**3. Results**

#### *3.1. E*ff*ect of Di*ff*erent Habitat Types and Season on Honey Bees*

Based on the analysis, we found that habitat type and season a ffected the forager abundance and colony weight of bees (*p* < 0.05), although the patterns were di fferent between species and region. In general, habitat types showed to be the most important factor that a ffected both forager abundance and colony weight of bees compared to season and observation time (Table 2). E ffect of habitat type on forager abundance and colony weight of *A. cerana* di ffered between Malang and Bogor. In Malang, the di fference of habitat type significantly influenced the forager abundance and colony weight of *A. cerana,* which were found higher in home gardens and forests compared to the agricultural areas (Figures 2a and 3a). In Bogor, forager abundance of *A. cerana* was not significantly di fferent among habitat type (Figure 2b), while the colony weight of *A. cerana* in forests was significantly higher than in home gardens (Figure 3b). The same pattern was observed with *A. cerana* in Malang for forager abundance of *A. mellifera* in Malang and *T. laeviceps* in Bogor, which were also found higher in home gardens and forests compared to agricultural areas (Figure 2c,d and Figure 3c,d).

**Table 2.** The results of Kruskal–Wallis tests of forager abundance and colony weight of bees in different seasons, habitat types, and observation times.


**Figure 2.** Forager abundance of honey bees and stingless bees in different habitats. (**a**) *A. cerana* in Malang, (**b**) *A. cerana* in Bogor, (**c**) *A. mellifera* in Malang, and (**d**) *T. laeviceps* in Bogor. Boxes with different letters are significantly different at *p* < 0.05 according to Fisher's least-significant difference.

**Figure 3.** Colony weight of honey bees and stingless bees in different habitats. (**a**) *A. cerana* in Malang, (**b**) *A. cerana* in Bogor, (**c**) *A. mellifera* in Malang, and (**d**) *T. laeviceps* in Bogor. Box with different letters are significantly different at *p* < 0.05 according to Fisher's least significant difference.

Based on non-parametric regression, we found that the forager abundance of *A. cerana* in Malang was prone to decrease with increasing observation time (*p* = 0.001) (Table 3). In addition, the colony weights of *A. cerana* and *T. laeviceps* in Bogor as well as *A. mellifera* in Malang tended to increase with increasing observation time (*p* < 0.05).


**Table 3.** Relationship between forager abundance and colony weight of bees and observation time (day) based on non-parametric regression.

#### *3.2. Detection Results of Insecticide Residue in Honey and Body of Honey Bees*

Based on residue analysis, we barely detected insecticide residue (imidacloprid content) both in the honey and the body of bees (Table 4). Two of 18 honey samples (11.11%) contained a small amount of insecticide residue (<5 μg/kg). Insecticide residues were also detected in three bee samples (16.66%). Surprisingly, we did not detect the insecticide residue in the agriculture habitat, ye<sup>t</sup> it was detected in forest and home garden habitats.

