Resolving the predator first paradox: Arthropod predator food webs in pioneer sites of glacier forelands

Abstract Primary succession on bare ground surrounded by intact ecosystems is, during its first stages, characterized by predator‐dominated arthropod communities. However, little is known on what prey sustains these predators at the start of succession and which factors drive the structure of these food webs. As prey availability can be extremely patchy and episodic in pioneer stages, trophic networks might be highly variable. Moreover, the importance of allochthonous versus autochthonous food sources for these pioneer predators is mostly unknown. To answer these questions, the gut content of 1,832 arthropod predators, including four species of carabid beetles, two lycosid and several linyphiid spider species caught in early and late pioneer stages of three glacier forelands, was screened molecularly to track intraguild and extraguild trophic interactions among all major prey groups occurring in these systems. Two‐thirds of the 2,310 identified food detections were collembolans and intraguild prey, while one‐third were allochthonous flying insects. Predator identity and not successional stage or valley had by far the strongest impact on the trophic interaction patterns. Still, the variability of prey spectra increased significantly from early to late pioneer stage, as did the niche width of the predators. As such the structure of pioneer arthropod food webs in recently deglaciated Alpine habitats seems to be driven foremost by predator identity while site and early successional effects contribute to a lesser extent to food web variability. Our findings also suggest that in these pioneer sites, predatory arthropods depend less on allochthonous aeolian prey but are mainly sustained by prey of local production.


S1: Environmental conditions
Figure S1: Environmental conditions in the three glacier forelands. Air temperature [°C] and relative humidity [%] were recorded in all three valleys (Rotmoostal, Gaisbergtal, Langtal) in the space between the two sampling sites. In Rotmoostal an additional data logger (Rotmoostal -L2) was operated recording air temperature [°C], precipitation [mm/15min], solar radiation [kW/m²] and wind speed [m/s]. Please note that data has not been validated and cleaned but is displayed as recorded (e.g. radiation error of one temperature logger during the first few days or partial malfunction of the humidity sensor in Rotmoostal following the precipitation on 18 th July 2010). Table S3: Number of caught (A) and analysed (B) individuals per taxon from the early and late pioneer stage (PS) respectively in each of the three investigated glacier forelands. Note: As the two lycosid spiders could not be identified in the field upon collection their catchnumbers were reconstructed from the ratio found in the analysed samples from each site.

S6: Prey detection rates for the predators from 3 valleys and 2 pioneer stages
Data

S7: Community composition of predators and prey
Data table Supplementary_Data_S7.xlsx: Community composition data of predators and prey. Predator data is given as average catch per live pitfall trap. Data of flying insects and collembolans is reported as total catch over the 2 weeks sampling period with Malaise trap and grey bowl (flying insects) and permanent pitfall traps (collembolans) respectively.

Assessment of the predator community:
Catch data from the pitfall grids described in the main article was used to describe the predator community.

Assessment of the prey community:
Flying insect prey was assessed by pooled catch totals from one Malaise trap and one grey bowl (23x23 cm) operated from 12 th to 23 rd July 2010 at each of the six sites. A saturated salt solution with a drop of odourless detergent was used as preservation liquid and all traps were emptied daily. Dipterans were determined to families, other insects are given as orders, except for the aphid Cinara sp.. Please note that numbers of Cinara sp. available to the predators are likely underestimated in the 'flying insect data' as these aphids are no regular member of the glacier foreland community but inhabit pine forests. The aphids were blown by the wind to the glacier forelands and provide thus a classical example of random Aeolian dropout. This means that when they were present, they were found in high numbers more or less immobilized on the ground. Collembola data stem from sets of each 5 permanent pitfall traps (50% ethylene glycol) surrounding the pitfall grids used for live catches. Permanent pitfall traps were emptied once at the end of the 2 weeks period. Collembolans were not determined further, but catches were dominated by surface-active Entomobryiidae.
Predator and prey communities from the six sites were compared by principal components analysis (PCA) using correlations of square root transformed catch counts in Canoco 5 (ter Braak & Šmilauer 2012).

Results of predator and prey community analysis (Figure S7):
For the predators, axis 1 separated Langtal from the other two valleys, mostly due to differences in the two Pardosa species and Nebria germari, while the much weaker axis 2 showed changes from the early to the late pioneer stage mostly caused by the near absence of N. rufescens in the early stage. Among the prey the differences between early and late pioneer stage were more important than valley differences and appeared on axis 1. Nearly all flying insect groups increased in the late pioneer stage, only the collembolans declined markedly. Valley differences, mostly of Gaisbergtal, were associated with a few groups only, such as the sporadically, but then massively occurring aphid Cinara sp., flying beetles, and three brachyceran families. Figure S7: Principal components analysis (PCA) of predator and prey communities from the early (yellow) and late (green) pioneer stages of the three valleys Gaisbergtal (G), Rotmoostal (R), and Langtal (L). Fractions of explained variance are given with the axis labels. Hodkinson et al. (2001) investigated the densities of chironomids in an Arctic glacier foreland by placing a series of water filled traps at the ground. To enable a comparison between those densities and the ones present in Alpine glacier forelands (this study), we considered here only catches from the grey bowls but not from the malaise traps. Numbers from both studies were normalized on caught chironomids per day and 500 cm² water surface. As no precise numbers are stated in Hodkinson et al. (2001), they were estimated from the presented figures.