Figure 10.

Results of “contests” between simulated male moths seeking a pheromone source: variation across memory timescale, τ, and advance parameter, $c_1^{\ast }$, with fixed plume-crossing probability, P _success =0.25. In contests, a subset of simulated moths is released within a plume at the same time and downstream position. Analogously to searching by male moths for pheromone-releasing females, the first moth to locate the source “wins”. The plots represent the number of contests won by simulated moths with corresponding parameters, normalized by the expected number if outcomes were random, with two, three or four contestants (left, middle and right, respectively). The overall higher winning metrics for three- and four-way contests reflect the relatively smaller frequency of contests with no winners (i.e., all contestants failed to locate the source). Depending on the number of contestants, two types of odor source location strategies are most successful. One successful strategy corresponds roughly to parameter values ($c_1^{\ast }\approx 0.4$, τ≥6) with long search times but nearly maximal search success probability (Figure 8), i.e., to “risk-averse” behaviors. The other successful strategy (roughly $c_1^{\ast }\approx 1$, τ≈3) maximizes neither search success probability nor search efficiency, but occupies a region of parameter space in which these metrics are balanced effectively. Because these behaviors fail in locating the source more frequently than risk-averse behaviors, but are faster when they do succeed, they reflect a more “risk-tolerant” strategy. The plots illustrate a shift in contest-winning behaviors away from risk-averse towards risk-tolerant strategies, as the number of contestants increases. This trend reflects strong density-dependence in performance metric tradeoffs: Risk-averse, low efficiency behaviors are likely to win only if there are no successful contestants with risk-tolerant behaviors. As the number of contestants increases, the likelihood that at least one risk-tolerant contestant is successful also increases, shifting the tradeoffs between performance metrics.