We found that herb and canopy cover as well as spatial effects were the best predictors of Diptera community com-position, consisting of 62 families, including 99 empi-doidea and 78 Phoridae species. abundance of empidoidea was positively influenced by herb diversity, indicating bot-tom-up control. a complex causal pathway influenced Dip-teran species richness: species-rich forest stands, with low beech cover, had lower canopy cover, resulting in higher Dipteran species richness. In addition, Diptera benefited from a more dense and diverse herb community. Individ-ual species responded differentially to herb layer diversity, indicating that effects of plant diversity on higher trophic levels depend on species identity. We conclude that tree and herb canopy cover as well as herb diversity predomi-nately shape Dipteran communities in temperate deciduous forests, which is in contrast to expectations from grassland studies exhibiting much closer relationships between plant and insect diversity.
Vegetation characteristics such as the diversity and produc-tivity of plants strongly influence the distribution and abun-dance of animals. For example, plant diversity has been shown to enhance insect diversity and abundance (Knops et al. 1999; haddad et al. 2009; scherber et al. 2010). more diverse plant communities provide a more heterogeneous resource base that allows a greater number of herbivorous insect species to coexist, an effect that can cascade up to higher trophic levels (hunter and Price 1992; siemann et al. 1998). In addition, higher structural complexity
Abstract Biodiversity experiments have shown that plant diversity has largely positive effects on insect diversity and abundance. however, such relationships have rarely been studied in undisturbed and more complex ecosystems such as forests. Flies (Diptera) are among the most domi-nant taxa in temperate ecosystems, influencing many eco-system processes. as it is unknown how Diptera respond to changes in forest biodiversity, we examined how com-munity characteristics of Diptera respond to varying lev-els of tree and herb diversity and vegetation structure. the study was conducted in the hainich national Park (Central germany) on 84 plots along a gradient of tree (from two to nine species) and herb (from two to 28 species) diversity.
Communicated by andreas Prinzing.
C. scherber and e. a. Vockenhuber contributed equally to this study.
Electronic supplementary material the online version of this article (doi:10.1007/s00442-013-2865-7) contains supplementary material, which is available to authorized users.
C. scherber (*) · e. a. Vockenhuber · t. tscharntke agroecology, Department of Crop sciences, university of göttingen, grisebachstrasse 6, 37077 göttingen, germanye-mail: [email protected]; [email protected]
a. stark seebener strasse 190, 06114 halle (saale), germany
a. stark senckenberg Deutsches entomologisches Institut, eberswalder strasse 90, 15374 müncheberg, germany
h. meyer Institut für Ökosystemforschung, abteilung angewandte Ökologie, Christian-albrechts-universität zu Kiel, Olshausenstrasse 75, 24118 Kiel, germany
associated with diverse plant communities can positively influence insect diversity and abundance (lawton 1983; siemann et al. 1998; Brose 2003).
While these plant–insect diversity relationships have been frequently studied in (experimental) grassland eco-systems (siemann et al. 1998; Koricheva et al. 2000; haddad et al. 2009), little is known about whether these patterns also hold true for forest ecosystems (Basset et al. 2012). Compared to grasslands, forests are characterised by a higher complexity due to an overall higher biovolume and the organisation of the plant community in distinct layers. tree and herb layer diversity often show linkages (Barbier et al. 2008; Vockenhuber et al. 2011) and both may affect the structure of insect communities in forests. however, effects of herb layer characteristics on insect communities have rarely been considered so far (but see humphrey et al. 1999; hirao et al. 2009), even though the herb layer contains most of a forest’s plant diversity (gil-liam 2007). While herb diversity may be expected to have a stronger influence on insect communities in the herb layer through direct bottom-up processes, the diversity of trees strongly determines overall environmental conditions in forest stands and may consequently also shape insect communities.
Diptera, a hyperdiverse insect order (Pape et al. 2009), represent a major component of all non-marine ecosys-tems and are often the most abundant animals in temperate habitats (hughes et al. 2000; thompson 2009). they fulfil a great variety of ecological functions, acting as predators, parasitoids, herbivores, detritivores and pollinators (Ooster-broek 2007). In spite of their high ecological importance, Diptera are frequently overlooked in ecological studies (Woodcock et al. 2003; allgood et al. 2009). While grass-land plant diversity effects on arthropods (including Dip-tera) have been well studied (haddad et al. 2009; scherber et al. 2010), it remains to be tested whether these relation-ships also hold in other ecosystem types, including forests (naeem et al. 2012). Plant diversity can be hypothesized to be positively related to insect taxa with detritivorous, soil-dwelling larvae because higher plant diversity above-ground leads to greater heterogeneity of carbon substrates belowground via the input of more diverse leaf litter and root exudates (Bais et al. 2006; hättenschwiler and gasser 2005). Predatory taxa, on the other hand, can benefit both from a greater diversity of prey and a greater structural complexity in more diverse plant communities. Finally, taxa that are predominantly flower visitors or pollinators have also been shown to respond positively to plant spe-cies richness and flower cover (hudewenz et al. 2012). as the Diptera comprise many groups including decomposers, predators, and flower visitors, both richness and abundance of Diptera can be expected to increase with increasing plant diversity.
In our study area, empidoidea and Phoridae represent two highly dominant and functionally diverse Dipteran taxa. this is why we chose them to test if taxa with dif-ferent life history traits react differently to gradients of plant diversity. With the exception of the phytophagous genus Thrypticus (Dolichopodidae), all larval stages of the empidoidea are predators. the adults of empidoidea use a wide range of food sources, but the vast majority of species are zoophagous; it can therefore be expected that empidi-dae are indirectly bottom-up controlled by plant diversity (scherber et al. 2010). especially within the empididae (the true ‘dance flies’), predatory behaviour has evolved into a complex courtship behaviour where males present prey to females to stimulate copulation (Cumming 1994). nutrition in empididae is therefore mostly restricted to nectar or pollen (Chvála 1983), and taxa are obligate flower visitors. In addition, several species of the empidoidea are strongly tied to closed forests.
the family Phoridae encompasses a particularly high variety of different life history traits (Disney 1994) and includes many detritivorous species. as a consequence, the flower-visiting or predatory empidoidea can be expected to be more directly linked to vegetation characteristics than the Phoridae.
In this study, we explore the effects of both tree layer and herb layer diversity on Diptera richness and abundance in germany’s largest connected deciduous forest. We also examine how Dipteran community composition changes in response to biotic and abiotic characteristics. In addi-tion, we test for indirect effects of tree diversity on Dip-tera, modulated by abiotic and biotic characteristics such as canopy structure, soil parameters or the herb layer. We test the following hypotheses:
1. Dipteran abundance and richness are positively affected by tree and herb diversity.
2. tree diversity indirectly influences Dipteran diversity via changes in canopy and herb layer structure.
3. Individual species of Diptera respond differently to changes in tree and herb diversity.
Materials and methods
study area
the study was conducted in the hainich national Park (thuringia, Central germany), near the village of Weber-stedt (51°05′28″n, 10°31′24″O). the hainich forest encompasses 16,000 ha, of which 7,500 ha have national park status. Climatic conditions are sub-atlantic with con-tinental influence; throughout the study period (2008), the mean daily temperature was 9.5 °C and annual precipitation
Oecologia
1 3
was 500.8 mm (meteomedia ag, appenzell, switzerland). the research area is situated at an elevation of approxi-mately 350 m a.s.l. the predominant soil type is (stagnic) luvisol developed from triassic limestone as bedrock with partial loess cover (leuschner et al. 2009).
sampled forest stands have been unmanaged during the last 40 years. all investigated forest stands were mature and held deciduous forest for at least 200 years (mölder 2009). Dominant tree species were Fagus sylvatica l., Tilia platyphyllos scop., Tilia cordata mill., and Fraxinus excel-sior l. Other species such as Carpinus betulus l., Acer campestre l., Acer platanoides l., Acer pseudoplatanus l., Prunus avium l., Sorbus torminalis (l.) Crantz, Quercus robur l., Quercus petraea liebl., and Ulmus glabra huds. were less abundant. the area is exceptionally suitable for the study of tree diversity effects because differences in historic forest-use practises (mostly coppicing and selective cutting) have created a small-scale mosaic of stands differ-ing in tree diversity, while exhibiting comparable climatic and edaphic conditions (leuschner et al. 2009).
study plots
In spring 2008, we established 100 plots based on a priori combinations of tree species (leuschner et al. 2009). Plots were arranged in two locations (each n = 50): lindig and thiemsburg (the distance between the locations is approxi-mately 1.5 km; see Online resource 1). For this study, we only included plots with a minimum inter-plot distance of 30 m to improve spatial independence of insect samples. this yielded 84 plots (44 plots at the lindig location and 40 plots at the thiemsburg location) (Online resource 1). spatial coordinates of each location were included in the statistical models (see below). Plots consisted of a circular area (radius 20 m, area 1,257 m2) for tree layer measure-ments, with herb layer measurements (on 2 × 1-m2 quad-rats) and sampling of Diptera conducted at the plot centre. Plots differed in their tree and herb diversity and vegetation cover (table 1). table 2 gives a summary of explanatory variables and selected abiotic plot characteristics.
Vegetation measurements
herb layer species richness and cover were recorded on two 1 × 1-m quadrats in the plot centre in June 2008. We visually estimated percentage cover of each vascular plant species present. all plants with a height of <1 m were con-sidered, as this threshold is commonly used to delimit the herb layer (gilliam 2007). For every pair of quadrats, we pooled the number of plant species present and calculated mean vegetation cover, giving a total of n = 84 values for herb species richness and cover. the number of herb layer species recorded per plot ranged from two to 28 plant
species, and vegetation cover from 2 to 79 % (table 1). We calculated plant diversity indices for every plot (see “statis-tical analyses” section). the most common herb layer plant species were A. platanoides, F. excelsior, A. pseudoplata-nus, and F. sylvatica (tree saplings), Lamium galeobdolon (l.) l. s. str., Stellaria holostea l., Viola reichenbachi-ana Boreau, and Primula elatior (l.) hill (forbs), and the graminoid Hordelymus europaeus (l.) harz.
to estimate tree diversity, we recorded all tree species within a 20-m radius around the plot centre with diameter at breast height (DBh) of >7 cm. In addition, we meas-ured DBh of all trees with circumference tape and cal-culated the stem area. the number of broad-leaved tree species per plot ranged from two (nearly monospecific F. sylvatica stands) to nine different broad-leaved tree spe-cies. no conifers were present. the most dominant species were F. sylvatica (lindig 5 %, thiemsburg 48 %), Tilia sp. (lindig 44 %, thiemsburg 16 %) and F. excelsior (lindig 23 %, thiemsburg 20 %). T. cordata/T. platyphyllos and Q.
Table 1 summary statistics of the explanatory variables and selected covariates, based on all 84 study plots
eH′ numbers equivalent (Jost 2007) of the shannon–Wiener diver-sity index (H′) under the assumption that all species have equal rela-tive abundance
SE standard error
Variable range mean ± se
Vegetation cover (%) 1.8–86.9 33.6 ± 2.2
Flower cover (%) 0–5 0.32 ± 0.08
Canopy cover (%) 69.1–96.4 90.2 ± 0.5
herb diversity (H′) 0.56–2.92 1.81 ± 0.05
tree diversity (H′) 0.27–1.88 1.26 ± 0.04
herb species richness (eH′) 1.75–18.54 6.11 ± 1.05
tree species richness (eH′) 1.31–6.55 3.54 ± 1.04
soil ph 4.05–6.98 5.33 ± 0.07
humus layer mass (g m−2) 2.5–180.1 51.6 ± 4.5
soil moisture (%) 18.9–35.3 25.5 ± 0.3
Table 2 summary statistics of the response variables
Prop. proportion, SE standard error
Variable minimum maximum mean ± se n
Diptera family richness 16 35 24.7 ± 0.5 70
Diptera abundance 132 643 281 ± 11 70
empidoidea species richness 6 28 15.5 ± 0.6 72
empidoidea abundance 17 248 60 ± 4 70
Prop. flower-visiting empi-doidea
11 56 31.3 ± 1.1 72
Prop. silvicolous empidoidea 50 100 74.7 ± 1.1 72
Phoridae species richness 2 15 6.3 ± 0.3 77
Phoridae abundance 18 255 71 ± 6 70
Oecologia
1 3
robur/Q. petraea were recorded at the genus level as these species could not be reliably distinguished in the field.
Diptera sampling and processing
We used both suction sampling and pan traps to sample Dip-tera with a broad range of habitat requirements. One yellow, one white and one blue plastic pan trap per plot (Fun In a Box ltd., Brentwood, essex, uK; 5 cm deep, 15 cm upper diameter, 8 cm lower diameter) were mounted on wooden posts at 50 cm height and arranged 1.5 m apart in a triangle around the plot centre. Pan trap colours corresponded to the prevailing floral colours in the herb layer. We filled the traps with 200 ml of a 50 % ethylene glycol solution in water with a few drops of unscented dishwashing detergent. traps were set out in the forest for 1 week in may and July 2008.
In July 2008, we performed suction sampling at two locations near the centre of each plot, using a leaf blower (stihl sh56; stihl ag & Co. Kg, Waiblingen, germany) fitted with a cloth bag at the nozzle to hold the insects. We quickly placed a plastic cylinder (basal area 0.25 m2, height 0.8 m) onto the ground vegetation to prevent insects from escaping, and took a suction sample for a duration of 30 s within the cage. Insect samples were sorted in the labora-tory and stored in 70 % ethanol.
all Diptera were identified to family level following Oosterbroek (2007). Family richness is a good indica-tor of species richness in Diptera (Baldi 2003). empi-doidea [including the families empididae, hybotidae, atelestidae, microphoridae, and Dolichopodidae, follow-ing Chvála (1983)] and Phoridae, the two most abundant Dipteran taxa in our study area, were identified to species level (see Online resource 2 for references of identification keys). all empidoidea were classified as to whether they were silvicolous species (predominantly inhabiting forests) and whether they were flower visitors (Online resource 3). Online resources 4 and 5 contain details on the Dipteran families and on the Phoridae species reported by us.
statistical analyses
tree diversity and herb diversity were expressed by the shannon–Wiener diversity index (H′) based on relative stem area and relative cover, respectively. H′ incorporates species richness as well as relative abundances of species (maurer and mcgill 2011). the numbers equivalent of H′, exp(H′), is used to characterize communities using num-bers of equally common species (Jost 2007).
For analyses of total Diptera family richness and abun-dance as well as abundances of empidoidea and Phori-dae, we pooled data from pan trapping (both sampling rounds; yellow, white, and blue pan traps) and suction sam-pling. species richness of empidoidea and Phoridae was
determined from pan traps only (Phoridae only from yellow pan traps). Female phorids of the genus Megaselia were not included in the analysis of phorid species richness since their identification to species level was not possible at the time. We excluded plots from the analyses where pan traps had been overturned or otherwise damaged during either of the two sampling rounds (see table 2 for final sample sizes). Because we were interested in the actual numbers of species in every group, species richness of all Diptera taxa was not transformed (and no diversity indices were calcu-lated), and we used abundance as a covariate in all models [for a justification of this approach see Knops et al. (1999) and sobek et al. (2009b)].
all data were analysed using generalized additive mod-els (gam) to account for non-linear effects of covariates on response variables while retaining all variables in their original units (Wood 2006). to model potential spatial dependencies in our data, we also included a smooth func-tion of the (x, y) coordinates of all sampling points into our models [r code: s(x, y)].
Initial models contained the following covariates in the following sequence: Diptera abundance (only when spe-cies richness was the response variable), canopy cover (%), herb H′ (excluding tree saplings), tree H′ (within a 20-m radius, based on stem area), herb cover (excluding tree saplings), humus layer mass (in kg/m2), flower cover (%), and a smooth term for the (x, y) coordinates (Fahrmeir et al. 2009; Bivand et al. 2013). these covariates were chosen based on our design (tree/herb diversity gradient) and biological knowledge about adult/larval behaviour and habitat preferences of Diptera. We additionally ran machine-learning approaches using model-based boosting (hastie et al. 2009) that turned out an almost identical set of covariates.
models were fitted using a shrinkage version of cubic regression splines with a maximum of four knots; the spatial term was fitted using thin plate regression splines (Wood 2006, pp 157–161). the degree of smoothness of each term was assessed using penalized likelihood maximi-zation. We checked diagnostic plots for homoscedasticity, normality of residuals, and presence of outliers. all explan-atory variables had pairwise spearman rank correlation coefficients <0.43.
For model simplification, we added an extra penalty to each term so that it could potentially be penalized to zero. as quasi-likelihood estimation precluded the calculation of information criteria [e.g. akaike′s information criterion (aIC)] that are based on maximum likelihood (mcCulloch and searle 2001), each model was simplified by sequen-tially removing non-significant terms from the model, comparing the residual deviance of each resulting model with its predecessor. When dropping terms, hierarchy was respected (that is, lower order terms were never removed
Oecologia
1 3
if they were present in higher order interactions). models were considered minimal adequate (Crawley 2013) if fur-ther deletion of terms lead to significant increases in devi-ance (see documentation on gam.selection in Wood 2012).
species-level responses of empidoidea and Phoridae were analysed using multinomial models, where a matrix containing the abundances of each species was used as the response variable. models were fitted using the mul-tinom() function in the nnet library in r (ripley 2013) with the same explanatory variables as in the gam mod-els described above. as multinomial models are inherently nonlinear, we refrained from introducing further nonlinear and/or smooth terms in the explanatory variables. multino-mial models were simplified automatically based on aIC until aIC reached a minimum. We used a strong penalty term [k = log(n)] to force the models to contain only terms with sufficient support by the data.
Finally, to test for direct and indirect pathways among var-iables, we employed structural equation models (sem) using IBm sPss amOs 20.0 (IBm software group, somers, ny). Data were centred and scaled by subtracting the column mean and dividing by the column sD (grace 2006, p 328), missing values were replaced by the mean and all analyses were performed with the resulting data matrix. In addition, we performed bootstrapping using a maximum likelihood bootstrap with 5,000 samples to account for potential non-normality (Blunch 2013, p 235 ff.). We constructed latent var-iables for soil characteristics (indicators: humus layer mass, soil moisture), herbs (indicators: species richness, height and cover) and Diptera (indicators: Family, empidoidea and Phor-idae richness). tree canopy characteristics were entered as observed variables (tree species richness, percent beech, and canopy cover). the initial model structure (following grace et al. 2010) is shown in Online resource 6. all analyses except for the sem analyses were carried out using r, version 3.0.1 (r Development Core team 2013).
Results
Overall Diptera abundance and family richness
In total, we collected 19,641 individuals belonging to 62 families (Online resource 4). Pan trap sampling yielded 94.5 %, while suction sampling yielded 5.5 % of individuals.
assuming that individuals are attracted to a pan trap within a radius of 25 m, an abundance of ca. 18,560 indi-viduals from pan trap sampling alone would correspond to ca. 1 Diptera individual m−2 day−1. Phoridae were by far the most abundant Dipteran family (25.2 % of individu-als), followed by Cecidomyiidae (10.9 %), Dolichopodidae (10.7 %), sciaridae (8.8 %), muscidae (5.5 %), empididae
Diptera abundance was strongly negatively influenced by tree canopy cover (table 3; Fig. 1) and declined from ca. 500 individuals (per plot, summed over 2 weeks) at 70 % canopy cover to ca. 300 individuals at >90 % canopy cover. herb and tree diversity were retained in minimal adequate models (table 3; Figs. 1, 2) but had no significant effects on overall Diptera abundance. In addition, there was a sig-nificant spatial trend in the data, with abundances generally increasing towards the outer margins of the study area (not necessarily the forest margin).
Family richness increased significantly and linearly with Diptera abundance (table 3); in addition, it increased slightly with herb cover (Fig. 2), while humus layer mass had a negative effect on family richness; ca. 25 fly fami-lies were present if the humus layer was poorly developed, while only 23 families were present when the humus layer mass was 0.15 kg m−2 or more.
abundance and species richness of empidoidea
Overall, 4,184 individuals of empidoidea were caught. Catches from pan traps, which were determined to species level, yielded 4,077 individuals belonging to 99 species (Online resource 3). about a quarter of all recorded empi-doidea species (26.6 %) were classified as flower visitors, and more than half of all species were silvicolous (54.5 %) (Online resource 3). empidoidea abundance was strongly positively affected by herb diversity (H′; table 3; Fig. 2): at low herb diversity (from one to two species), only ca. 50 individuals were caught; however, if herb diversity increased to around 10–12 species, 100–200 individuals were caught (Fig. 2).
empidoidea species richness depended on sample size (Diptera abundance; table 3), and there was a significantly negative effect of tree canopy cover (Fig. 1). the propor-tion of flower-visiting empidoidea (based on individuals) was significantly negatively affected by canopy cover: at 70 % canopy cover, 40 % of individuals were flower visi-tors, while only 20 % of individuals were flower visitors at higher canopy cover. the proportion of flower-visiting indi-viduals increased linearly with herb cover: 20 % of individ-uals were flower visitors at <10 % herb cover, while around 40 % were flower visiting if the herb cover reached 50 % (Fig. 2). at the species level, proportions of flower visitors showed more complex patterns: at low and high values of canopy cover, around 40 % of species were flower visitors, while more species were flower visitors at intermediate levels of canopy cover. tree species richness had a slightly negative effect on the proportion of flower-visiting species; humus layer mass had a complex non-linear effect similar to the canopy cover effect. Finally, across all empidoidea
Oecologia
1 3
data, there was an additional significant spatial trend in the data with a local minimum between sampled forest stands.
abundance and species richness of Phoridae
In total, we recorded 4,955 phorids, of which 2,483 indi-viduals were caught in yellow pan traps and analysed to species level. altogether, 78 species were found (Online resource 5). Phorid abundance increased very slightly with
herb cover, without significant spatial effects. Phorid spe-cies richness was significantly affected by sample size (table 3) and showed a significant quadratic relationship with herb species diversity (Fig. 2) with a maximum at intermediate levels of herb diversity. In addition, phorid richness increased quadratically with flower cover, with highest species numbers at 0 or 5 % flower cover, and lower numbers in between. In addition, there was a spatial trend in the data as in the other analyses described above.
Table 3 summary of terms for generalized additive models
note that intercept terms are on the scale of the linear predictor (log). F-values are given for smooth terms, t-values for parametric terms. P-val-ues test the null hypothesis that the intercept is zero, or give the approximate significance of each smooth term
s smooth function, Est. df estimated degrees of freedom of the smooth term, est. pp estimate of the parametric parameter, Ref. df reference degrees of freedom for smooth terms; for other abbreviations, see tables 1 and 2a approximate parametric interpretation of the smooth functions
response variable Parameter est. df (est. pp) effecta ref. df (se) F-value (t-value) P Deviance explained
multinomial models showed that only herb diversity, can-opy cover, herb and flower cover, and distance to the for-est edge influenced individual species (Fig. 3). Below, we describe patterns in the relative abundance of the five most abundant species in our samples.
Metopina galeata haliday (Phoridae) declined strongly with herb diversity but increased with herb, flower or can-opy cover. In addition, the relative abundance of M. galeata increased deep inside the forest (700–800 m distance to the forest edge). Metopina braueri strobl (Phoridae) showed almost the opposite patterns: it was very strongly positively affected by herb diversity, decreased with herb, canopy and flower cover and was found rather close to the forest edge. Chrysotus angulicornis Kowarz (empidoidea) often showed hump-shaped responses, peaking at intermediate levels of herb diversity, herb cover or distance to the forest edge. Sciapus platypterus Fabricius (empidoidea) declined with herb diversity and cover, and showed a hump-shaped response to the forest edge. Finally, Leptopeza flavipes meigen (empidoidea) increased with herb diversity and
declined with herb cover; it mainly occurred deep inside the forest (>700 m).
Direct and indirect pathways among variables
structural equation modelling showed that the Diptera communities (family richness, Phoridae and empidoidea richness) were largely driven by canopy and herb layer characteristics (Fig. 4; Online resources 6–12). the most parsimonious model had χ2 = 40.16 and P = 0.375 at 38 df, residual mean square of approximation = 0.026, range 0.00–0.083; P = 0.692. Because there were some indica-tions of non-normality (skewness and kurtosis, see Online resource 11), we additionally performed a Bollen–stine bootstrap with 5,000 samples that resulted in a P-value of 0.465—i.e. the assumed causal structure was still well supported by our data. there were no severe outliers in the structural equation model (Online resource 12). tree spe-cies richness was negatively related to percent beech (path coefficient r = −0.62); percent beech had a positive effect on canopy cover (r = 0.36) which negatively affected the Dipteran communities (r = −0.46). soil parameters (e.g.
0.5 1.0 1.5 2.0 2.5
100
200
300
400
500
600
700
Herb diversity (H´)
Dip
tera
abu
ndan
ce
0 10 20 30 40 50
15
20
25
30
35
Herb cover (percent)
Num
ber
of D
ipte
ra fa
mili
es
0.5 1.0 1.5 2.0 2.5
50
100
150
200
250
Herb diversity (H´)
Em
pido
idea
abu
ndan
ce
0 10 20 30 40 50
0
50
100
150
200
Herb cover (percent)
Pho
ridae
abu
ndan
ce
0.5 1.0 1.5 2.0 2.50
5
10
15
Herb diversity (H´)
Pho
ridae
spe
cies
ric
hnes
s
0 10 20 30 40 50
0.0
0.2
0.4
0.6
0.8
1.0
Herb cover (percent)
Pro
p. o
f flo
wer
−vi
s. E
mpi
doid
eaFig. 1 effects of herb diversity (H′) and herb cover on Diptera in the hainich national Park. Solid lines show model predictions, dashed lines show ±1 se of the predictions
Oecologia
1 3
humus layer mass or soil moisture) were affected by per-cent beech and only vaguely related to Diptera, but none of these effects were significant. By contrast, herb layer characteristics (species richness, height and cover) had significantly positive effects on the Dipteran community (r = 0.53). Overall, variables related to the herb layer had the strongest direct effect on the Dipteran community, closely followed by canopy cover.
Discussion
We have shown that characteristics of the herb and tree layer in forests are closely linked to community charac-teristics of Diptera, an often overlooked insect taxon. Our findings are particularly noteworthy because Diptera are often (falsely) considered unspecialized detritivores with few relationships to aboveground habitat characteristics (savage 2002). It should be noted that the scope of this study included sampling of adult Diptera only. however, tree and herb layer characteristics can also affect soil-dwelling larvae via altered soil and litter layer heterogene-ity due to plant-specific root exudates and differences in litter quality.
effects of herb diversity
We found that herb diversity influenced empidoidea and Phoridae. the empidoidea responded more strongly to herb diversity than the Phoridae, which indicates that the empidoidea, which include several flower visitors (Chvála 1983), may be more strongly affected by bottom-up pro-cesses (scherber et al. 2010). Overall, we had hypothesized that a diverse herb layer should support higher insect abun-dances due to a greater variety of (alternative) food sources available for herbivores, an effect which could cascade up to higher trophic levels (hunter and Price 1992). Further, the higher structural complexity in diverse plant commu-nities may create habitat conditions suitable for holding greater insect numbers (siemann et al. 1998), for instance because more structures and surfaces for egg deposition, sheltering, hunting or overwintering are available (strong et al. 1984). given the results of the present study, the empidoidea were more strongly affected by herb diver-sity than the Phoridae, which may be due to their preda-tory behaviour (hunting on leaf surfaces, e.g. hybotidae) or flower visitation (especially empididae).
Contrary to our expectations, Dipteran family richness as well as species richness of empidoidea and abundance
Fig. 2 effects of tree canopy cover and tree diversity (H′) on Diptera in the hainich national Park. Solid lines show model predictions, dashed lines show ±1 se of the predictions
70 75 80 85 90 95
100
200
300
400
500
600
700
Canopy cover (percent)D
ipte
ra a
bund
ance
0.5 1.0 1.5
100
200
300
400
500
600
700
Tree diversity (H´)
Dip
tera
abu
ndan
ce
0.5 1.0 1.5
0.0
0.2
0.4
0.6
0.8
1.0
Tree diversity (H´)P
ropo
rtio
n f f
low
er−
visi
ting
Em
pido
idea
spe
cies
70 75 80 85 90 95
10
20
30
40
Canopy cover (percent)
Em
pido
idea
spe
cies
ric
hnes
s
Oecologia
1 3
of Phoridae did not respond to herb diversity. this was sur-prising, as a positive association between plant and insect diversity is predicted as a bottom-up effect of resource heterogeneity and has frequently been reported in observa-tional as well as experimental studies (murdoch et al. 1972; haddad et al. 2009; Woodcock and Pywell 2009; scher-ber et al. 2010). however, most studies that show positive effects of plant diversity on insect diversity were conducted in grassland ecosystems. Forest ecosystems display differ-ent and more complex habitat characteristics compared to grasslands, conditions under which plant diversity might contribute less to insect diversity patterns. In addition, her-bivorous taxa generally react most strongly to changes in plant diversity, whereas plant diversity effects get weaker with increasing trophic level and proportion of omnivores (haddad et al. 2009; scherber et al. 2010). as a conse-quence, taxa that encompass few herbivores, such as many Diptera families, may be less affected. the few studies which specifically analysed the effect of herb layer diver-sity on insect richness yielded mixed results: in agreement with our findings, parasitoid wasp richness (sperber et al. 2004; Fraser et al. 2007) and carabid richness (humphrey
et al. 1999) were not affected by herb layer diversity. how-ever, Woodcock et al. (2003) found higher Dipteran species richness with increasing plant species richness in the herb layer.
effects of tree diversity
tree diversity had weak and inconclusive effects on Dip-teran communities in our study. structural equation mod-elling showed that tree diversity effects were mediated via changes in beech abundance, the dominant tree species in the forests analysed in this study. note that an alternative path (from beech abundance to tree species richness) was not supported, indicating that tree species richness was driving beech abundance (and not the other way around). the weak or non-existent effect of tree diversity is not sur-prising, given that (1) tree canopies are distant from the locations at which flies were sampled, and (2) tree diversity may act on a larger spatial scale than herb diversity. Other studies comparable to ours also showed mixed results: while tree diversity enhanced the abundances of beetles, true bugs, cavity-nesting bees, ants, and parasitoid wasps
Fig. 3 species-level responses to a herb diversity (H′), b herb cover (%), c canopy cover (%), d herb height (cm) and e soil moisture (%). Curves show predictions from a single minimal adequate multinomial model on the 50 most abundant fly species from the families empi-
doidea and Phoridae. Met gal Metopina galeata haliday (Phoridae), Met bra Metopina braueri strobl (Phoridae), Chr ang Chrysotus angulicornis Kowarz (empidoidea), Sci pla Sciapus platypterus Fab-ricius (empidoidea), Lep fla Leptopeza flavipes meigen (empidoidea)
Oecologia
1 3
(riihimaki et al. 2005; Fraser et al. 2007; Kaitaniemi et al. 2007; sobek et al. 2009a, b, c), spider abundances remained unaffected (riihimaki et al. 2005; schuldt et al. 2008).
Our results do not support our hypothesis that Dipteran richness is positively associated with tree diversity due to a greater heterogeneity of structures and resources in diverse forest stands. Other studies have found positive tree diver-sity effects on different taxa. For example, richness of parasitoid wasp taxa increased with increasing tree species richness in agroforestry systems (sperber et al. 2004) and temperate woodlands (Fraser et al. 2007). similarly, spe-cies richness of beetles was higher in stands with higher tree diversity in a near-natural forest (sobek et al. 2009b), while diversity of spiders, true bugs and cavity-nesting bees was not affected by tree diversity (schuldt et al. 2008; sobek et al. 2009a, c). In general, responses of arthropod communities to variations in tree diversity appear to be mixed and taxon dependent. however, it must be noted that in our study, Diptera were sampled in the herb layer only, while previous studies that detected significant effects of tree diversity on insect communities often sampled in the
canopy (e.g. sobek et al. 2009c). Certainly, by employing traps in the canopy, different conclusions about the impor-tance of tree diversity might have been drawn. however, we considered it worthwhile to include tree diversity as an explanatory variable since the tree layer has the potential to modify herb layer characteristics and thus indirectly insect communities in the understorey of forests (Vockenhuber et al. 2011, 2013).
effects of canopy, vegetation and flower cover
From all explanatory variables tested by us, tree canopy cover was among the most important predictors; a dense canopy had mostly negative effects on Diptera abundance, family richness, empidoidea richness, and proportions of flower visitors. a high canopy cover reduces herb cover and herb diversity, which negatively affects abundance and diversity of Diptera. similarly, many of the taxa studied by us positively respond to sunlight for mating or hunting, which may again explain the negative effects of dense tree canopies.
Fig. 4 structural equation model showing that Diptera species rich-ness (families, empidoidea and Phoridae) is driven by tree species richness, percent beech, canopy cover and herb layer characteristics. Diptera, Soil and Herbs are latent variables with from two to three indicator variables each. error terms are indicated by small empty cir-
cles. significant causal paths are indicated by bold arrows. Numbers next to arrows are standardized path coefficients (i.e. correlations). For unstandardized coefficients and more detailed output, see Online resources 10–12
Oecologia
1 3
In contrast to canopy cover, herb layer cover (and partly also flower cover) had generally positive effects on fam-ily richness, Phorid abundance and richness and propor-tions of flower visitors. these results support our initial hypothesis that herb cover is positively associated with Dipteran richness and abundance due to increased food resource availability and a greater available vegetation sur-face. On the whole, vegetation cover appears to be more important for Dipteran community characteristics than the plant diversity of the herb or tree layer. Previous stud-ies also emphasised the greater importance of plant bio-mass/cover compared to plant diversity in shaping insect species richness or abundance (Koricheva et al. 2000; haddad et al. 2001; Ober and hayes 2008; Woodcock and Pywell 2009). according to the more individuals hypoth-esis (Preston 1962; srivastava and lawton 1998), more productive sites harbour higher insect species richness, since productive sites support bigger populations of each species, making even rare species less extinction prone. accordingly, a greater vegetation cover has the potential to support higher insect numbers by providing a greater resource base for herbivores, which can subsequently lead to higher abundances of predators and parasitoids. In addi-tion, insect species richness as well as abundance may be enhanced if a greater vegetation cover is linked to a higher vegetation surface area or structural complexity. similar to our results, syrphid richness in forests has been shown to increase with vegetation cover (humphrey et al. 1999). In addition, Bährmann (1984) pointed out that a pronounced vegetation cover (together with the presence of decaying plant material) can enhance the abundances of certain Dip-teran taxa in grasslands.
Previous studies mentioned a preference of empidoidea for habitats with well-developed herb layers (Pollet and grootaert 1987; meyer 2009), with other environmental factors such as humidity/soil moisture, light intensity, and the availability of dead wood also contributing to diversity patterns of empidoidea (Pollet and grootaert 1987; meyer 2005; stark 2008). Phorid abundance was enhanced by a higher vegetation cover. Commonly, phorids engage in leaf-searching behaviour when looking for traces of hon-eydew, moisture, dead insects, or plant sap, thus larger leaf surfaces concomitant with higher vegetation cover may prove beneficial. Durska (2006) found that phorid commu-nities were influenced by the physical structure of the habi-tat, but he also pointed out the importance of microclimatic factors, in particular the light regime.
effects of other environmental variables
to our surprise, abiotic habitat characteristics played a minor role in our study; the depth of the humus layer was almost the only abiotic variable that consistently (and
negatively) influenced Dipteran communities. now critics might argue that we simply did not include enough envi-ronmental covariates; however, additional analyses using model-based boosting, essentially based on machine-learning algorithms, revealed that biotic parameters were always more important than abiotic ones. In these models, spatial coordinates, canopy and herb cover were selected in ca. 50 % of cases, while other variables, such as soil ph or humus layer mass were selected only in 0–10 % of cases. thus, environmental variables were overall of minor importance when compared with biotic parameters; this should be considered in future studies comparable to ours.
response of individual Diptera species
Our multinomial models showed that some species responded positively to herb diversity, while others even showed strong negative responses to herb diversity. two species showed a clear preference for forest interiors (M. galeata and L. flavipes), while three showed abundance maxima at the forest edge (M. braueri, C. angulicornis and S. platypterus). Indeed, C. angulicornis has been reported mainly from grassland and arable land. L. flavipes is a sil-vicolous species whose larvae inhabit deadwood. surpris-ingly, most species positively responded to canopy cover, which had not been visible from an inspection of Diptera species richness alone. these results show that species-level responses may differ from community-level attributes such as Diptera abundance or species richness. modelling the response of whole insect communities as a matrix of multinomial response variables may be more enlightening than other multivariate techniques, because exact predic-tions for each taxon can be made.
trapping methodology
like almost every method employed to trap insects at a given location, pan trap sampling has its strengths and weaknesses. One potential weakness is that flying insects are attracted to the colour of the pan trap, so that the result-ing data are not an unbiased sample of the population. a separate analysis of our data, separated by pan trap col-ours, showed that suction samples and pan trap sam-ples were highly significantly related (t = 2.81, df = 79, P = 0.0 063), and all trap colours (white, yellow, blue) dif-fered only in the mean amount of individuals captured, but not in species richness. In addition, the variance in overall community composition (sensu legendre and De Cáceres 2013) was 0.31–0.33 in all cases (based on Bray–Curtis distances). these analyses show that pan trap sampling and suction sampling yielded very similar results, and our data can indeed be used to study the presence of individual taxa in response to the environment.
Oecologia
1 3
Observation vs. experimentation
Because our study is largely observational, it is not pos-sible to infer strong causality. however, we used a gradi-ent in tree species richness that was selected a priori, and in that sense our study differs from purely observational approaches. using such an a priori gradient has proven to be an appropriate approach to biodiversity studies in natural ecosystems (Kahmen et al. 2005; unsicker et al. 2006). While experimental forest plantations can help reduce the influence of confounding factors (scherer-lor-enzen et al. 2007), these comparatively young plantations would need long periods of time to resemble the equilib-rium conditions of old growth, natural forests (leuschner et al. 2009).
another possible caveat of our study is that the tree diversity gradient actually represents a dilution gradient of the species F. sylvatica, which is present in different pro-portions on nearly all plots (nadrowski et al. 2010). thus, plots with low tree diversity are always almost pure stands of F. sylvatica, with the typical conditions created by this species, such as low light availability, reduced soil ph, low vegetation cover, and a thick, slowly decaying litter layer (ellenberg and leuschner 2010). as a consequence, it is impossible to completely disentangle the species identity effect of F. sylvatica on Dipteran communities in the herb layer from the effect of tree diversity. Future studies in for-ests differing in tree diversity, but not dominated by F. syl-vatica, could help to test the generality of our results.
Our structural equation modelling approaches allowed us to impose what has been termed ‘statistical control’ on our data (grace 2006). this allowed us to show that most effects are indeed driven by herb layer characteristics that are in turn indirectly affected by tree species richness and canopy properties (getzin et al. 2012).
Conclusion
We have shown that herb layer characteristics and canopy cover are related to Dipteran community characteristics in germany’s largest connected deciduous forest. While plant diversity effects were not as prominent as in several grassland studies, Dipteran abundance did respond, albeit moderately, to the diversity of the herb layer, and especially empidoidea showed a strong increase in stands character-ized by a more species-rich herb layer. Individual species showed strong and opposing effects to environmental gra-dients, which may have resulted in weaker patterns at the community level.
Overall, patterns of Dipteran abundance and richness may be better explained by tree canopy and herb cover than by tree or herb diversity per se. the linkages between
herb layer characteristics and community attributes of a large and ecologically important taxon such as the Diptera indicate that we are still just beginning to understand bio-diversity in complex ecosystems such as natural forests. the types of analyses and models employed by us, such as structural equation modelling, combined with multi-nomial models for whole communities, will help us to understand networks of interacting entities across taxa and ecosystems.
Acknowledgments this study was funded by the german research Foundation (DFg) within the framework of the research training group graduiertenkolleg 1086. We thank s. Prescher and g. Weber for the identification of Phoridae. t. Wommelsdorf provided valuable data on tree layer composition. D. seidel provided data on gPs coor-dinates of the plots. We are grateful to V. eißfeller and other members of the research training group for field assistance, to m. gollnow and e. Dyson for assisting with the sorting of insect samples, and to a. Prinzing, C. normann, g. everwand and P. Kabouw for helpful comments on a previous version of this manuscript. the experiments comply with the laws of germany, where they have been conducted.
References
allgood DW, miller Da, Kalcounis-rueppell mC (2009) Influence of intensive pine management on Dipteran community structure in coastal north Carolina. environ entomol 38:657–666
Bais hP, Weir tl, Perry lg, gilroy s, Vivanco Jm (2006) the role of root exudates in rhizosphere interactions with plants and other organisms. ann rev Plant Biol 57:233–266
Bährmann r (1984) Die Zweiflügler (Diptera Brachycera) der Kraut- und strauchschicht des leutratales bei Jena/thüringen—ein öko-faunistischer Vergleich. Zool Jb syst 111:175–217
Baldi a (2003) using higher taxa as surrogates of species richness: a study based on 3700 Coleoptera, Diptera, and acari species in Central-hungarian reserves. Basic appl ecol 4:589–593
Barbier s, gosselin F, Balandier P (2008) Influence of tree species on understory vegetation diversity and mechanisms involved—a critical review for temperate and boreal forests. For ecol manage 254:1–15
Basset y, Cizek l, Cuénoud P, Didham rK, guilhaumon F, missa O, novotny V et al (2012) arthropod diversity in a tropical forest. science 338:1481–1484
Bivand rs, Pebesma e, gómez-rubio V (2013) applied spatial data analysis with r, 2nd edn. springer, heidelberg
Blunch nJ (2013) Introduction to structural equation modelling using IBm sPss statistics and amOs. sage, london
Brose u (2003) Bottom-up control of carabid beetle communities in early successional wetlands: mediated by vegetation structure or plant diversity? Oecologia 135:407–413
Chvála m (1983) the empidoidea (Diptera) of Fennoscandia and Denmark. II. general part. the families hybotidae, atelestidae and microphoridae. Fauna entomol scand 12:1–279
Crawley mJ (2013) the r book, 2nd edn. Wiley, ChichesterCumming Jm (1994) sexual selection and the evolution of dance fly
mating systems. Can entomol 126:907–920Disney rhl (1994) scuttle flies: the Phoridae. Chapman and hall,
londonDurska e (2006) Diversity of the scuttle fly (Diptera: Phoridae) com-
munities in the plantations of moist pine forests of the Bialowieza Primeval Forest and the tuchola Forest (Poland). Biodivers Con-serv 15:385–393
Oecologia
1 3
ellenberg h, leuschner C (2010) Vegetation mitteleuropas mit den alpen in ökologischer, dynamischer und historischer sicht, 6th edn. ulmer, stuttgart
Fahrmeir l, Kneib t, lang s (2009) regression. springer, heidelbergFraser sem, Dytham C, mayhew PJ (2007) Determinants of parasi-
toid abundance and diversity in woodland habitats. J appl ecol 44:352–361
getzin s, Wiegand K, schoning I (2012) assessing biodiversity in forests using very high-resolution images and unmanned aerial vehicles. methods ecol evol 3:397–404
gilliam Fs (2007) the ecological significance of the herbaceous layer in temperate forest ecosystems. Bioscience 57:845–858
grace JB (2006) structural equation modeling and natural systems. Cambridge university Press, Cambridge
grace JB, anderson tm, Olff h, scheiner sm (2010) On the specifi-cation of structural equation models for ecological systems. ecol monogr 80:67–87
haddad nm, tilman D, haarstad J, ritchie m, Knops Jmh (2001) Contrasting effects of plant richness and composition on insect communities: a field experiment. am nat 158:17–35
haddad nm, Crutsinger gm, gross K, haarstad J, Knops Jmh, til-man D (2009) Plant species loss decreases arthropod diversity and shifts trophic structure. ecol lett 12:1029–1039
hastie t, tibshirani r, Friedman J (2009) the elements of statistical learning, 2nd edn. springer, heidelberg
hättenschwiler s, gasser P (2005) soil animals alter plant litter diver-sity effects on decomposition. Pnas 102:1519–1524
hirao t, murakami m, Kashizaki a (2009) Importance of the under-story stratum to entomofaunal diversity in a temperate deciduous forest. ecol res 24:263–272
hudewenz a, Klein am, scherber C, stanke l, tscharntke t, Vogel a, Weigelt a, Weisser WW, ebeling a (2012) herbivore and pollina-tor responses to grassland management intensity along experimen-tal changes in plant species richness. Biol Conserv 150:42–52
hughes JB, Daily gC, ehrlich Pr (2000) Conservation of insect diversity: a habitat approach. Conserv Biol 14:1788–1797
humphrey J, hawes C, Peace aJ, Ferris-Kaan r, Jukes mr (1999) relationships between insect diversity and habitat characteristics in plantation forests. For ecol manage 113:11–21
hunter mD, Price PW (1992) Playing chutes and ladders: heteroge-neity and the relative roles of bottom-up and top-down forces in natural communities. ecology 73:724–732
Jost l (2007) Partitioning diversity into independent alpha and beta components. ecology 88:2427–2439
Kahmen a, Perner J, audorff V, Weisser W, Buchmann n (2005) effects of plant diversity, community composition and environ-mental parameters on productivity in montane european grass-lands. Oecologia 142:606–615
Kaitaniemi P, riihimaki J, Koricheva J, Vehvilainen h (2007) experi-mental evidence for associational resistance against the european pine sawfly in mixed tree stands. silva Fenn 41:259–268
Knops Jmh et al (1999) effects of plant species richness on invasion dynamics, disease outbreaks, insect abundances and diversity. ecol lett 2:286–293
Koricheva J, mulder CPh, schmid B, Joshi J, huss-Danell K (2000) numerical responses of different trophic groups of invertebrates to manipulations of plant diversity in grasslands. Oecologia 125:271–282
lawton Jh (1983) Plant architecture and the diversity of phytopha-gous insects. annu rev entomol 28:23–39
legendre P, De Cáceres m (2013) Beta diversity as the variance of community data: dissimilarity coefficients and partitioning. ecol lett 16:951–963
leuschner C, Jungkunst hF, Fleck s (2009) Functional role of forest diversity: pros and cons of synthetic stands and across-site com-parisons in established forests. Basic appl ecol 10:1–9
mcCulloch Ce, searle sr (2001) generalized, linear, and mixed models. Wiley, new york
meyer h (2005) langbein-, tanz- und schwebfliegen (Diptera: empi-doidea: Dolichopodidae, empididae, hybotidae; syrphidae) im totholz von laubwäldern schleswig-holsteins. Faun Ökol mitt 8:363–382
meyer h (2009) Bestandsaufnahme und Bibliographie der langbein-, tanz- und rennraubfliegen schleswig-holsteins mit angaben zur Ökologie ausgewählter arten (Diptera, empidoidea: Dolichopo-didae s. l., Brachystomatidae, empididae, hybotidae). Faun Ökol mitt 9:17–60
mölder I (2009) Diversity and tree neighborhood effects on the growth dynamics of european beech and the stand seed bank in temperate broad-leaved forests of variable tree diversity. PhD dis-sertation, university of göttingen, göttingen
murdoch WW, Peterson Ch, evans FC (1972) Diversity and pattern in plants and insects. ecology 53:819–829
nadrowski K, Wirth C, scherer-lorenzen m (2010) Is forest diversity driving ecosystem function and service? Curr Opin environ sus-tain 2:75–79
naeem s, Duffy Je, Zavaleta e (2012) the functions of biological diversity in an age of extinction. science 336:1401–1406
Ober hK, hayes JP (2008) Influence of forest riparian vegetation on abundance and biomass of nocturnal flying insects. For ecol manage 256:1124–1132
Oosterbroek P (2007) the european families of the Diptera. KnnV, utrecht
Pape t, Bickel D, meier r (eds) (2009) Diptera diversity: status, chal-lenges and tools. Brill, leiden
Pollet m, grootaert P (1987) ecological data on Dolichopodidae (Diptera) from a woodland ecosystem. I. Colour preference, detailed distribution and comparison of different sampling tech-niques. Bull Inst r sci nat Belg entomol 57:173–186
Preston FW (1962) the canonical distribution of commonness and rarity. ecology 43:185–215
r Development Core team (2013) r: a language and environment for statistical computing. r Foundation for statistical Computing, Vienna, http://www.r-project.org
riihimaki J, Kaitaniemi P, Koricheva J, Vehvilainen h (2005) testing the enemies hypothesis in forest stands: the important role of tree species composition. Oecologia 142:90–97
ripley BD (2013) Feed-forward neural networks and multinomial log-linear models. nnet package, version 7.3-6, url http://www.stats.ox.ac.uk/pub/mass4/
savage J (2002) Cleaning up the world: Dipteran decomposers. Bio-diversity 3:12–15
scherber C et al (2010) Bottom-up effects of plant diversity on multitrophic interactions in a biodiversity experiment. nature 468:553–556
scherer-lorenzen m, schulze eD, Don a, schumacher J, Weller e (2007) exploring the functional significance of forest diver-sity: a new long-term experiment with temperate tree species (BIOtree). Perspect Plant ecol evol syst 9:53–70
schuldt a, Fahrenholz n, Brauns m, migge-Kleian s, Platner C, schaefer m (2008) Communities of ground-living spiders in deciduous forests: does tree species diversity matter? Biodivers Conserv 17:1267–1284
siemann e, tilman D, haarstad J, ritchie m (1998) experimental tests of the dependence of arthropod diversity on plant diversity. am nat 152:738–750
sobek s, goßner mm, scherber C, steffan-Dewenter I, tscharntke t (2009a) tree diversity drives abundance and spatiotemporal β-diversity of true bugs (heteroptera). ecol entomol 34:772–782
sobek s, steffan-Dewenter I, scherber C, tscharntke t (2009b) spa-tiotemporal changes of beetle communities across a tree diversity gradient. Divers Distrib 15:660–670
sobek s, tscharntke t, scherber C, schiele s, steffan-Dewenter I (2009c) Canopy vs. understory: does tree diversity affect bee and wasp communities and their natural enemies across forest strata? For ecol manage 258:609–615
sperber C, nakayama K, Valverde m, neves F (2004) tree species richness and density affect parasitoid diversity in cacao agrofor-estry. Basic appl ecol 5:241–251
srivastava Ds, lawton Jh (1998) Why more productive sites have more species: an experimental test of theory using tree-hole com-munities. am nat 152:510–529
stark a (2008) species of the genus Oedalea meigen, 1820 (Diptera: hybotidae): an element of the canopy fauna in european for-ests? In: Floren a, schmidl J (eds) Canopy arthropod research in europe. Bioform entomology, nürnberg, pp 105–117
strong Dr, lawton Jh, southwood tre (1984) Insects on plants: community patterns and mechanisms. Blackwell scientific, Oxford
thompson FC (2009) nearctic Diptera: twenty years later. In: Pape t, Bickel D, meier r (eds) Diptera diversity: status, challenges and tools. Brill, leiden, pp 3–46
unsicker sB, Baer n, Kahmen a, Wagner m, Buchmann n, Weis-ser WW (2006) Invertebrate herbivory along a gradient of plant
species diversity in extensively managed grasslands. Oecologia 150:233–246
Vockenhuber ea, scherber C, langenbruch C, meißner m, seidel D, tscharntke t (2011) tree diversity and environmental con-text predict herb species richness and cover in germany’s larg-est connected deciduous forest. Perspect Plant ecol evol syst 13:111–119
Vockenhuber e, Kabouw P, tscharntke t, scherber C (2013) Plant–animal interactions in two forest herbs along a tree and herb diversity gradient. Plant ecol Divers 6:205–216
Wood sn (2006) generalized additive models. Chapman & hall/CrC, london & new york
Wood sn (2012) Package mgcv (version 1.7-22). mixed gam com-putation vehicle with gCV/aIC/reml smoothness estimation. url: http://www.cran.r-project.org/web/packages/mgcv/mgcv.pdf. accessed 22 July 2013
Woodcock Ba, Pywell rF (2009) effects of vegetation structure and floristic diversity on detritivore, herbivore and predatory inverte-brates within calcareous grasslands. Biodivers Conserv 19:81–95
Woodcock Ba, Watt aD, leather sr (2003) Influence of manage-ment type on Diptera communities of coniferous plantations and deciduous woodlands. agric ecosyst environ 95:443–452
