Mutual cheating strengthens a tropical seed dispersal mutualism

<b>Analyses reported in the article of "</b><b>Mutual
cheating strengthens a tropical seed dispersal mutualism</b><b>" can be reproduced using </b><b>the data. </b><b><br></b><b><i>Abstract</i>. </b>While cheating can cause the degradation or collapse of mutualisms, mutualisms may theoretically stabilize or strengthen if the cheating is mutual. Here, we present an asymmetric two-player game model to explore the evolutionary dynamics of mutual cheating in a mutualistic interaction. We found that the interaction evolved towards mutual cheating if cheating can help both partners obtain higher benefits or if counter-cheating yields more benefits to victims than simply tolerating exploitation by partners. Then, we present empirical evidence for such mutual cheating strengthening a seed dispersal mutualism in which rodents disperse seeds by scatter hoarding, rodents sabotage seed germination by pruning radicles, and seeds escape rodents by resprouting. By tracking &gt;8000 Pittosporopsis kerrii seeds throughout the dispersal process in a tropical forest in southwest China, we found that rodents provided better dispersal to seeds that they pruned, i.e., pruned seeds were dispersed farther and were more likely to establish seedlings than unpruned seeds. Compared to unpruned seeds, pruned seeds retained more of their nutrients, i.e., dry mass of pruned seeds was greater than that of unpruned seeds, and were stored for longer by rodents. These findings indicate that mutual cheating benefited both partners. Payoffs estimated from the field experiments indicated that mutual cheating was indeed favored in rodents and plants P. kerrii, and that neither partner was enslaved by the other under mutual cheating. Rather, the mutualism remained stable because the partners were able to exploit each other, and each partner attempted to gain the maximum benefits from the interaction. Our findings indicate that mutual cheating between two mutualists can enhance and stabilize mutualisms.<br><b><br></b><b><i>Data analysis</i></b><br>Generalized linear mixed models (GLMMs) were used to analyze the relationships between the probability of seedling establishment from scatter-hoarded seeds or 2-yr seedling survival and their potential influencing factors (i.e., radicle pruning, number of times pruned, number of movements, deposition microhabitats, burial substrates, dispersal distance and seed size), respectively. Seedling establishment from scatter-hoarded seeds and 2-yr seedling survival were modeled as a binomial variable with a logit-link function (i.e., 1 for establishment or survival and 0 for death). Radicle pruning, number of times pruned, number of movements, deposition microhabitats, burial substrates, dispersal distances and seed sizes were included as fixed effects, year and seed station nested within stand were included as random effects.<br>A GLMM with gamma errors was used to analyze the differences in the dispersal distances between unpruned and pruned seeds. The action of radicle pruning (i.e., unpruned vs. pruned) was included as a fixed effects, year and seed station nested within stand were included as random effects.<br>A Cox regression model was used to compare the differences in seed survival time (storage time in caches) between unpruned and pruned seeds.<br><b></b>