Fig. 5From: What can ecosystems learn? Expanding evolutionary ecology with learning theoryEcological dynamics after evolution in varying environmental conditions. The evolved ecosystem exhibits two attractor states (rightmost frames) that are reached from various initial species densities (leftmost frames). a Random initial species densities develop into one of two possible attractors corresponding to the patterns of forcing experienced in the evolutionary past. b Initial configurations that resemble a small part of E 1 (i and ii) or E 2 (iii and iv) develop into equilibria that fully recreate E 1 and E 2 respectively. c Initial configurations that are partially randomised versions of E 1 (i. 20 %, ii. 80 %) or E 2 (iii. 80 %, iv. 20 %) develop into equilibria that ‘repair’ the corresponding state. e For initial conditions between E 1 and E 2, (E 1:E 2 ratio = i.80:20, ii.55:45, iii.45:55 iv.20:80) the dynamics ‘recognise’ the pattern that is resembled most closelyBack to article page