Insect Conservation and Diversity (2016) 9, 456–469 doi: 10.1111/icad.12183
Complementary distribution patterns of arthropod detritivores (woodlice and millipedes) along forest edge-to-interior gradients
PALLIETER DE SMEDT,1 KAREN WUYTS,2 LANDER BAETEN,1 AN DE SCHRIJVER,1 WILLEM PROESMANS,1 PIETER DE FRENNE,1 EVY AMPOORTER,1 ELYN REMY,1 MERLIJN GIJBELS,1 MARTIN HERMY,3 4 1 DRIES BONTE andKRISVERHEYEN 1Forest & Nature Lab, Department of Forest and Water management, Ghent University, Melle (Gontrode), Belgium, 2ENdEMIC Research Group, Department of Bioscience Engineering, University of Antwerp, Antwerp, Belgium, 3Division Forest, Nature and Landscape, Department of Earth and Environmental Sciences, University of Leuven, Leuven, Belgium and 4Terrestrial Ecology Unit (TEREC), Department of Biology, Ghent University, Ghent, Belgium
Abstract. 1. Worldwide, forest fragmentation induces edge effects, thereby strongly altering the forest microclimate and abiotic characteristics in the forest edge compared to the forest interior. The impact of edge-to-interior gradients on abiotic parameters has been extensively studied, but we lack insights on how biodiversity, and soil communities in particular, are structured along these gra- dients. 2. Woodlice (Isopoda) and millipedes (Diplopoda) are dominant macro-detriti- vores in temperate forests with acidic sandy soils. 3. We investigated the distribution of these macro-detritivores along forest edge-to-interior gradients in six different forest stands with sandy soils in north- ern Belgium. 4. Woodlouse abundance decreased exponentially with distance from the for- est edge, whereas millipede abundance did not begin to decrease until 7 m inside the forest stands. Overall, these patterns were highly species specific and could be linked to the species’ desiccation tolerance. Whereas the observed abundance patterns were independent from forest stand and dominant tree species, tree species had a large effect on community structure. 5. Edge gradients in macro-detritivores may consequently have implications for nutrient cycling, especially in smaller forest fragments with a large edge-to- interior ratio. Key words. Detritivores, diplopoda, edge effects, gradients, isopoda, soil fauna.
Introduction (Janzen, 1986; Reed et al., 1996; Gascon et al., 2000; Harper et al., 2005; Fletcher et al., 2007; Echeverria et al., Forest habitats are profoundly fragmented around the 2008). Forest edges are characterised by enhanced light world (Wade et al., 2003). Such fragmentation induces a availability (Delgado et al., 2007), higher wind speeds reduction in forest patch sizes and strengthens edge effects (Wuyts et al., 2008a), higher air and soil temperatures (Delgado et al., 2007; Heithecker & Halpern, 2007), and lower relative humidity and soil moisture (Chen et al., Correspondence: Pallieter De Smedt, Forest & Nature Lab, 1995; Gehlhausen et al., 2000). They are also relative ‘hot- Department of Forest and Water management, Ghent University, spots’ for the deposition of eutrophying and acidifying Geraardsbergsesteenweg 267, 9090 Melle (Gontrode), Belgium. atmospheric pollutants (Weathers et al., 2001; Wuyts E-mail: [email protected] et al., 2008b,c) compared to the forest interior. The
456 Ó 2016 The Royal Entomological Society Arthropod detritivores in forest edges 457 magnitude and depth of influence of these edge effects are edge effects and macro-detritivore communities, we put strongly affected by the structure and composition of the forward the following hypotheses about the influence of edge itself (Weathers et al., 2001; Wuyts et al., 2008b,c). edge effects on woodlouse and millipede distribution: These abiotic edge gradients then give rise to secondary 1 More favourable environmental conditions (higher effects on biotic effects at the edge of forest ecosystems temperatures and higher litter quality) at the forest (Murcia, 1995; Harper et al., 2005). edge will result in a higher abundance of macro-detri- Soil macro-invertebrates are dominant detritivores in tivores at this boundary, with abundance steadily temperate forests, which breakdown dead organic material declining towards interiors. If true, we expect the mass (Lavelle, 1997), thus affecting the physico-chemical char- of the ectorganic horizon to show the inverse trend acteristics of soil (Snyder & Hendrix, 2008). By reducing due to increasing rates of decomposition from the inte- the size of dead organic material on the forest floor rior to the forest edge. (Anderson, 1988; Grelle et al., 2000), they increase the 2 The response of macro-detritivores to edge proximity accessible surface area for further decomposition by is species specific, as each species exhibits different microbes (Harper et al., 2005). This results in a more temperature and humidity preferences. stable soil organic matter layer (Wolters, 2000). The trans- 3 Responses of detritivores to forest edge proximity can formation of fallen leaves into macro-detritivore faeces be related to changes in relevant abiotic parameters also has strong effects on the microbial response and con- (i.e. food quality, cation content of the soil, etc.). sequently on the breakdown of the leaf material (Joly et al., 2015). Exclusion of these soil macro-arthropods slows down decomposition rates (Riutta et al., 2012; Slade & Riutta, 2012), and their presence is therefore of vital Material and methods importance for nutrient cycling in forest ecosystems. The distribution of macro-arthropods, such as woodlice and Site description millipedes, within forests is highly scattered (Hornung, 2011) and aggregated towards forest edges (Riutta et al., We selected six forest stands in the northern part of 2012). The occurrence of abiotic edge effects raises the Belgium, located on poor, acidic, well-drained sandy soils question of whether the response from macro-arthropods (Haplic podzols) of the Campine and Sandy region. All also varies gradually as the forest edge is approached, and were recently created forests, formerly managed as heath- whether this is reflected in the accumulation of forest floor lands until 80–90 years ago. This heathland management material. Yet our understanding of how biotic factors, practice resulted in a significant depletion of soil nutrients such as different taxonomic groups of the soil fauna com- through sheep grazing and turf cutting on these already munity, change along edge-to-interior gradients and how naturally nutrient-poor soils. After some years of aban- this could affect the litter decomposition process is extre- donment, the sites were afforested with monocultures of mely limited (Hattenschwiler€ et al., 2005). Temperature pedunculate oak (Quercus robur L.; stands Q1 and Q2), and humidity are important, and highly species specific, silver birch (Betula pendula Roth.; stands B1 and B2), environmental triggers for survival and distribution of Corsican pine (Pinus nigra ssp. laricio Maire; stand P1), macro-detritivores (Warburg, 1964; Haacker, 1968; Meyer or Austrian pine (P. nigra ssp. nigra Arnold; stand P2) & Eisenbeis, 1985; Dias et al., 2013). On the other hand, (Table 1). The stands all had an abrupt forest edge, i.e. the spatial distribution of detritivores is also strongly they lacked a gradual transition with the adjacent open influenced by soil acidity and exchangeable base cations land, and were always oriented towards the southwest, (Kime, 1992; Van Straalen & Verhoef, 1997), as well as perpendicular to the prevailing wind direction. All were food quality (C/N ratio) (Hassall et al., 2002; David & located in the periphery of forest complexes, within a Handa, 2010; Gerlach et al., 2014). These environmental fragmented landscape dominated by agriculture and inten- parameters vary strongly along edge-to-interior gradients, sive livestock breeding. Forest edges were bordered by and we would therefore expect species distribution pat- grass pasture, extensively managed meadow (B1, Q1, and terns to be highly influenced by distance from the forest P2), or by arable land (B2, Q2, and P1). A road with edge. Detailed empirical data that could be used to inves- roadside verges (~20 m in total) was present between the tigate these patterns is, however, to our knowledge very grassland or arable land and the forest edge of stands Q1 scarce. and Q2. The stands bordering arable land could have In northern Belgium, where forests are strongly frag- been exposed to drift of lime or fertiliser from agricultural mented, we investigated the distribution patterns of woo- applications. Under-storey vegetation was absent in all dlice and millipedes in transects that stretched from the stands except for (i) brambles (Rubus fruticosus agg.), forest edge towards the forest interior, for several differ- which occurred in the first 20 m from the edge in Q1 and ent forest stands (De Schrijver et al., 2007). On these further than 50 m from the edge in P1, and (ii) creeping acidic sandy forest soils, earthworms are very scarce or soft grass (Holcus mollis L.), which was present in the first even absent (Muys & Lust, 1992), and therefore woodlice 10 m from the edge in both P1 and P2. No significant and millipedes are the major macro-detritivore groups quantity of coarse woody debris was present in the stands. (Jeffery et al., 2010). Based on our knowledge of abiotic The edge patterns of soil nitrogen leaching, soil
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 458 Pallieter De Smedt et al.
Table 1. Overview of the characteristics of the six investigated forest stands.
Stand Location Dominant tree species Other species* Stand age (y) Soil pH(KCl)
Q1 51°24044″N Quercus robur L. Alnus glutinosa L. 74 2.88 05°02045″E Sorbus aucuparia L. Q2 50°52008″N Quercus robur L. Alnus glutinosa L. 96 3.35 03°27059″E Prunus serotina Erhr. Sorbus aucuparia L B1 51°23030″N Betula pendula Roth Pinus nigra ssp. laricio Maire 35 3.40 05°02031″E Larix spec. Mill. B2 51°090220’N Betula pendula Roth Quercus robur L. 36-46 2.93 03°040480’E Sorbus aucuparia L. P1 51°26037″N Pinus nigra ssp. nigra Arnold Sorbus aucuparia L. 49 2.89 05°05014″E P2 51°08026″N Pinus nigra ssp. laricio Maire Betula pendula Roth 71 2.91 03°06036″E Quercus robur L. Sorbus aucuparia L.
*subdominant, in the shrub layer. acidification, litter and soil chemistry, stand structure, and collect any remaining organisms. Once complete, collected the atmospheric deposition of nitrogen, acidifying ions, arthropods were again stored in 70% ethanol, and the dry and base cations in the studied forests were previously mass of the ectorganic horizon subsample was determined described by Wuyts et al. (2008b,c, 2011, 2013). after drying as described previously. Afterwards, the abundance data of the Berlese-Tullgren subsamples were converted into sample level by multiplying arthropod Experimental set-up and sampling abundance data with the total dry mass of the sample, divided by the dry mass of the subsample. The total abun- In each of the forest stands, a transect was laid out dance of woodlice and millipedes (numbers per square perpendicular to the forest edge and heading towards the metre) was determined as the sum of both the visual forest interior. Along each transect, samples were taken at counts and the converted data from the Berlese-Tullgren 0, 2, 4, 8, 16, 32, 64, and 128 m from the forest edge. No subsamples. The collected woodlice and millipedes were samples could be taken at 128 m in the birch stands due identified to the species level according to Berg and Wijn- to their smaller size. According to Murcia (1995) and De hoven (1997) and Andersson et al. (2005), respectively. Schrijver et al. (2007), edge effects are negligible at dis- We are aware that summer sampling is biased towards tances greater than 50 m from the forest edge, so we larger species as many small species retreat deeper into assumed distances of 64 m and 128 m to be representative the soil during this period (Gregory, 2009). of the forest interior. At every distance, three samples of The fermentation and humus layer of the forest floor, the ectorganic layer (including the litter, fermentation, as well as the upper 5 cm of the mineral soil, were sam- and humus layer) were taken, with a spacing of about pled for chemical analyses in previous studies. This sam- 5 m from each other. Samples were collected in plastic pling campaign was carried out for all stands and at all bags. We used a wooden frame (25 9 25 cm) to cut out sampling distances, and the pH(KCl) of the upper mineral the ectorganic layer. All forests were sampled once soil is available at all of these locations (Wuyts et al., between 12 July 2011 and 12 August 2011, before the 2008b, 2013). In addition, the following data, described peak litter fall period. Whereas July 2011 was rather dry by Wuyts et al. (2011, 2013), are available for all stands and cold, weather conditions in August 2011 were normal except B1: (i) potassium (K), calcium (Ca), magnesium in the context of the previous 10 years. (Mg), carbon (C), and nitrogen (N) concentration in the Dry mass of the ectorganic horizon was determined ectorganic horizon; (ii) exchangeable amounts of K, Ca, after removing the arthropods (see further) and drying the Mg, and aluminium (Al) (subsequent to a BaCl2 extrac- samples at 70°C for 2 days. No specific sampling was per- tion); and (iii) C and N concentration in the upper min- formed on decaying wood due to the lack of coarse eral soil (0–5 cm depth). For stand B1, only data on the woody debris. Fine woody debris was included in the C and N concentration of the forest floor were available ectorganic samples. (Wuyts et al., 2008b). The samples were visually inspected for macro-arthro- Late summer leaf area index (LAI) was measured at all pods in the laboratory within 1 day after collection, and sampling distances in previous studies (Wuyts et al., the found organisms were removed and stored in 70% 2008b,c). LAI is used as a measure of light in the forest ethanol. Immediately following the visual inspection, an and is a crucial factor for woodlouse and millipede beha- ectorganic horizon subsample was taken ( 500 cm³) and viour, as they are mostly night active and negative photo- transferred into a Berlese-Tullgren funnel for 7 days to tactic (they move away from light) (Sutton, 1972; Hopkin
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 Arthropod detritivores in forest edges 459
& Read, 1992). The LAI in stand B1 was determined species-specific abundance patterns differ between stands using the LAI-2000 Plant Canopy Analyzer in August with different dominant tree species, individual macro- 2004. For all other stands, digital hemispherical pho- detritivore species did not generally occur in sufficient tographs, taken in August–September 2006 and processed numbers in each of the stands to make this possible. The with Gap Light Analyzer 2.0, were used (www.caryinsti species-specific graphs for woodlice were ordered accord- tute.org). Although LAI measurements and chemical anal- ing to the desiccation resistance of the species, based on yses were performed 7 and 5 years prior to the arthropod the experiments by Dias et al. (2013). For millipedes, collection, no major disturbances in the canopy (i.e. forest these data have not yet been published, but the ordering management, storm damage, insect damage, etc.) was done after personal communication with M. P. Berg, occurred, and therefore we considered the values as repre- who is measuring desiccation tolerance on millipedes sentative of the current situation in the studied forests. using the same method as Dias et al. (2013). To test the effect of distance from the forest edge, tree species, and forest stand on the community structure of Data analyses woodlice, millipedes, and the two groups combined, we used the Bray-Curtis distance measure to calculate com- Abundance patterns along forest edge-to-interior gradi- positional dissimilarities between sampling locations with ents were not expected to show linear relationships, and different distances from the forest edge, between different were therefore modelled with generalised additive models dominant tree species, and between forest stands. This (GAM), using the gam-function in the mgcv-package in R metric quantifies compositional variation driven by both 3.2.2 (Wood, 2006; R Core Team, 2015). Abundance compositional differences (i.e. different species are pre- values used in the models were calculated for each stand sent) as well as differences in relative abundances (Ander- as the sum of the three samples taken at each sampling son et al., 2006). A log(distance + 1), tree species, and distance. By comparing three models of increasing stand effect were fitted using a permutational multivariate complexity, we sequentially tested how the macro-detri- analysis of variance with 999 permutations (PERMANOVA; vore abundances (dependent variable) were affected by function adonis in the vegan-package) (Oksanen et al., taxonomic group, distance from the forest edge, and taxo- 2015). We ran all models with only one explanatory vari- nomic group-specific distance effects (predictor variables). able at a time to have a proper comparison between the First, a model with taxonomic group as the predictor different variables. As PERMANOVA confounds location variable tested differences in mean abundance between (compositional dissimilarities between groups) and disper- woodlice and millipedes. Second, log(distance + 1) was sion effects (compositional dissimilarities within groups) added as a predictor variable to test how abundance (Anderson, 2001; Warton et al., 2012), however, we tested changes from forest edge to interior. Because abundance separately for multivariate homogeneity of dispersions does not change in a linear way along the edge-to-interior between groups using the function betadisper (vegan-pack- gradient, we modelled it using a smooth function of the age; centroid analysis). This is a multivariate analogue of distance predictor variable. Smooth functions are useful Levene’s test for homogeneity of variances (Anderson for this application because they can be used to model a et al., 2006). If multivariate dispersion is significant, PER- wide range of trends. Finally, we allowed the distance MANOVA results must be handled with care as we cannot effects to vary per taxonomic group (woodlice and milli- distinguish between dissimilarities between and within pedes); that is, each taxonomic group was allowed to have groups. a different distribution pattern from edge to interior. The Lastly a principal component analysis was performed models were compared using an analysis of deviance table using the vegan-package (Oksanen et al., 2015), with to successively test the significance of each of the predic- woodlouse and millipede species abundances as response tor variables. An F-test statistic was used to test the variables and LAI, mineral topsoil, and ectorganic hori- change in deviance across the three models. Tree species zon variables, and mass of the ectorganic horizon as envi- or stand may also have an influence on woodlouse and ronmental variables. These environmental factors were millipede abundance patterns from edge to interior. This retained after stepwise exclusion of factors with variance was tested for each taxonomic group separately, by com- inflation factors higher than 10. After a factor was paring (via change in deviance) models that allowed the excluded, the analysis was re-run. Subsequently, these distance effect to vary between the six stands or between four variables and their relation to distance from the for- the three tree species (oak, birch, pine; two stands each). est edge were analysed according to the same statistical We used the same methods to analyse differences in spe- methods used for the arthropod analyses. cies richness of woodlice and millipedes as well as mass of the ectorganic horizon. Species-specific distance effects were tested for species with more than 50 individuals Results in the data set. Again, we used GAM models with log(distance + 1) as the main effect to test how species’ We found six woodlouse species belonging to five different abundances change from forest edge to interior. families, and ten millipede species belonging to six differ- Although it would be interesting to further test whether ent families (Table 2). The dominant woodlouse species
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 460 Pallieter De Smedt et al.
Table 2. Species list, with their family, abbreviation (Abb.), and relative occurrence within their taxonomic group. Nomenclature follows Berg et al. (2008).
Species Family Abb. Rel. occurrence (%)
Woodlice Porcellio scaber Latreille, 1804 Porcellionidae PORCSCAB 56.7 Philoscia muscorum (Scopoli, 1763) Philosciidae PHILMUSC 22.1 Oniscus asellus Linnaeus, 1758 Oniscidae ONISASEL 8.2 Haplophthalmus danicus Budde-Lund, 1880 Trichoniscidae HAPLDANI 6.9 Trichoniscus pusillus s.l. Trichoniscidae TRICPUSI 6.0 Ligidium hypnorum (Cuvier, 1792) Ligiidae LIGIHYPN 0.1 Millipedes Proteroiulus fuscus (Am Stein 1857) Blaniulidae PROTFUSC 32.7 Cylindroiulus punctatus (Leach 1815) Julidae CYLIPUNC 16.9 Glomeris marginata (Villers 1789) Glomeridae GLOMMARG 12.0 Polydesmus denticulatus C.L. Koch 1847 Polydesmidae POLYDENT 11.1 Craspedosoma rawlinsii Leach 1814 Craspedosomatidae CRASRAWL 4.0 Melogona spec. Chordeumatidae MELOSPEC 2.0 Julus scandinavius Latzel 1884 Julidae JULUSCAN 1.9 Ommatoiulus sabulosus (Linnaeus 1758) Julidae OMMASABU 0.5 Chordeuma sylvestre C.L. Koch 1847 Chordeumatidae CHORSYLV 0.3 Leptoiulus belgicus (Latzel 1884) Julidae LEPTBELG 0.1 Millipede spec. juvenile / MILLIJUV 18.5
were Porcellio scaber (56.7% of individuals) and Philoscia the forest edge. We therefore focus further analysis on muscorum (22.1%). The dominant millipede species was abundance patterns. Proteroiulus fuscus (32.7%). Abundance patterns of both woodlice and millipedes varied considerably between stands (woodlice: F = 42.40, P < 0.001; millipedes: F = 5.717, P < 0.01). In the case of Woodlouse and millipede abundance patterns and trends in woodlice, they varied in a consistent way between stands mass of the ectorganic horizon dominated by different tree species (F = 37.45, P < 0.001), but millipede abundance patterns were not consistent The overall abundance of woodlice did not differ from between the different tree species (F = 3.245, P = 0.072) that of millipedes (model with vs. without effect of taxo- (Fig. 2b). In fact, only three of six stands showed patterns nomic group: t = 1.598, P = 0.111). The effect of distance resembling the overall trend for millipedes (Fig. 1a). It was significant (model with vs. without distance effect; should be noted that the fitted lines (smoothers) were only F = 20.634, P < 0.001), however, and was also signifi- significant for oak stands in the millipede data (P < 0.01), cantly different for the two taxonomic groups (with or however. For woodlice, all smoothers were significant without different distance effect per taxonomic group; (P < 0.001) (Fig. 2a). F = 10.58, P < 0.01). Indeed, the total abundance of The mass of the ectorganic horizon was significantly woodlice steeply decreased from the forest edge towards influenced by distance from the forest edge (F = 6.333, the forest interior, whereas the abundance of millipedes P < 0.05) and tree species (F = 8.555, P < 0.001) (Fig. 3), did not begin to decrease until about 7 m inside the whereas no difference between stands (F = 1.111, forest stands (Fig. 1). The abundance of woodlice P > 0.05) was found. For all stands, we found an increase (326 ind. m 2) and millipedes (36 ind. m 2) differed in ectorganic horizon mass going from the forest edge significantly at the forest edge (Intercept, t = 26.67, towards the forest interior (Fig. 3). P < 0.001), with higher abundances of woodlice compared to millipedes (Fig. 1a). The same patterns were found for species richness of woodlice and millipedes; that is, aver- Species-specific patterns age species richness was the same (Intercept, t = 1.656, P = 0.099), but species richness of woodlice (1.96 0.22) Species-specific distance-to-edge models were fitted for was higher at the forest edge compared to millipedes a total of five woodlouse species and five millipede species (1.14 0.22) (Intercept, t = 9.901, P < 0.001). Distance (Fig. 4). The smoothers in this sequence reveal an interest- to the forest edge had a significant effect on species rich- ing pattern for woodlice (Fig. 4a). Species such as ness (F = 16.492, P < 0.001) and was also different for Haplophthalmus danicus, or even Trichoniscus pusillus and the two taxonomic groups (F = 5.5612, P = 0.019). As P. muscorum, which have low desiccation tolerances, are can be seen in Fig. 1, species richness (Fig. 1b) and abun- actually more abundant a certain distance away from the dance (Fig. 1a) show similar patterns with distance from forest edge than they are at the edge itself, whereas species
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 Arthropod detritivores in forest edges 461
(a) the variation in community structure explained by stand alone (Table 3). For both taxa combined (woodlice + millipedes), the variance explained by stand was 22.5%. Tree species also explained a large percentage of the variation in community structure of woodlice (11.9%), millipedes (11.2%), and both groups combined (9.5%). Distance only explained a minor part of community varia- tion, with 3.16% for woodlice and 3.64% for both groups (abundance)
e combined. For millipedes, distance did not contribute significantly to the observed variation in community struc- Log ture. The multivariate dispersion of some of the commu- Woodlice nity data, however, was significantly heterogeneous in relation to distance (woodlice + millipedes), tree species Millipedes (millipedes and woodlice + millipedes), and stand (woo- dlice, millipedes, and woodlice + millipedes) (Table 3).
Loge (distance) Hence, variation in community structure across stands results from both effective changes in community compo- sition and dispersion (difference in average distance to (b) group centre between tested groups). In the principal component analysis (Fig. 5), variables with variance inflation factors higher than 10 were excluded. The results showed that species of woodlice and millipedes have higher abundances at higher pH values, higher Mg content, and lower C/N ratio of the forest floor litter (Fig. 5). They are less correlated with LAI, except for the woodlouse P. scaber, whose abundances
Species richness are not highly correlated with any other species or envi- Woodlice ronmental driver but are closely linked to LAI. In the lower part of the graph, species with low desiccation resis- Millipedes tance, such as the woodlice H. danicus and T. pusillus and the millipede Craspedosoma rawlinsii, cluster together, but also the more drought-tolerant G. marginata is found in Loge (distance) this corner. Differences among tree species are mostly shown along the second axis, from higher to lower: Pine, Fig. 1. Modelled abundance (ln-transformed) (a) and number of Oak, and Birch. This axis is also almost perfectly followed species (b) of woodlice and millipedes versus distance from the forest edge (ln-transformed) across all stands. Dotted lines denote by the very common woodlouse P. scaber, which was the 1 standard error. only abundant woodlouse in the pine stands, although the millipede P. fuscus was also very abundant there. Oak with high desiccation tolerance decrease exponentially stands cover a wide range (Fig. 5) as all species (except with distance from the forest edge. The species with the the millipede Leptoiulus belgicus) were present in these highest desiccation tolerances show the strongest stands (see also supplementary material 1). Models of pH, decreases. The same analysis was done for millipedes, but C/N, Mg, and LAI were all significantly influenced by dis- the observed patterns are less clear (Fig. 4b). In general, tance from the forest edge (P < 0.05): pH, Mg, and LAI most millipedes show a slower decrease compared to decreased with increasing distance, whereas C/N ratio of woodlouse species, although the very drought tolerant the mineral top soil increased (Fig. 6). millipede Glomeris marginata shows a very strong decrease after moving further than 16 m into the forest stand. Whereas the trends for individual millipede species vary Discussion considerably, the significant trend for all millipedes (Fig. 1a) resembles the species-specific response of the Arthropod detritivore abundances and species richness most abundant millipede in the data set: P. fuscus. Patterns of abundance of woodlice and millipedes dif- fered as one moved from the edge to the interior of the Community structure of species along edge-to-interior forests. Woodlouse numbers were up to 40 times higher at gradients the forest edge relative to the forest interior, whereas mil- lipede numbers only began to decrease after a certain dis- Community structure of woodlice and millipedes was tance from the forest edge. Riutta et al. (2012) also found most significantly affected by stand, with around 25% of more woodlice in forest edges (7 m) compared to forest
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 462 Pallieter De Smedt et al.
(a) Birch*** Oak*** Pine*** 2 (abundance)/m e Log
(b) Birch Oak** Pine 2 (abundance)/m e Log
Loge (distance)
Fig. 2. Modelled abundance (ln-transformed) of woodlice (a) and millipedes (b) versus distance from the forest edge (ln-transformed) for six forest stands (solid lines, two per tree species) and three dominant tree species (Birch, Oak, Pine), with dots showing the actual data points. Dotted lines denote 1 standard error. All smoothers of tree species contributing significantly to the model are indicated with one or more asterisk(s). Significance levels: *P < 0.05, **P < 0.01, ***P < 0.001. interiors (107 m), but did not find a difference in milli- abundance and reaffirms that forest edges provide more pede abundance in woodlots in the UK. The species optimal conditions for woodlice. A potential explanation composition for woodlice was almost the same as in our is that higher productivity (i.e. higher growth rates; Har- study; however, millipede species composition was differ- per et al., 2005), more niche diversity (through a more ent, making it difficult to compare the results. Tajovsky complex structure of the understory; Harper et al., 2005), et al. (2012) found higher abundances of woodlice in and a higher herb species richness (especially in stands smaller forest fragments (up to 0.8 ha) compared to larger with fewer tree species; Normann et al., 2016) at the edge ones (>4.5 ha). The maximum distance to an edge in the allow for the coexistence of more species, thereby result- small fragments was only 50 m, and therefore the ing in larger total abundances. A higher litter production observed relationship between woodlouse numbers and (Vasconcelos & Luizao,~ 2004) and LAI (Beier & Gunder- forest size could be attributed to a shorter distance to the sen, 1989; Wuyts et al., 2008b) at the forest edge relative forest edge. to the interior has also been reported, but is not reflected Whereas millipede species richness did not change along in the dry mass of the ectorganic horizon in our study. the studied gradient, the number of woodlouse species So, although one would expect higher litter input where decreased towards the forest interior. Although no such the LAI is higher, in deciduous forests (Jonckheere et al., trend was found by Riutta et al. (2012) in the UK, the 2004), at least, this effect does not seem to be translated observed pattern is similar to the one found for total into a larger litter build-up. We attribute this to the
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 Arthropod detritivores in forest edges 463
Birch Oak Pine Mass ectorganic horizon (g)
Loge (distance)
Fig. 3. Modelled mass of the ectorganic horizon versus distance from the forest edge (ln-transformed) for the tree dominant tree species (solid lines), with dots showing the actual data points. Dotted lines denote 1 standard error. All smoothers contributed significantly to the model (P < 0.05). higher abundances and functionality of macro-detritivores because of drier conditions and thus show behavioural in forest edges. Many studies indeed demonstrated a fas- adaptive responses rather than physiological ones (own ter rate of decomposition when macro-detritivores are observations; Gregory, 2009). Based on our sampling present (Wolters, 2000; Vasconcelos & Luizao,~ 2004; scheme, it is difficult to investigate whether the lower Hattenschwiler€ et al., 2005 Riutta et al., 2012; Slade & abundances at the forest edge are real or due to move- Riutta, 2012). ment into the soil. The maximum in the overall millipede abundance at intermediate distances from the edge can be attributed to Species-specific responses the presence of P. fuscus. This typical forest species with- stands acidic environments (Berg et al., 2008), which Considering edge responses at the species level reveals could explain why it was most common species in our highly species-specific patterns. In general, the abundance study on acidic sandy soils. Although P. fuscus seems to of species with high desiccation resistance declined more have a peak in abundance at about 14 m inside the forest strongly along edge-to-interior gradients, and small spe- (and thus resembles the pattern shown by woodlouse spe- cies with very low desiccation resistance found optimal cies with low drought tolerance), it has been reported to conditions deeper inside the forest, with some even show- have a broad ecological amplitude towards humidity ing what appears to be a maximum abundance a few (Berg et al., 2008). The other millipede species showed an metres inside the forest stand. Dias et al. (2013) also sug- exponential decrease in abundance in moving towards the gested that a negative relation between site moisture centre of the forests, similar to the responses observed for (which is lower at forest edges) and body size could most woodlouse species. G. marginata was the exception, reduce the relative abundance of smaller species (such as showing no changes in abundance at distances smaller H. danicus and T. pusillus). These species, however, are than 15 m from the forest edge. This is likely the result of known to move deeper into the soil during summer the species’ ability to withstand very dry conditions by
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 464 Pallieter De Smedt et al.
(a) T. pusillus H. danicus P. muscorum O. asellus P. scaber 2 (abundance)/m e Log
2.4 ± 0.6 h 2.8 ± 0.4 h 22.1 ± 7.3 h 31.8 ± 12.1 h 51.3 ± 14.8 h (b) C. rawlinsii P. denticulatus P. fuscus C. punctatus G. marginata 2 (abundance)/m e Log
Loge (distance)
Fig. 4. Abundance (ln-transformed) of woodlouse (a) and millipede (b) species versus distance from the forest edge (ln-transformed), showing modelled (solid lines) and actual data points (dots). Dotted lines denote 1 standard error. Species were ordered from left to right according to increasing desiccation resistance at 85% relative humidity (average survival time, in hours) (Dias et al., 2013). Millipedes were ordered based on yet unpublished data, using the same methods as for woodlice (M. P. Berg, pers. comm.). folding its body into a sphere to conserve moisture. The Soil moisture, temperature, and leaf litter quality are of abundance of this drought-tolerant species decreases uttermost importance for the abundance and activity of strongly from 15 m onwards, similarly to the responses of woodlice and millipedes (Warburg, 1964; Hopkin & Read, the more drought-tolerant woodlouse species. Drought 1992; David & Handa, 2010; Hornung, 2011). From forest tolerance thus appears to be an important trait in deter- edge to interior, soil moisture increases (Chen et al., 1995; mining the distribution of arthropod detritivores along Gehlhausen et al., 2000), whereas temperature and litter edge-to-interior gradients. quality tend to decrease (Matlack, 1993; Chen et al.,
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 Arthropod detritivores in forest edges 465
Table 3. Results of a permutational analysis of variance, relating pattern could not be detected for millipedes in our study, the variation in community structure of woodlice and millipede which may indicate different strategies to conserve mois- communities to distance (loge), tree species, and stand. ture. A microcosm experiment by Collison et al. (2013) demonstrated that under low moisture conditions, litter P-value P-value decomposition was solely realised by two woodlouse spe- % var. location dispersion cies, P. scaber and P. muscorum, whereas the millipede explained effect effect species G. marginata was found to be rolled up and less Woodlice Distance 3.16 * NS active. G. marginata, however, can survive around 12 Tree species 11.9 *** NS times longer under dry circumstances than, e.g. P. scaber Stand 24.5 *** ** (M. P. Berg, pers. comm.). Therefore, behaviour may be Millipedes Distance 2.48 NS NS an important factor in governing the distribution patterns Tree species 11.2 *** ** of millipedes along forest edge-to-interior gradients. *** *** Stand 25.2 Although both woodlouse and millipede species contribute Woodlice Distance 3.64 *** * to litter decomposition under higher moisture conditions, + Millipedes Tree species 9.53 *** *** Stand 22.5 *** * under dry conditions millipedes burrow deeper into the soil and remain inactive for a longer period than woodlice Results show percentage of variance explained and P-values of (David & Handa, 2010). Sterzynska et al. (2015) reported the assumed location effect. P-value dispersion effect represents that woodlouse communities were more affected by the P-value for the multivariate homogeneity of dispersions test. changes in soil moisture than millipede communities. The Significance levels: *P < 0.05, **P < 0.01, ***P < 0.001, NS: woodlouse community that they investigated, however, > P 0.0.5. consisted mostly of species with low desiccation tolerance, such as trichoniscid and ligiid species. Overall, our results suggest that millipedes find a more favourable microcli- st PORCSCAB 1 axis: 35.4% mate at a certain distance from the edge, where humidity LAI 2nd axis: 20.5% is higher and temperature lower.
PROTFUSC Mg litter
PHILMUSC JULUSCAN Effects on community structure OMMASABU CYLIPUNC ONISASEL POLYDENT pH Whereas individual species showed specific responses C/N 0-5 TRICPUSI CRASRAWL GLOMMARG HAPLDANI regarding forest edge effects, communities were overall less variable along the studied gradients. Moreover, speci- fic stand effects, irrespective of forest typology, explained one fourth of all variation in community structure across Oak Birch Pine all studied gradients. Other local factors are thus likely to be important for local arthropod detritivore composition. Fig. 5. PCA of arthropod detritivores distribution patterns and We found a dominance of woodlice over millipedes in related variables along a forest edge-to-interior gradient. Symbols pine stands and the reverse pattern in birch stands. In oak (further described in figure) represent dominant tree species. Nor- stands, neither group was dominant over the other. Topp mal arrows represent woodlouse and millipede species. Bold et al. (2006) also reported differences in woodlouse and arrows represent environmental variables: pH: pH(KCl) of min- millipede species composition in primeval oak forests eral topsoil, Mg litter: Mg concentration of the ectorganic hori- (Quercus polycarpa/cerris) compared to beech/fir forests zon, C/N 0–5: C/N ratio of mineral topsoil, LAI, Leaf Area (Fagus sylvatica/Abies alba). For tropical forests in Puerto Index in August. Percentage of variation explained by the first and second axis is indicated in the figure. Only species occurring Rico, Richardson et al. (2005) also found significant in at least three samples across all distances and stands are pre- effects of forest type on woodlouse and millipede abun- sented. dances. They suggested that this effect was more impor- tant than direct effects of temperature and rainfall. The importance of tree species and stand in determining the 1999; Vasconcelos & Luizao,~ 2004; Harper et al., 2005; community structure of macro-detritivores is supported Wuyts et al., 2013). Although soil moisture and tempera- by our data. ture measurements were not incorporated in the design of Soil acidity, C/N ratio of the mineral top soil, Mg con- this study, Remy (pers. comm.) did record higher temper- tent in the ectorganic layer, and the LAI in late summer atures and lower soil moisture in the edges of the same are all related to changes in macro-detritivore species forest stands. A warmer forest edge with higher litter composition among and within forests. Indeed, soil acidity quality could favour the overall abundance of macro-det- and exchangeable base cation concentrations in the soil ritivores, whereas highly species-specific responses to soil can strongly influence the spatial distribution of woodlice moisture could explain differences in edge-to-interior (Van Straalen & Verhoef, 1997) and millipedes (Kime, patterns between the different woodlouse species. This 1992). For both taxonomic groups, calcium and
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 466 Pallieter De Smedt et al.
(a) (b) 4 pH C/N
(c) (d) LAI Mg (meq/kg)
Loge (distance)
Fig. 6. Models of environmental variables (solid lines) significantly related to species’ composition (as determined by PCA) versus dis- tance (ln-transformed) from the forest edge, with dots showing the actual data points. Black lines and dots represent birch stands, light grey are oak stands and dark grey lines are pine stands. Subplots show (a) pH(KCl) of the mineral topsoil, (b) C/N ratio of mineral top- soil, (c) Mg concentration of the ectorganic horizon, (d) Leaf Area Index in August. For all four variables, distance contributed signifi- cantly to the model. Dotted lines denote 1 standard error. Data have already been published in previous studies by Wuyts et al., 2008b, c, 2011, 2013. magnesium are required for the construction of the as well). Another important factor for woodlice and milli- exoskeleton, and these macro-nutrients are (together with pedes is the quality of available food sources. Organic sodium and potassium) important cations in the haemo- matter with low C/N ratio is preferred (Hassall et al., lymph (Hopkin & Read, 1992). Soil acidification causes 2002; David, 2009; David & Handa, 2010; Gerlach et al., base cation concentrations to decrease and exchangeable 2014), as lower C/N ratios have been shown to result in aluminium concentrations to increase (Bowman et al., higher woodlouse population stability (Kautz et al., 2000) 2008). Therefore, given that all studied forest types were and higher assimilation efficiency (Loureiro et al., 2006). found to have very acidic soils (pH(KCl) <3.5), we would Wuyts et al. (2011) previously reported lower C/N ratios expect low exchangeable base cation concentrations and of the ectorganic horizon and the mineral topsoil at the high availability of aluminium in the soil solution. Soe- edge of the studied forests compared to the forest interi- jono Sastrodihardjo and Van Straalen (1993) studied pH ors. They suggested that this was a result of higher N preferences of woodlice and showed that most of our deposition at the forest edge. Therefore, it is expected that studied species preferred pH(H2O) values ranging between litter at forest edges should be a better food source for 5 and 6 (equivalent to a pH(KCl) of about 4–5 in our woodlice and millipedes, resulting in larger populations. study). Soil pH, as well as the exchangeable base cation Surprisingly, no strong correlation was found between concentration, was higher at the forest edge compared to species abundance and LAI, except for P. scaber, which the interior (Wuyts et al., 2013). More favourable chemi- was the most abundant species in the forest edge. This cal soil conditions at the forest edge could help to drive could indicate that light availability is of minor impor- larger abundances of woodlice (and probably millipedes tance for most species, or that its effect is overruled by
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 Arthropod detritivores in forest edges 467 other factors, such as the ones mentioned above. A more Table S1. Abundances of woodlice and millipedes per likely explanation is that the range in LAI is not big m2 in the different forest types (mean standard error) enough to detect effects on species composition, however. averaged across all distances. It remains very difficult to draw conclusions about the main abiotic factors driving macro-arthropod communi- ties at forest edges, as most of these abiotic factors are References related to distance from the forest edge and are therefore strongly inter-correlated. Anderson, J.M. (1988) Invertebrate-mediated transport processes in soils. Agriculture, Ecosystems and Environment, 24,5–19. Anderson, M.J. (2001) A new method for non-parametric multi- 26 – Conclusion variate analysis of variance. Austral Ecology, ,32 46. Anderson, M.J., Ellingsen, K.E. & McArdle, B.H. (2006) Multi- variate dispersion as a measure of beta diversity. Ecology Let- Woodlice and millipedes showed clear differences in abun- ters, 9, 683–693. dance with distance from the forest edge. Whereas wood- Andersson, G., Meidell, B.A., Scheller, U., Winqvist, J.-A., louse numbers decreased exponentially with distance from Osterkamp Madesen, M., Djursvoll, P., Budd, G. & Garden-€ the edge, millipede abundances only began to decline at fors, U. (2005) Nationalnyckeln till Sveriges flora och fauna. distances greater than about 7 m from the edge. Similar Mangfotingar. Myriapoda. ArtDatabanken, SLU, Uppsala, patterns were found for species richness, showing that Sweden. arthropod detritivores are rarer in the centre than at for- Beier, C. & Gundersen, P. (1989) Atmospheric deposition to the est edges. The observed patterns were consistent within edge of a spruce forest in Denmark. Environmental pollution, 60 – monocultures of different tree species and between forest , 257 271. Berg, M.P., Soesbergen, M., Tempelman, D. & Wijnhoven, H. stands for woodlice, but less for millipedes. Species com- (2008) Verspreidingsatlas Nederlandse landpissebedden, duizend- position was, however, predominantly determined by for- poten en miljoenpoten (Isopoda, Chilopoda, Diplopoda). EIS- est typology. Despite this general pattern, species-specific Nederland, Leiden & Vrije Universiteit – Afdeling Dierecologie, responses were found. For the five dominant species of Amsterdam, The Netherlands. woodlice in our study, we observed that species with a Berg, M.P. & Wijnhoven, H. (1997) Landpissebedden – Weten- lower desiccation resistance retreated deeper into the for- schappelijke Mededelingen KNNV 221. KNNV Uitgeverij, est and showed a different edge pattern than more Utrecht, The Netherlands. drought-tolerant species. The observed edge-to-interior Bowman, W.D., Cleveland, C.C., Halada, L., Hre sko, J. & gradients in arthropod detritivores, which are key in litter Baron, J.S. (2008) Negative impact of nitrogen deposition on 1 – decomposition, are hypothesised to also give rise to edge soil buffering capacity. Nature Geoscience, , 767 770. Chen, J., Franklin, J.F. & Spies, T.A. (1995) Growing-season gradients in nutrient mineralisation and nutrient turnover microclimatic gradients from clearcut edges into old-growth rates, thereby influencing nutrient cycling in fragmented Douglas-fir forests. Ecological Applications, 5,74–86. forest landscapes. Small forest fragments with a large pro- Chen, J., Saunders, S.C., Crow, T.R., Naiman, R.J., Brosofske, K.D., portion of edge to interior area are thus expected to have Mroz, G.D., Brookshire, B.L. & Franklin, J.F. (1999) Microclimate a disproportional and stronger impact on this functioning in forest ecosystems and landscape ecology. Variations in local cli- relative to more continuous forests. mate can be used to monitor and compare the effects of different management regimes. BioScience, 49,288–297. Collison, E.J., Riutta, T. & Slade, E.M. (2013) Macrofauna Acknowledgements assemblage composition and soil moisture interact to affect soil ecosystem functions. Acta Oecologia, 47,30–36. David, J.-F. (2009) Ecology of millipedes (Diplopoda) in the con- P.D.S. and W.P. each hold a doctoral fellowship of the 81 – – text of global change. Soil Organisms, , 719 733. Research Foundation Flanders (FWO). A.D.S. and David, J.-F. & Handa, I.T. (2010) The ecology of saprophagous P.D.F. each hold a postdoctoral fellowship of the macroarthropods (millipedes, woodlice) in the context of global Research Foundation – Flanders (FWO). We would like change. Biological Reviews, 85, 881–895. to thank the Agency for Nature and Forest (ANB) and De Schrijver, A., Devlaeminck, R., Mertens, J., Wuyts, K., the private forest owners and their ground keepers for Hermy, M. & Verheyen, K. (2007) On the importance of incor- allowing access to their forests. L.B. and D.B. are funded porating forest edge deposition for evaluating exceedance of by the FWO research network on the eco-evolutionary critical pollutant loads. Applied Vegetation Science, 10, 293– dynamics of biotic interactions. Special thanks to Luc 298. Willems for technical support during the research. Delgado, J.D., Arroyo, N.L., Arevalo, J.R. & Fernandez-Pala- cios, J.M. (2007) Edge effects of roads on temperature, light, canopy cover, and canopy height in laurel and pine forests (Tenerife, Canary Islands). Landscape and Urban Planning, 81, Supporting Information 328–340. Dias, A.T.C., Krab, E.J., Mari€en, J., Zimmer, M., Cornelissen, Additional Supporting Information may be found in the online J.H.C., Ellers, J., Wardle, D.A. & Berg, M.P. (2013) Traits version of this article under the DOI reference: doi: 10.1111/ underpinning desiccation resistance explain distribution pat- icad.12183: terns of terrestrial isopods. Oecologia, 172, 667–677.
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 468 Pallieter De Smedt et al.
Echeverria, C., Coomes, D.A., Hall, M. & Newton, A.C. (2008) leaf area index determination Part I. Theories, sensors and Spatially explicit models to analyse forest loss and fragmenta- hemispherical photography. Agricultural and Forest Meteorol- tion between 1976 and 2020 in southern Chile. Ecological Mod- ogy, 121,19–35. elling, 212, 439–449. Kautz, G., Zimmer, M. & Topp, W. (2000) Responses of the Fletcher, R.J., Ries, L., Battin, J. & Chalfoun, A.D. (2007) The parthenogenetic isopod, Trichoniscus pusillus (Isopoda: Onisci- role of habitat and edge in fragmented landscapes: definitively dae), to changes in food quality. Pedobiologia, 44,75–85. distinct or inevitably interwined? Canadian Journal of Zoology, Kime, R.D. (1992) On abundance of West-European millipedes 85, 1017–1030. (Diplopoda). Proceedings of the 8th International Congress of Gascon, C., Williamson, G.B. & da Fonseca, G.A.B. (2000) Myriapodology. Berichte des Naturwissenschaftlich-medizi- Receding forest edges and vanishing reserves. Science, 288, nischen Vereins in Innsbruck, 10, 393–399. 1356–1358. Lavelle, P. (1997) Faunal activities and soil processes: adaptive Gehlhausen, S., Schwartz, M.W. & Augspurger, C.K. (2000) strategies that determine ecosystem function. Advances in Eco- Vegetation and microclimatic edge effects in two mixed-meso- logical Research, 27,93–132. phytic forest fragments. Plant Ecology, 147,21–35. Loureiro, S., Sampaio, A., Brandao,~ A., Nogueira, A.J.A. & Gerlach, A., Russell, D.J., Jaeschke, B. & Rombke,€ J. (2014) Soares, A.M.V.M. (2006) Feeding behaviour of the terrestrial Feeding preferences of native terrestrial isopod species (Onis- isopod Porcellionides pruinosus Brandt, 1833 (Crustacea, Iso- cidea, Isopoda) for native and introduced litter. Applied Soil poda) in response to changes in food quality and contamina- Ecology, 83,95–100. tion. Science of the Total Environment, 369, 119–128. Gregory, S. (2009) Woodlice and Waterlice (Isopoda: Oniscidea & Asel- Matlack, G.R. (1993) Microenvironment variation within and lota) in Britain and Ireland. FSC Publications, Shrewsbury, UK. among forest edge sites in the eastern United States. Biological Grelle, C., Fabre, M.-C., Lepr^etre, A. & Descamps, M. (2000) Conservation, 66, 185–194. Myriapod and isopod communities in soils contaminated by Meyer, E. & Eisenbeis, G. (1985) Water relations in millipedes heavy metals in northern France. European Journal of Soil from some alpine habitat types (Central Alps, Tyrol) (Diplo- Science, 51, 425–433. poda). Bijdragen tot de Dierkunde, 55, 131–142. Haacker, U. (1968) Descriptive, experimental and comparative Murcia, C. (1995) Edge effects in fragmented forests: implications investigations on the ecology of Diplopoda of the Rhein-Main- for conservation. Trends in Ecology and Evolution, 10,58–62. Region. Oecologia, 1,87–129. Muys, B. & Lust, N. (1992) Inventory of the earthworm commu- Harper, K.A., Macdonald, S.E., Burton, P.J., Chen, J., Bro- nities and the state of litter decomposition in the forests of sofske, K.D., Saunders, S.C., Euskirchen, E.S., Roberts, D., Flanders, Belgium, and its implications for forest management. Jaiteh, M.S. & Esseen, P.-A. (2005) Edge influence on forest Soil Biology and Biochemistry, 24, 1677–1681. structure and composition in fragmented landscapes. Conserva- Normann, C., Tscharntke, T. & Schreber, C. (2016) How forest tion Biology, 19, 768–782. edge-center transitions in the herb layer interact with beech Hassall, M., Tuck, J.M., Smith, D.W., Gilroy, J.J. & Addison, dominance vs. tree diversity. Journal of Plant Ecology, R.K. (2002) Effects of spatial heterogeneity on feeding beha- doi:10.1093/jpe/rtw004. [Epub ahead of print]. viour of Porcellio scaber (Isopoda: Oniscidea). European Jour- Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, nal of Soil Biology, 38,53–57. P.R., O’Hara, R.B., Simpson, G.L., Solymos, P., Stevens, Hattenschwiler,€ S., Tiunov, A.V. & Scheu, S. (2005) Biodiversity M.H.H. & Wagner, H. (2015) vegan: Community Ecology and litter decomposition in terrestrial ecosystems. Annual Package. R package version 2.2-1.
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469 Arthropod detritivores in forest edges 469
Sterzynska, M., Tajovsky, K. & Nicia, P. (2015) Contrasting Wolters, V. (2000) Invertebrate control of soil organic matter sta- responses of millipedes and terrestrial isopods to hydrologic bility. Biology and Fertility of Soils, 31,1–19. regime changes in forested montane wetlands. European Journal Wood, S.N. (2006) Generalized Additive Models: An Introduction of Soil Biology, 68,33–41. with R. Chapman and Hall/CRC, Boca Raton, Florida. Sutton, S.L. (1972) Woodlice. Ginn & company limited, London, Wuyts, K., De Schrijver, A., Staelens, J., Gielis, M., Geudens, G. UK. & Verheyen, K. (2008b) Patterns of throughfall deposition Tajovsky, K., Ho sek, J., Hofmeister, J. & Wytwer, J. (2012) along a transect in forest edges of silver birch and Corsican Assemblages of terrestrial isopods (Isopoda, Oniscidea) in a pine. Canadian Journal of Forest Research, 38, 449–461. fragmented forest landscape in Central Europe. ZooKeys, 176, Wuyts, K., De Schrijver, A., Staelens, J., Gielis, L., Vanden- 189–198. bruwane, J. & Verheyen, K. (2008c) Comparison of forest edge Topp, W., Kappes, H., Kulfan, J. & Zach, P. (2006) Distribution effects on throughfall deposition in different forest types. Envi- pattern of woodlice (Isopoda) and millipedes (Diplopoda) in ronmental Pollution, 156, 854–861. four primeval forests of the Western Carpathians (Central Wuyts, K., De Schrijver, A., Staelens, J., Van Nevel, L., Adri- Slovakia). Soil Biology & Biochemistry, 38,43–50. aenssens, S. & Verheyen, K. (2011) Soil inorganic N leaching Van Straalen, N.M. & Verhoef, H.A. (1997) The development of in edges of different forest types subject to high N deposition a bioindicator system for soil acidity based on arthropod pH loads. Ecosystems, 14, 818–834. preferences. Journal of Applied Ecology, 34, 217–232. Wuyts, K., De Schrijver, A., Staelens, J. & Verheyen, K. (2013) Vasconcelos, H.L. & Luizao,~ F.J. (2004) Litter production and Edge effects on soil acidification in forests on sandy soils under litter concentrations in a fragmented Amazonian landscape. high deposition load. Water, Air, and Soil Pollution, 224,1–14. Ecological Applications, 14, 884–892. Wuyts, K., Verheyen, K., De Schrijver, A., Cornelis, W.M. & Wade, T.G., Riitters, K.H., Wickham, J.D. & Jones, F.B. (2003) Gabriels, D. (2008a) The impact of forest edge structure on Distribution and causes of global forest fragmentation. Conser- longitudinal patterns of deposition, wind speed, and turbulence. vation Ecology, 7,7. Atmosperic Environment, 42, 8651–8660. Warburg, M.R. (1964) The response of isopods towards tempera- ture, humidity and light. Animal Behavior, 12, 175–186. Accepted 7 June 2016 Warton, D.I., Wright, S.T. & Wang, Y. (2012) Distance-based First published online 1 July 2016 multivariate analyses confound location and dispersion effects. Methods in Ecology and Evolution, 3,89–101. Editor: Raphael Didham Weathers, K.C., Cadenasso, M.L. & Pickett, S.T.A. (2001) Forest Associate editor: Thomas Bolger edges as nutrient and pollutant concentrators: potential syner- gisms between fragmentation, forest canopies, and the atmo- sphere. Conservation Biology, 15, 1506–1514.
Ó 2016 The Royal Entomological Society, Insect Conservation and Diversity, 9, 456–469