This commentary is addressed only to John et al.'s application of their ideas to ecological situations. I contend that two of their examples of failure are, in fact, not correct. However, this does not mean that “proxy failure” could not be relevant to other ecological situations. I suggest two other possible and common situations.
The first proposed proxy failure is “runaway sexual selection,” where as a hypothetical example, peahens might have a “goal” of producing the healthiest offspring, but simply “like” to mate with peacocks with the largest, brightest tails (Andersson, Reference Andersson1994), this despite such tails make males less healthy (e.g., imposing a cost of not being able to effectively avoid predators or skimping on investing in their immune systems). However, if one realizes that the true goal of a peahen is to produce more grandchildren than any other peahen (i.e., have the highest evolutionary fitness) then there is no proxy failure. It is unrewarding for a peahen to produce survival-maximizing, or the “healthiest”, sons either through her own genotype or from a chosen healthy but less colorful male. Those sons will likely mate with few or no females and, therefore, are wasted investment from the standpoint of their mother. In contrast, a son with a large bright tail will produce many grandchildren. Hence, even in a runaway selection scenario, the proxy measure of a big, bright tail is accurate to the goal of reproductive success through the generations. The authors suggest that proxy traits are stable with runaway sexual selection because the trait and female preference are genetically linked. I suggest the reverse: This linkage evolves because the proxy trait accurately predicts a fitness-enhancing choice.
The second proposed proxy failure is the example of a marsupial neonate that succeeds with stronger arms in securing its mother's teat even if the initial arm investment comes at a cost for other parts of its development that eventually result in a less viable baby. This example, however, has a problem that the authors do not address – the same individual will be an agent (as a neonate) and a regulator (as a mother) in the same evolutionary conflict. How is this to be resolved; and to advantage of which part of the individual's life history? Nevertheless, if we again simply change the perspective and the goal, the proxy failure goes away. Consider that the neonate's goal is to have the highest chance of becoming a future mother relative to all other neonates in the population. It then makes no sense, evolutionarily, to gain a teat at a cost of dying before reproducing. Thus, selection on neonates should result in the optimal trade-off between investment to gain a teat and the resulting downstream probability to survive and reach motherhood. In essence, this is exactly the same goal as the neonate's mother.
I propose to the authors that there are two better ecological examples for their hypotheses. The first is kin recognition, particularly in the case of creating paternal certainty. Hypothetically, a male bird could be certain that the chick hatching from his mate's egg is indeed his offspring, if that offspring produces one or several genetically determined physical traits that could only have come from the male. These would be “greenbeards” and act as proxies for genome-level genetic kinship (Gardner & West, Reference Gardner and West2010; Nonacs, Reference Nonacs2011) – with the result that “father” invests more heavily in providing parental care for those he accurately “knows” are his offspring. Although this would seem to be advantageous for both father and offspring, greenbeards and true genetic kin recognition appear to be quite rare in nature. The evolutionary problem is that not all hatchlings may be the offspring of the male, and there is a differential cost to revealing one's genetics. A true offspring might get some more care, but offspring revealed not to be the male's, could get killed. Thus, a non-informative system may, on average, best serve all offspring. Again, realize that a male may have survived to become a father only because his “father” failed to recognize that he was not. The overall result appears to be that males are often left relying on some environmental or behavioral proxy (e.g., “I mated with this female, so the chicks could be mine”), which is not foolproof. Therefore, how honest any paternity proxy is, may rely on whether or not females prefer genetic monogamy over nonmonogamy (Møller, Reference Møller2020). Note also that the failure and lack of a genetic proxy here may not be because of deceptive signals by the agents – that is, “I am your offspring,” when that is untrue – but to the advantages of not signaling at all.
The second is warning coloration. Many species that are dangerous or poisonous advertise this with bright colors and high visual contrasts. This is a proxy for potential predators to warn them that it will cost more to attack than to avoid. Therefore, the predator and the dangerous individuals may have the same shared goal: Do not engage and attack (Aubier & Sherratt, Reference Aubier and Sherratt2020). This proxy, however, is easily and often cheated by many prey species that “mimic” the danger signals and colors without actually being either dangerous or poisonous. Yet the proxy communication systems rarely collapse from the presence of cheaters, even though the proxy signals may be very inaccurate relative to the agent's true state.
The above two examples illustrate the importance of cost in the evolution (or nonevolution) of proxy signaling systems. In the first, an accurate and honest proxy signal may result in a massive loss of fitness if it is presented to the wrong individual. Here, it is the agent that risks paying a disproportionate cost. In the case of warning coloration, the costs can be extreme for the regulator, in mistaking an honest proxy signal for a dishonest one. At least as far as ecological systems are concerned, proxy success or failure will strongly depend on the costs of mistakes and which parties have to differentially pay them.
You have
Access
This commentary is addressed only to John et al.'s application of their ideas to ecological situations. I contend that two of their examples of failure are, in fact, not correct. However, this does not mean that “proxy failure” could not be relevant to other ecological situations. I suggest two other possible and common situations.
The first proposed proxy failure is “runaway sexual selection,” where as a hypothetical example, peahens might have a “goal” of producing the healthiest offspring, but simply “like” to mate with peacocks with the largest, brightest tails (Andersson, Reference Andersson1994), this despite such tails make males less healthy (e.g., imposing a cost of not being able to effectively avoid predators or skimping on investing in their immune systems). However, if one realizes that the true goal of a peahen is to produce more grandchildren than any other peahen (i.e., have the highest evolutionary fitness) then there is no proxy failure. It is unrewarding for a peahen to produce survival-maximizing, or the “healthiest”, sons either through her own genotype or from a chosen healthy but less colorful male. Those sons will likely mate with few or no females and, therefore, are wasted investment from the standpoint of their mother. In contrast, a son with a large bright tail will produce many grandchildren. Hence, even in a runaway selection scenario, the proxy measure of a big, bright tail is accurate to the goal of reproductive success through the generations. The authors suggest that proxy traits are stable with runaway sexual selection because the trait and female preference are genetically linked. I suggest the reverse: This linkage evolves because the proxy trait accurately predicts a fitness-enhancing choice.
The second proposed proxy failure is the example of a marsupial neonate that succeeds with stronger arms in securing its mother's teat even if the initial arm investment comes at a cost for other parts of its development that eventually result in a less viable baby. This example, however, has a problem that the authors do not address – the same individual will be an agent (as a neonate) and a regulator (as a mother) in the same evolutionary conflict. How is this to be resolved; and to advantage of which part of the individual's life history? Nevertheless, if we again simply change the perspective and the goal, the proxy failure goes away. Consider that the neonate's goal is to have the highest chance of becoming a future mother relative to all other neonates in the population. It then makes no sense, evolutionarily, to gain a teat at a cost of dying before reproducing. Thus, selection on neonates should result in the optimal trade-off between investment to gain a teat and the resulting downstream probability to survive and reach motherhood. In essence, this is exactly the same goal as the neonate's mother.
I propose to the authors that there are two better ecological examples for their hypotheses. The first is kin recognition, particularly in the case of creating paternal certainty. Hypothetically, a male bird could be certain that the chick hatching from his mate's egg is indeed his offspring, if that offspring produces one or several genetically determined physical traits that could only have come from the male. These would be “greenbeards” and act as proxies for genome-level genetic kinship (Gardner & West, Reference Gardner and West2010; Nonacs, Reference Nonacs2011) – with the result that “father” invests more heavily in providing parental care for those he accurately “knows” are his offspring. Although this would seem to be advantageous for both father and offspring, greenbeards and true genetic kin recognition appear to be quite rare in nature. The evolutionary problem is that not all hatchlings may be the offspring of the male, and there is a differential cost to revealing one's genetics. A true offspring might get some more care, but offspring revealed not to be the male's, could get killed. Thus, a non-informative system may, on average, best serve all offspring. Again, realize that a male may have survived to become a father only because his “father” failed to recognize that he was not. The overall result appears to be that males are often left relying on some environmental or behavioral proxy (e.g., “I mated with this female, so the chicks could be mine”), which is not foolproof. Therefore, how honest any paternity proxy is, may rely on whether or not females prefer genetic monogamy over nonmonogamy (Møller, Reference Møller2020). Note also that the failure and lack of a genetic proxy here may not be because of deceptive signals by the agents – that is, “I am your offspring,” when that is untrue – but to the advantages of not signaling at all.
The second is warning coloration. Many species that are dangerous or poisonous advertise this with bright colors and high visual contrasts. This is a proxy for potential predators to warn them that it will cost more to attack than to avoid. Therefore, the predator and the dangerous individuals may have the same shared goal: Do not engage and attack (Aubier & Sherratt, Reference Aubier and Sherratt2020). This proxy, however, is easily and often cheated by many prey species that “mimic” the danger signals and colors without actually being either dangerous or poisonous. Yet the proxy communication systems rarely collapse from the presence of cheaters, even though the proxy signals may be very inaccurate relative to the agent's true state.
The above two examples illustrate the importance of cost in the evolution (or nonevolution) of proxy signaling systems. In the first, an accurate and honest proxy signal may result in a massive loss of fitness if it is presented to the wrong individual. Here, it is the agent that risks paying a disproportionate cost. In the case of warning coloration, the costs can be extreme for the regulator, in mistaking an honest proxy signal for a dishonest one. At least as far as ecological systems are concerned, proxy success or failure will strongly depend on the costs of mistakes and which parties have to differentially pay them.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.