Supporting dataset for elemental traits of plant-invertebrate food web components of oilseed rape fields

This dataset is provided in support of the paper "Edge effect imprint on elemental traits of plant-invertebrate food web components of oilseed rape fields" (Orłowski et al., 2019). Supplementary data are given on the following: (1) the full taxonomic list of invertebrates (n = 12 916) classified into food guilds and functional groups, which were sampled in 34 oilseed rape fields in SW Poland in spring 2015; (2) concentrations of 12 chemical elements measured in invertebrates; (3) the relationships between abundance and percentage (%) in the community of major invertebrate groups, and habitat variables; (4) the statistical tests comparing the concentrations of chemical elements between the different groupings of organisms; (5) the relationships between the elemental traits of oilseed rape plant samples and major functional invertebrate groupings or main taxonomic insect groups, and the habitat variables of oilseed rape fields.


b s t r a c t
This dataset is provided in support of the paper "Edge effect imprint on elemental traits of plant-invertebrate food web components of oilseed rape fields" (Orłowski et al., 2019). Supplementary data are given on the following: (1) the full taxonomic list of invertebrates (n ¼ 12 916) classified into food guilds and functional groups, which were sampled in 34 oilseed rape fields in SW Poland in spring 2015; (2) concentrations of 12 chemical elements measured in invertebrates; (3) the relationships between abundance and percentage (%) in the community of major invertebrate groups, and habitat variables; (4) the statistical tests comparing the concentrations of chemical elements between the different groupings of organisms; (5) the relationships between the elemental traits of oilseed rape plant samples and major functional

Data
The data presented here (Figs. 1 and 2; Tables 1e7) constitute the basis for the article by Orłowski et al. [1]. The dataset provides detailed information on: (1) the full taxonomic list of invertebrates (n ¼ 12 916) classified into food guilds and functional groups (Annex 1 in Ref [1]; Table 4), which were sampled in 34 oilseed rape fields in SW Poland in spring 2015 ( Fig. 1; Tables 1 and 2); (2) concentrations of 12 chemical elements measured in invertebrates (Table 3); (3) the relationships between abundance and percentage (%) in the community of major invertebrate groups, and habitat variables (Table 5); (4) the statistical tests comparing the concentrations of chemical elements between the different groupings of organisms (Table 6); (5) the relationships between the elemental traits of oilseed rape plant samples and major functional invertebrate groupings or main taxonomic insect groups, and the habitat variables of oilseed rape fields (Table 7).
The most numerous of the invertebrates, classified into six functional groups, were herbivores, which made up on average 39.4% (95% CI ¼ 34.9e43.9%) of all the specimens sampled from one field.
Specifications Table   Subject area Ecology, Biological Sciences, Biogeochemistry, Agroecology More specific subject area Biogeochemistry of invertebrates Type of data Tables, figures How data was acquired Through field work and laboratory work Data format Raw, filtered and analysed Experimental factors Investigation of chemical composition of insects and plant tissues, and variability in landcover.

Experimental features
Quantification of the abundance of invertebrates and measurements of the elemental composition (K, Na, Ca, Mg, Cu, Zn, Fe, Mn, As, Cd, Co and Pb) of 15 different organisms within the plant-invertebrate food web: plant e oilseed rape pests/herbivores e pollinators ¼ wild bees e saprovores e predators e parasitoids. These were then related to the individual field edge habitat features (including typically anthropogenic ones like dirt and tarred roads) measured within a 100 m radius around the invertebrate sampling sites. Data source location The dataset presented in this data paper were gathered in spring

Value of the data
The data on elemental traits of organisms relate to the individual field edge habitat features (including typically anthropogenic ones like dirt and tarred roads) measured within a 100 m radius around the invertebrate sampling sites. The data in this article demonstrate that the elemental traits of the plant-invertebrate food web components in oilseed rape crops varied owing to the habitat specificity determined at the relatively small spatial scale of an individual field, and that the elemental traits of these organisms differed from both an inter-and an intra-guild perspective. These data may be useful for explaining the sources of variation in both the quality of agricultural products (including food for human consumption) and the dietary flow of essential macronutrients and non-essential trace elements within plantinvertebrate food webs in agroecosystems.

Relationships between landscape and habitat variables, and abundance of invertebrates
There was a weak but insignificant relationship between the (log-transformed) area of a field and the number of invertebrates sampled there (r ¼ 0.155, P ¼ 0.406). On average, 6.2% more invertebrates were sampled in the large fields (i.e. area >13 ha; n ¼ 24) than in the small ones (i.e. area <8 ha; n ¼ 10): 387 (95% C.I. ¼ 334e440) vs. 363 (237e488), respectively. More pronounced differences emerged when the landscape context, i.e. the type of field surroundings, was taken into account. In an open landscape, 12% fewer invertebrates were sampled in the small fields than in the large ones ( Fig. 1 in Ref [1]). In a mosaic-like landscape, by contrast, 20% more invertebrates were sampled in the large fields than in the small ones ( Fig. 1 in Ref [1]). 18% fewer invertebrates were sampled in small fields with mosaic surroundings than in small fields with open surroundings (Fig. 1 in Ref [1]). None of these differences, however, were statistically significant (Mann-Whitney test, P ! 0.201), presumably because of the small sample size and the confounding influence of edge habitat variability (see below). Analysis of the eight edge habitat variables measured within a 100 m radius of the invertebrate sampling points (Table 1) in the 34 oilseed rape fields ( Fig. 1) with respect to the number or percentage (%) of six functional invertebrate groupings and the most numerous insect orders sampled there yielded only a few statistically significant relationships. But the P-value of none of them met the threshold for multiple comparisons (at P 0.0043; k ¼ 12). Specifically, we found that the number of all invertebrates sampled was positively correlated with the area of dirt roads (Pearson correlation Fig. 2. Examples of statistically significant relationships between the number of major invertebrate/insect groups sampled and the habitat variables of 34 oilseed rape fields; woodlot ¼ wood.       The only negative statistically significant relationship was between the percentage (%) of herbivorous insects and the area of woods (r ¼ À0.462, P ¼ 0.009).
The concentrations of Na, Ca, Mg, Cu, Zn, Fe, As, Co and Pb varied significantly between two oilseed rape pest taxa, M. aeneus and Ceutorhynchus spp., sampled on the same fields (t-test for dependent samples, in all cases, P 0.008).

The study area
The 35 winter oilseed rape fields were managed using conventional amounts of agrochemicals, including pesticides and fertilisers. The landowner (Top Farms Wielkopolska Co., Poland) supplied management data on agricultural practices in a few large fields for the study year; both the timing Table 5 Pearson correlation coefficients testing the relationships between abundance and percentage (%) in the community of major invertebrate groups, and habitat variables of 34 oilseed rape fields. The relationships that meet the FDR-adjusted P-value are shown in red font; black font e P 0.05; grey font e P > 0.05.     Table 7 Results of Spearman rank correlations (r s ) and associated p-values testing the relationship between the concentrations of 12 elements measured in oilseed rape plant samples (determined as the whole plant) and major functional invertebrate groupings or main taxonomic insect groups, and habitat variables of 34 oilseed rape fields in SW Poland in spring 2015. The relationships that meet the FDR-adjusted P-value are shown in red font; black font e P 0.05; grey font e P > 0.05.  and use of agrochemicals were similar in all the other fields. The fields were sown (winter oilseed rape cultivar: PRW 31 F-1) in the second half of September 2014. Each of three mineral fertilisers (Polifoska G, Saletrosan 26% N, ammonium sulphate 34% N) was used in a dose of 300 kg ha À1 ; foliar fertiliser (OSD Bor, 1.5 kg ha À1 ) and magnesium sulphate (3 kg ha À1 ). Also applied were herbicides (Butisan Star Max; 2.5 L ha À1 ); insecticides between March and May (Ammo Super, Decis, Alfacet, Mospilan; each 0.1e0.15 L ha À1 ) targeting herbivorous insects, mostly stem weevils Ceutorhynchus spp. and pollen beetles Meligethes spp.; and fungicides (Caryx, 0.6 L ha À1 ; Pictor; 0.5 L ha À1 ).

Chemical analysis
We used reference materials for each of the AAS measurements. These were blind tests, i.e. they contained the same chemical composition as a particular sample, but were devoid of the analysed biological material. The analytical procedure for preparing these samples was the same as in the case of our ones. We used standardised samples obtained from SGAB Analytica, Luleå Technical University, Luleå, Sweden and Fürst Medical Laboratory, Billingstad, Oslo, Norway, Certified Values and Uncertainty NCS ZC, i.e. standards of a particular quality for each kind of tissue and chemical element. The analytical measurement process was validated using reference materials, i.e. CVU (bovine liver, kidney, muscles, lung, bone) provided by SGAB Analytica, Luleå Technical University, Luleå, Sweden and Fürst Medical Laboratory, Billingstad, Oslo, Norway, Certified Values and Uncertainty NCS ZC. Reference values amounted to 0.25 ± 0.05e32.7 ± 1.8 for the target chemical elements. The average (±SD) values determined for the target elements (20 measurements in the invertebrate and plant samples) were 0.19 ± 0.05e34.8 ± 1.07. The precision of the method, understood as the degree of conformity between the results of multiple analyses performed on the same sample, was 5% (relative standard deviation, RSD).

Data treatment
Pre-analysis of our habitat variables quantified for the individual oilseed rape fields showed strong collinear associations (tested by Spearman's and Pearson's correlation coefficients) between some of these variables. Principal Component Analysis (PCA) was applied to reduce collinearity among them (see Table 2). But because PCA outputs identified PC-axes that clustered structurally distinct variables  (e.g. dirt and tarred roads or coverage of arable land and forest; Table 2), for which we wanted to assess their individual influence on particular invertebrate groups, the PCA-derived variables were of limited use in our subsequent analyses. Further, because the habitat variables were generally only loosely related to abundance data and the elemental traits of the studied organisms (and only single such relationships met the threshold of statistical significance), we assumed that the results of a univariate analysis (with P-values adjusted to multiple comparisons) would be justified, thus permitting a robust biological interpretation of our observations.