Skip to main content

Advertisement

Table 1 The influence of familiarity and experience treatments on subordinates’ strategies in Contest 2

From: Opponent familiarity and contest experience jointly influence contest decisions in Kryptolebias marmoratus

   (1) Displaying/attacking (2) Attacking
vs. vs.
Nonaggressive Displaying
(N= 240) (N= 128)
Effect df b ±SE χ 2 P b±SE χ 2 P
Familiarity (UF vs. F) 1 1.20±0.30 16.35 <0.001* −0.26±0.50 0.27 0.607
Experience 2   1.54 0.462   9.91 0.007*
 SW-DN vs. SN-DN 1 0.06±0.37 0.03 0.872 2.05±0.68 9.15 0.003*
 SN-DL vs. SN-DN 1 −0.35±0.36 0.96 0.327 0.11±0.46 0.06 0.810
Familiarity × experience 2   2.07 0.356   1.91 0.385
Dominant attacking directly (Y vs. N) 1 −2.31±0.58 16.00 <0.001* −0.03±1.36 0.00 0.980
Size 1 −0.08±0.06 1.41 0.236 −0.05±0.10 0.20 0.657
Lineage 4   11.36 0.023*   0.70 0.952
  1. Generalized linear models evaluating the influence of familiarity (F: familiar; UF: unfamiliar) and experience (SW-DN: subordinates received a winning experience and dominants received a no-contest experience; SN-DL: subordinates received a no-contest experience and dominants received a losing experience; SN-DN: both opponents received a no-contest experience) treatments as well as the interaction between them on the most aggressive behavior the subordinates exhibited in Contest 2. Because very few subordinates attacked without displaying first (Level 4), these individuals were pooled with those that attacked after first displaying (Level 3) to form an “attacking” group. Consequently, the subordinates were classified into 3 groups based on the most aggressive behavior exhibited in Contest 2: nonaggressive, displaying and attacking. We examined (1) the likelihood that subordinates would act aggressively (by either exhibiting opercular displays or launching attacks toward their dominant opponents; Levels 2–4) compared with the likelihood of them not acting aggressively (exhibiting none of these behaviors; Level 1) and (2) for the aggressive subordinates (Levels 2–4) only, the likelihood that they would launch attacks (Levels 3 and 4) compared with the likelihood of going no further than exhibiting opercular displays (Level 2). The pair’s body size and lineage and whether the dominant opponent’s launched attacks directly (Y: yes; N: no) were included in the models to account for their influences. Contrast analyses were used to evaluate the differences between the effects of different levels of a treatment. (χ2: Ward χ2; *: P < 0.05).