Results
Short-term effects of ant occupants and herbivores on A. drepanolobium physiology- In wet seasons, Control trees had lowerAmax-leaf , lowerEleaf , and higherψleaf in 2018 than in 2017, but Transition trees exhibited no significant differences in any of those traits (Amax-leaf : F1,131 = 5.16, P = 0.0231, Fig. 1A;Eleaf : F1,131 = 8.99, P = 0.0027, Fig. 1C; ψleaf : F1,122 = 3.88, P = 0.0488; Fig. 1E). During dry seasons,Amax-leaf was substantially lower in 2018 than in 2017 for newly-invaded trees, while Control trees exhibited a significantly smaller interannual decline (F1,127 = 17.78, P < 0.01; Fig. 1B).Eleaf was significantly higher for Control trees in 2018 than in 2017, while Transition trees remained consistent across years(F1,127 = 4.53, P = 0.0332; Fig. 1D), trees at both sites exhibited consistentψleaf in both years (F1,127 = 0.00, P = 0.99; Fig. 1F).
Long-term effects of ant occupants and herbivores on Acacia drepanolobium physiology - Trees occupied by C. mimosae vs.P. megacephala in long-term Uninvaded and Invaded sites differed substantially in leaf water potential ranges, leaf gas exchange rates, and canopy gas exchange capacities. During wet seasons, trees in Invaded areas had reduced Amax-leaf(F1,205 = 11.55, P < 0.001; Fig. 2A),Eleaf (F1,205 = 15.68, P <0.0001; Fig. S1A), Amax-canopy (F1,94 = 42.33, P < 0.0001; Fig. 2A), andEcanopy (F1,94 = 42.70, P < 0.0001; Fig. S1A); however,ψleaf did not significantly vary (F1,193 =1.39, P = 0.24; Fig. S1C). During dry seasons, trees in Invaded areas had higherAmax-leaf (F1,212 = 4.55, P = 0.0329; Fig. 2B) andEleaf (F1,212 = 5.96, P < 0.0001; Fig. S1B), but also slightly decreasedψleaf (F1,208 = 4.52, P = 0.0334; Fig. S1D) by ca . 0.1 MPa; we did not find differences in Amax-canopy (F1,53 = 0.06, P = 0.80; Fig. 2B) or Ecanopy(F1,53 = 0.49, P = 0.87; Fig. S1B) for long-term sites.
During the rainy season, trees in Invaded areas that were exposed to vertebrate herbivores had lower Amax-leaf (F2,205 = 5.03, P = 0.025; Fig. 3A),Amax-canopy (F2,94 = 53.07, P < 0.0001; Fig. 3C), andEcanopy (F2,94 = 50.30, P < 0.0001; Fig. 4A), and non-significantly lowerEleaf (F2,205 = 2.99, P= 0.084; Fig. S2A); ψleaf did not differ (F2,193 = 0.20, P = 0.66). During the dry season, invaded trees in each herbivory treatment did not differ inAmax-leaf (F2,212 = 1.94, P = 0.16; Fig. 3B), Amax-canopy(F2,53 = 1.71, P = 0.19; Fig. 3D), orψleaf (P = 0.93; Fig. 4D), but tree in Invaded areas that were exposed to vertebrates had slightly higherEleaf (F2,212 = 3.99, P= 0.0459; Fig S2B) and Ecanopy(F2,53 = 5.70, P = 0.0189; Fig. 4B).
Through ant removal in our factorial experiment, we linked occurrence ofP. megacephala ants with differences in water potential range and gas exchange in both wet and dry seasons. During wet seasons, Invaded trees that were experimentally cleared of P. megacephala had higher Amax-canopy(F2,94 = 10.57, P < 0.001; Fig. 3C) andEcanopy (F2,94 = 10.77, P < 0.001; Fig 3A), and slightly smallerψleaf (F2,193 = 4.32, P = 0.0378; Fig. 4C) by ca . 0.15 MPa; we did not find significant differences in Amax-leaf(F2,205 = 0.19, P = 0.67; Fig. 3A) orEleaf (F2,205 = 0.07, P = 0.79; Fig. S2A) on trees cleared of P. megacephala . In dry seasons, trees occupied by P. megacephala had slightly higherEleaf (F2,212 = 6.36, P = 0.0117; Fig S2B) and Ecanopy (F2,53 = 8.38, P = 0.0038; Fig. 4B) than P. megacephala removal trees, but we found no differences between these treatments in Amax-leaf(F2,212 = 0.01, P = 0.92; Fig. 3B),Amax-canopy (F2,53 = 3.11, P = 0.08; Fig. 3D), or ψleaf (F2,208 = 2.57, P = 0.11; Fig. 4D). In contrast to trees from long-term Invaded sites, trees occupied by C. mimosaeat Uninvaded sites did not exhibit different gas exchange or leaf water potential traits due to ant removal, herbivore exclusion or their interaction (Note S3).