Skip to main content
. 2016 Jul 8;11(7):e0158937. doi: 10.1371/journal.pone.0158937

Table 1. Model selection for factors affecting bee abundance in highbush blueberry fields of British Columbia (BC) and Michigan (MI).

Only suboptimal models with ΔAICc ≤ 3 are shown.

Response Region Explanatory variables df logLik AICc delta Model weight
Wild bees BC Honey bees + Land use (2000 m) + Row orientation 9 -207.07 431.27 0 0.32
Land use (2000 m) + Row orientation 7 -208.31 431.49 0.21 0.29
Honey bees + Land use (2000 m) 7 -208.34 431.55 0.27 0.28
MI Distance from edge + Honey bees + Land use (300 m) + Row orientation 11 -226.74 476.88 0 0.52
Distance from edge + Honey bees + Land use (300 m) + Row orientation + Orientation/distance interaction 14 -224.08 478.41 1.54 0.24
Distance from edge + Land use (300 m) + Row orientation 10 -229.07 479.31 2.43 0.16
Honey bees BC Land use (300 m) + Row orientation 7 -345.20 705.27 0 0.40
Land use (300 m) 6 -346.66 705.97 0.70 0.28
Stocking rate + land use (300 m) + Row orientation 8 -345.18 707.49 2.22 0.13
Stocking rate + land use (300 m) 7 -346.58 708.03 2.76 0.10
MI Land use (1000 m) 6 -717.93 1448.30 0 0.28
Land use (1000 m) + Row orientation 7 -717.38 1449.34 1.04 0.17
(null) 5 -719.74 1449.79 1.49 0.13
Stocking rate + land use (1000 m) 7 -717.92 1450.42 2.12 0.11
Row orientation 6 -719.16 1450.75 2.45 0.08