Figure 4. Relationship between responses and population trends.Panel (a) shows the mean population trend (±SEM) for species with different combinations of responses in phenology and NRB. The pgsl model b) indicated that a range shift had a significant positive effect on the population trend (Est.= 0.37; t=2.15; p<0.05). Panel (b) shows the mean population trend (±SEM) for species with different hypothesis-wise responses (dark grey= hypothesis 0 ̵̶- the species neither shift NRB nor phenology; dark blue= hypothesis 1 ̵̶- the species shifts either NRB or phenology; light blue= hypothesis 2 ̵̶- the species shifts both NRB and phenology). Species that responded according to Hypothesis 2 showed the strongest population trends (Est.= 0.55; t=3.08; p<0.01).
Table 1. Conversion of phenology and NRB estimates into categorical response groups. a) The slope coefficients from the phenological analyses and NRB shift estimates were used to convert continuous responses into categorical response groups. b) To enable tests of our hypotheses, we further combined these phenology and NRB response groups into 1) a four-level category describing combined response groups (RG) and 2) a three-level category describing hypothesis-wise responses (Hypotheses 0, 1, or 2), both corresponding to the potential combined responses shown in Fig. 1b.