Combining DNA gut content analysis with Structural Equation Modelling (SEM) based on arthropod activity density at a Greenland glacier foreland give us the opportunity to combine qualitative data from the metabarcoding technique with an analysis technique based on quantitative data. SEM shows bottom-up and top-down controlled food chains as bottom-up control was important for spider and harvestman populations while top-down control was important for ground beetle populations. These mechanisms are closely related to the hunting strategies of the predators as bottom-up mechanisms are connected to a sit-and-wait behavior while top-down mechanisms are related to active-search behavior. Bottom-up controlled population developments were important in the early phase of the vegetation development while top-down prevailed in the later phase of the vegetation development away from the glacier snout. The shift from bottom-up to top-down cascades between arthropod predators and their potential prey populations was mainly driven by increasing temperatures away from the glacier. A consequence of the strong top-down cascades in the later phase of the succession is high rates of intra-guild predation (IGP) among all arthropod predators. Particularly in the guts of the linyphiid spider, Collinsia holmgreni Thorell 1871, trophic linkages to other linyphiid and lycosid spiders were detected. The IGP ratio of C. holmgreni was negatively correlated with the activity-density of available ground-living prey. Probably as a consequence of the high IGP among the linyphiid spiders, cold-adapted linyphiid species like C. holmgreni decreased in numbers downhill and became extinct in the warmer climax vegetation, where lycosid spiders dominated.

Data was collected using wet pitfall traps set up across various patches in the glacier foreland areas during the summers of 2015 and 2016. Soil samples were also taken for environmental characterization. Captured arthropods were identified to species or family level. Soil samples were analyzed for water and organic matter content. The activity density of arthropods was measured based on pitfall trap catches, and environmental variables were correlated with arthropod population data. NDVI measurements were used to assess vegetation biomass.

Qassinnguit Dates and Sampling Information:

Soil samples were taken at the Qassinnguit glacier foreland on August 7, 2015, and August 5, 2016, to measure the environmental variables of soil water and organic matter during dry weather periods.

NDVI scanning was performed on August 1, 2016.

Wet pitfall traps for sampling arthropods were established on July 8, 2015, with continuous sampling until August 7, 2015, when sampling was postponed until August 12 due to rainy weather. From then (August 12), sampling was done until August 20, 2015.

Wet pitfall traps were established on July 6, 2016, with continuous sampling until August 5, 2016, when sampling was postponed until August 10 due to rainy weather. From then (August 10), sampling was done until August 19, 2016.

Data Transformations

IGP-ratios were arcsine and square root transformed. Distance to glacier snout and time since deglaciation were log-transformed. Arthropod activity densities were log-transformed. NDVI values were used untransformed.