3. Results
We observed a total of 9476 birds belonging to 69 species (Supporting Information 1). The most abundant species were House Sparrow (Passer domesticus ), Spotless Starling (Sturnus unicolor ) and Feral Pigeon (36% of all observed birds; Supporting Information 1). Out of the 189 points, 100 points were found to be inhabited bynon-invaded communities, while 89 were inhabited byinvaded communities (Supporting Information 1).
3.1. Comparison of diversity metrics between community types
The comparison between non-invaded and invaded communities (Test 1) revealed significantly higher TD (species richness), FD (FDiv and FEve), but not MPD, in the non-invaded communities (all p-values < 0.041, df = 2; Fig. 2; Supporting Information 3). We found similar results between non-invaded and invaded no alien communities (Test 2). In addition, MPD was significantly higher in non-invaded communities (all p-values < 0.017, df = 2; Fig. 2; Supporting Information 3). When we compared invaded vsinvaded no alien communities (Test 3), there were no significant differences except for TD and FDiv. In these cases, the metrics were always lower for invaded no alien than invaded communities (all p-values < 0.048, df =2; Fig. 2; Supporting Information 3). There was a significant difference between observed and null communities for MPD only for invaded no alien communities when we considered the Feral Pigeon as an alien species (negative sesMPD values; V = 12380, p-value < 0.01). Finally, the three tests performedwithin land-uses showed similar patterns as the testsacross land-uses, except for MPD that was higher ininvaded than non-invaded communities in urban areas (Supporting Information 2-3).
Trends for TD (species richness) and FD (FDiv and FEve) were similar but clearer in communities considering Feral Pigeon as alien compared to communities considering Feral Pigeon as native (for all three Tests). Regarding MPD, the trend for Test 1 & 2 remained unaltered considering feral pigeon as alien while, for Test 3, invaded communities showed a significantly higher MPD metric than invaded no aliencommunities (Supporting Information 2-3).
Effect of landscape configuration and composition on diversity metrics
TD was positively associated with agricultural and forest areas, along with SHDI, in non-invaded communities (Fig. 3; Supporting Information 3). Native forest and SHDI were positively correlated with TD both ininvaded and invaded no alien communities, whereas rice field areas were negatively correlated in both communities (Fig. 3; Supporting Information 3). In non-invaded communities, FDiv was positively correlated with SHDI, while intensive agricultural areas negatively affected this metric (Fig. 3; Supporting Information 3). Ininvaded no alien communities, FDiv was negatively correlated with agricultural (intensive and extensive) and rice field areas along with SHDI (Fig. 3; Supporting Information 3). The FEve (i.e. resilience) was negatively affected by urban (non-invaded and invaded no alien communities) and intensive agricultural areas (non-invadedcommunities), and was positively associated with plantation forest in both invaded and invaded no alien communities (Fig. 3; Supporting Information 3)
Rice field areas were negatively associated with MPD in the three communities (Fig. 3; Supporting Information 3). MPD was also negatively correlated with intensive agricultural areas and SHDI in invadedcommunities, while was negatively correlated with urban areas ininvaded no alien communities (Fig. 3; Supporting Information 3).
Averaged GLMM models showed similar results, in both approaches, for TD on non-invaded communities, and the MPD of invadedcommunities (Supporting Information 2-3). The major differences between the two approaches were found for FDiv and FEve of invaded no alien communities, and MPD of non-invaded communities (Supporting Information 2-3).