tylianakis, T. Rand and three anonymous reviewers provided helpful comments on the manuscript. This research and G.P. were supported by the Marsden Fund (UOC-0802). C.M.F. was supported by the Natural Sciences and Engineering Research Council of Canada, Education New Zealand, and the University of Canterbury.
AUTHORS’ CONTRIBUTIONS
G.P., C.M.F., R.K.D. designed the study; G.P., C.M.F. collected data; G.P. and R.K.D. conducted the analyses. G.P. wrote the manuscript and all authors contributed substantially to revisions. All authors gave final approval for publication.
DATA ACCESSIBILITY
Data are available in the Dryad Digital Repository https://doi.org/10.5061/dryad.92n4m (Peralta, Frost, & Didham, 2017).
