**3. Results**

#### *3.1. Sentinel Prey*

Overall, 72.13% (1731/2400) of the dummy caterpillars were attacked. On the basis of the identification of attack marks, 87.52% of the predators were ants, 7.93% beetles, 5.21% reptiles, 3.40% wasps, 3.17% birds, 1.25% mammals, and 0.79% were bees. The e ffect of landscape and local variables was not unequivocal for the two ecosystems. Fragment size, transect position, and their interaction had the highest relative importance for insect predation in forest-steppes (Table 1). We detected higher predation rates in edges in the central transects of small forest-steppes; however, we detected the opposite pattern in large fragments (Figure 2A). We found no interaction e ffect between fragment size and transect position in kurgans (Table 1, Figure 2B). Variables that best explained insect predation in kurgans were connectivity, transect position, and their interaction according to relative importance values (Table 1). Increasing connectivity had a negative e ffect on the predation rates of edges but not in kurgan centres (Figure 2D).

**Table 1.** Summary table for GLMM results after multimodel averaging of the best candidate models showing relative importance of each explanatory variable (fragment size: large (L) vs. small (S), transect position: centre (C) vs. edge (E), and connectivity), and their interactions on insect and seed predation rates in forest-steppes (FS) and kurgans (KU).


a Models fitted with binomial distribution (number of candidate models, ΔAIC < 6). b Each variable's importance within bestcandidatemodels(ΔAIC<6).c Significancelevels: \*: <0.05,\*\*:<0.01,\*\*\*:<0.001.

*Forests* **2021**, *12*, 54

**Figure 2.** Insect predation. Interacting effect of (**A**) fragment size and edge effect on insect predation in forest-steppes; (**B**) fragment size and edge effect on insect predation in kurgans (mean ± SE); (**C**) connectivity and edge effect on insect predation in forest-steppes; and (**D**) connectivity and edge effect on insect predation in kurgans. Green, centre position; orange, edge position. Significance levels: \*: <0.05, \*\*: <0.01.

#### *3.2. Seed Predation*

In total, 77.58% (3724/4800) of the seeds had predation marks or been carried away. Similar to the results of insect predation, we found a significant interaction between fragment size and edge effect in case of forest-steppes, but not in kurgans (Table 1, Figure 3A,B). We found higher seed predation rates in edge in centre transects in forest-steppes, and this difference was more pronounced in small than in large fragments (Figure 3A). Connectivity and transect position did not affect seed predation in forest-steppes (Table 1, Figure 3C); however, seed predation in kurgans was affected by connectivity, transect position, and their interaction (Table 1, Figure 3D). In kurgans, we found a negative effect of connectivity on seed predation rates in edge and positive effect in centre transects (Figure 3D).

*Forests* **2021**, *12*, 54

**Figure 3.** Seed predation. Interacting effect of (**A**) fragment size and edge effect on seed predation in forest-steppes; (**B**) fragment size and edge effect on seed predation in kurgans (mean ± SE); (**C**) connectivity and edge effect on seed predation in forest-steppes; and (**D**) connectivity and edge effect on seed predation in kurgans. Green, centre position; orange, edge position. Significance levels: \*\*: <0.01, \*\*\*: <0.001.
