Fig. 3: Behavior of longfin damselfish indicative of domesticator role.

Number of chases by damselfish (Stegastes diencaeus) towards other fishes during (a) predation experiment (significant effect of location (zero-inflated GLMM): \(\chi _1^2\) = 68.95, P < 0.001; significant effect of treatment (zero-inflated GLMM): \(\chi _2^2\) = 6.07, P = 0.048, n = 30 damselfish farms with 30 trials inside the farm and 30 trials conducted outside of the farm) (post-hoc test included an mvt correction) and (b) 30-min behavioral observations (significant effect of farm type (glm): F_{1, 57} = 5.4, P = 0.024, n = 60 fish observed, including 30 that hosted mysids within their farm and 30 that did not host mysids within their farm). (c) Number of bites on farmed substrate by damselfish during 30-min behavioral observations (significant effect of farm type (glm): F_{1, 58} = 8.14, P = 0.006, n = 60 fish observed, including 30 that hosted mysids within their farm and 30 that did not host mysids within their farm) and (d) hepatosomatic index (HSI) of fish that either did or did not host mysids (significant effect of farm type (glm): F_{1, 58} = 82.2, P < 0.001, n = 60 fish, including 30 that hosted mysids within their farm and 30 that did not host mysids within their farm). Treatments for (a) are outlined in the Fig. 2 legend. Boxplots show median values (horizontal lines), interquartile range (boxes) and minimum and maximum values (whiskers). Asterisks indicate significant differences (P < 0.05 = *, P < 0.01 = **, P < 0.001 = ***). Note that while our analysis found a significant effect of treatment on the number of chases in the predation experiment (3a, P = 0.048), the more conservative multiple comparison post-hoc test found a slightly non-significant result (P = 0.089).