Social versus nonsocial cues and responses: a reply to Alizon


  • F. Harrison

    Corresponding author
    1. Centre for Biomolecular Science, School of Molecular Medical Sciences, University of Nottingham, University Park, Nottingham, UK
    • Correspondence: Freya Harrison, School of Molecular Medical Sciences, Centre for Biomolecular Science, University of Nottingham, University Park, Nottingham NG7 2RD, UK. Tel.: 0115 8232010; fax: 115 846 8002; e-mail:

    Search for more papers by this author

Alizon (2013) provides an interesting and constructive reply to my experimental demonstration of plastic responses to siderophore cheating in Pseudomonas aeruginosa (Harrison, 2013). I fully agree with his statement that ‘[t]he increase in siderophore production in the presence of cheaters cannot (yet) be interpreted in the context of social evolution theory’, and I hope that the debate we have started highlights gaps in the field that are ripe for exploration.

That an apparently ‘social’ behaviour can be a response to aspects of the nonsocial environment is a valid and important point, and Alizon's simple model of the effects of cheats as siderophore sinks is a neat contribution to understanding the mechanistic underpinnings of the responses I measured. However, the model may have limited applicability because pyoverdine is recycled (Imperi et al., 2009; Kümmerli & Brown, 2010), meaning that uptake of a molecule by a cell is not necessarily equivalent to its loss from the environment. Further, I would stress that behaviours that evolved as responses to nonsocial cues can still have social consequences. This is exemplified by the recent diffusion sensing/quorum sensing (QS) debate (Redfield, 2002; West et al., 2012) where it was argued (Redfield, 2002) that it is most parsimonious to explain autoinducer molecules as inferring diffusion properties of their environment rather than gauging the density of their producer clone. It is quite possible that QS molecules actually evolved to sense diffusion, but in sensing local molecular concentrations they unavoidably also gather information on social conditions (number of cells). This has consequences for other social behaviours because the responses controlled by QS include the production of factors that are undeniably public goods and so evolve under social as well as nonsocial selection pressures (see West et al., 2012). Additionally, Cornforth & Foster (2013) have explored the potential for responses to environmental stresses typically thought of as asocial to have a strong social cause. In particular, these authors suggest that nutrient limitation can correlate with competitor density and so could be co-opted into a ‘competition sensing’ response.

In the case of the Pseudomonas siderophore system, the critical points to note are:

  1. Siderophore production is social under the conditions I use, because siderophores can be used by nonproducing cells, and these show exactly the dynamics we expect of social ‘cheats’ (Griffin et al., 2004; Harrison, 2013); and
  2. Producer (cooperator) cells alter their level of siderophore production in the presence of cheats, so the presence of cheats cues a change in social behaviour by cooperators – whether or not the response to cheats is direct or simply due to the effect of cheats on siderophore concentration.

The key question is therefore neither ‘is the response of bacteria to the presence of cheats social?’ nor ‘do bacteria directly sense the presence of cheats?’ but ‘what consequences (if any) do responses to cheats have for the evolutionary dynamics of social behaviours?’ It would certainly be ideal to conduct an explicit empirical test of the effects of siderophore loss to diffusion versus cheating, but it would be nontrivial to find a way to ‘syphon off’ siderophores from experimental populations without introducing other, potentially confounding effects such as loss of other exoproducts and disturbance of populations.

Models that explore optimal response rules in multiplayer public goods scenarios are conspicuous by their absence from the published cooperation literature. I certainly do not intend to argue that what seems to work in the case of dyadic interactions between parents must also work in the context of bacterial public goods – as Alizon notes, there are key differences in how fitness pay-offs of the players are related that may critically determine the evolutionary success of different response rules in the two scenarios. However, the question of how individuals should dynamically adjust their investment over an extended period of social interaction really only been studied in detail in the case of biparental care, and the observation that the pattern of P. aeruginosa's responses to cheats is so reminiscent of the response rule shown to be optimal in the former scenario is surprising. Whether this is coincidence, artefact or evolutionarily important can only be determined by further work on the plasticity of bacterial responses to changes in social investment and by the development of models that explicitly address the likely benefits of different response rules in multiplayer cooperation games. These are both areas that would add a novel dimension to the field of social evolution.


I would like to thank Daniel Cornforth and Steve Diggle for helpful discussion.