Abstract
In this paper, I identify two general positions with respect to the relationship between environment and natural selection. These positions consist in claiming that selective claims need and, respectively, need not be relativized to homogenous environments. I then show that adopting one or the other position makes a difference with respect to the way in which the effects of selection are to be measured in certain cases in which the focal population is distributed over heterogeneous environments. Moreover, I show that these two positions lead to two different interpretations—the Pricean and contextualist ones—of a type of selection scenarios in which multiple groups varying in properties affect the change in the metapopulation mean of individual-level traits. Showing that these two interpretations stem from different attitudes towards environmental homogeneity allows me to argue: (a) that, unlike the Pricean interpretation, the contextualist interpretation can only claim that drift or selection is responsible for the change in frequency of the focal trait in a given metapopulation if details about whether or not group formation is random are specified; (b) that the traditional main objection against the Pricean interpretation—consisting in arguing that the latter takes certain side-effects of individual selection to be effects of group selection—is unconvincing. This leads me to suggest that the ongoing debate about which of the two interpretations is preferable should concentrate on different issues than previously thought.
Notes
Since this debate only concerns MLS1 cases, this paper will do so as well, leaving aside the type of cases known as MLS2 (in which the target of interest is the change in the average value of group-level traits). Moreover, since they are not integral to the focal debate, this paper also leaves aside MLS1 cases that involve group characters like population density, division of labor etc., which are not mere averages of the individual traits within the group.
Sober (2011) seemed somewhat reluctant to agree that the two approaches define group selection differently; but that no longer seems to be the case, as he recently stated (Sober 2015, 836) that “the two approaches disagree about what group selection is”, before going on to note that his own “definitions of group and individual selection are compatible with the Price approach but not with the contextual approach”.
Even proponents of the recent “statisticalist” view of natural selection seem to agree with this, since they avowedly do not contest statements like “variation in camouflage causes evolutionary change” (Matthen and Ariew 2009, 203).
I borrow the term “Causal Condition” from Glymour (2014), but I attach a different meaning to it.
To avoid unnecessary complications, I assume that all eggs hatch and that, on the whole, an equal number of eggs are laid in environments E1 and E2.
Because they are conditioned to do so by some factor, be it genetic or not.
Let me add that I am not claiming that a researcher that is only interested in explaining why a type tolerates one environment (say E 1 ) better would be wrong in studying only E 1 and in considering it as a homogenous selective environment in its own right. What I am claiming—following Brandon—is that, if one’s goal is that of providing an explanation for the total evolutionary change in this population, one would be wrong to consider E 1 and E 2 as distinct homogenous selective environments.
This entails that, in cases in which the within-environment difference in fitness has a “single-cause explanation” (Frank 2012, 1014), nothing stands in the way of considering Price’s equation to be causally relevant, as Frank implicitly agrees—an example of this sort was Brandon’s Drift-Case analyzed above. Moreover, this is the type of case that lies at the heart of the debate about the interpretation of MLS1 that I will deal with in the second part of this paper. Indeed, this debate revolves exclusively around MLS1 cases in which it is assumed that the within-group fitness differences between individuals are due to their difference in a single relevant trait. I will uphold this assumption here. But I will also add that, if we were to shift our attention towards cases in which within-group fitness differences do not have a “single-cause explanation”, as Frank puts it, speaking of “Pricean” and “contextualist” interpretations of MLS1 would become strained, since both would actually make use of some form of multiple regression analysis. However, the distinction I make here between a position that adopts EHR and one that rejects it would remain intact in such cases as well, and this corroborates my point that the Pricean and contextualist interpretations of MLS1 need to be seen as diverging positions with respect to EHR.
This example is similar to the one I used in Jeler (2015), but the conclusions I will draw from it differ from and even contradict the ones from that paper (see footnote 16 below).
I will loosen this assumption in “The wider version” section.
The fitness values are rounded to the second decimal point, as will also be the case below for the values of frequency changes and relative fitnesses.
By this I do not mean to say that, according to these authors, the Pricean interpretation necessarily sees groups as strongly identified or highly cohesive entities.
Though, of course, there will be no response to selection, because the green color will not be passed on to the next generation.
See Okasha (2004b, 2005, 2006) for a discussion—based on Nunney (1985a)—of the cases in which this choice of variables is appropriate. The application of contextual analysis by using, as an independent variable, the “neighborhood character” and, respectively, the “group character” is what Okasha calls the “neighborhood” and, respectively, the “contextual approaches”. But the two are essentially the same approach and only differ—very slightly—in their choice of variables (the neighborhood character being the preferable variable when an individual’s fitness depends on the average character of its neighbors, whereas the group character is the preferable variable for cases in which an individual’s fitness depends on the average character of its group, the focal individual included).
In Jeler (2015, 2016), I argued that what the contextualist interpretation calls “group selection” (i.e. “selection on neighborhood/group character”, depending on the details of the case) can be seen as a form of frequency-dependent selection in multi-group scenarios. If true, this idea would hold irrespective of whether the non-zero value of the covariance between individual character and neighborhood/group character reflects a chance event or a phenotypic trait. However, this amounts to nothing more than claiming that the only factor that is responsible for this “selection on neighborhood/group character” is the fact that one type is found more often in the richer environment. But, as shown above, the differential fitness that comes from the uneven distribution of types into environments of different quality cannot be called selection unless this distribution pattern reflects a difference in phenotypic trait (e.g. assortative preferences).
Gardner (2015b) makes a similar point, but brings little arguments in its support.
Bourrat (2015) recently made a very similar claim, but his argument is seriously undermined by the assumption that natural selection can only act on “intrinsic-invariable properties”. This is tantamount to claiming, for example, that frequency-dependent selection is not natural selection.
Without further dwelling on this point, I will just indicate that, as Okasha (2006, 89) has shown, Eq. (5) may be derived from the above discussed Eq. (2) of multiple regression analysis if we keep in mind that in the MLS1 cases at issue here the fitness of a group is the average fitness of its members—and consequently \(Cov(Z_{j} ,W_{j} ) = Cov(Z_{j} ,w_{ij} )\)—and a group’s character is the average character of its individual members—and consequently \(Cov(z_{ij} ,Z_{j} ) = Var(Z_{j} )\).
Obviously, in cases in which the indirect effects of the focal trait on fitnesses are very small, the differences between the relative fitnesses of types across the given groups might be extremely small and one could reasonably discount these “grey-area” cases as involving multiple homogenous environments.
And this is, of course, the basic assumption of the contextualist interpretation. Once we move the litigious term \(\beta_{I} Var(Z_{j} )\) from “group” to “individual” selection, the Pricean partition of the overall change in z into components due to individual and group selections collapses into the contextualist partition because, as it can easily be shown, for cases in which individual and contextual characters linearly affect fitnesses, \(\frac{{E(Cov_{j} (z_{ij} ,w_{ij} ))}}{{\bar{w}}} + \frac{{\beta_{I} Var(Z_{j} )}}{{\bar{w}}} = \frac{{\beta_{I} Var(z_{i} )}}{{\bar{w}}}\). This equation also holds for cases of single-level selection with accidental distribution of types across environments of different quality, like Brandon’s Drift-Case from “Adopting and rejecting EHR in a single-level selection scenario” section (of course, in such cases the subscript j will refer to the different environments of the case—and not to groups—, and \(Z_{j}\) will thus denote the average value of the phenotypic character in the jth environment).
Sober (2015, 839)—followed by McLoone (2015)—suggests that the wider version of the objection “may beg the question”, but he does not bring further arguments in support of this claim. In Jeler (2016), I contested Sober’s point, but on grounds that become questionable once we bring EHR into the discussion.
References
Abrams M (2014) Environmental grain, organism fitness and type fitness. In: Barker G, Desjardin E, Pearce T (eds) Entangled life. Organism and environment in the biological and social sciences. Springer, Dordrecht, pp 127–151
Antonovics J, Ellstrand NC, Brandon R (1988) Genetic variation and environmental variation: expectations and experiments. In: Gottlieb LD, Jain SK (eds) Plant evolutionary biology. Chapman and Hall, London, pp 275–303
Bourrat P (2015) Distinguishing natural selection from other evolutionary processes in the evolution of altruism. Biol Theory 10:311–321
Brandon RN (1990) Adaptation and environment. Princeton University Press, Princeton
Damuth J (1985) Selection among ‘species’: a formulation in terms of natural functional units. Evolution 39:1132–1146
Damuth J, Heisler L (1988) Alternative formulations of multilevel selection. Biol Philos 3:407–430
Earnshaw E (2015) Group selection and contextual analysis. Synthese 192:305–316
Frank SA (2012) Natural selection. IV. The Price equation. J Evol Biol 25:1002–1019
Gardner A (2015a) The genetical theory of multilevel selection. J Evol Biol 28:305–319
Gardner A (2015b) More on the genetical theory of multilevel selection. J Evol Biol 28:1747–1751
Glymour B (2008) Correlated interaction and group selection. Br J Philos Sci 59:835–855
Glymour B (2014) Adaptation, adaptation to, and interactive causes. In: Barker G, Desjardin E, Pearce T (eds) Entangled life. Organism and environment in the biological and social sciences. Springer, Dordrecht, pp 105–126
Godfrey-Smith P (2009) Darwinian populations and natural selection. Oxford University Press, Oxford
Goodnight CJ (2015) Multilevel selection theory and evidence: a critique of Gardner, 2015. J Evol Biol 28:1734–1746
Goodnight CJ, Schwartz JM, Stevens L (1992) Contextual analysis of models of group selection, soft selection, hard selection and the evolution of altruism. Am Nat 140:743–761
Hamilton WD (1975) Innate social aptitudes of man: an approach from evolutionary genetics. In: Fox R (ed) Biosocial anthropology. Wiley, New York, pp 133–155
Heisler IL, Damuth J (1987) A method for analyzing selection in hierarchically structured populations. Am Nat 130:582–602
Jeler C (2015) Is there such a thing as “group selection” in the contextual analysis framework? Hist Philos Life Sci 36:484–502
Jeler C (2016) Do we need a new account of group selection? a reply to McLoone. Biol Theory 11:57–68
Kerr B (2009) Theoretical and experimental approaches to the evolution of altruism and the levels of selection. In: Garland T Jr, Rose MR (eds) Experimental evolution. Concepts, methods, and applications of selection experiments. University of California Press, Berkeley, pp 585–630
Lande R, Arnold SJ (1983) The measurement of selection on correlated characters. Evolution 37:1210–1226
Matthen M, Ariew A (2009) Selection and causation. Philos Sci 76:201–224
Matthewson J (2015) Defining paradigm Darwinian populations. Philos Sci 82:178–197
McLoone B (2015) Some criticism of the contextual approach, and a few proposals. Biol Theory 10:116–124
Millstein R (2010) The concepts of population and metapopulation in evolutionary biology and ecology. In: Bell MA, Futuyama DJ, Eanes WF, Levinton S (eds) Evolution since Darwin: the first 150 years. Sinauer, Sunderland, pp 61–85
Nunney L (1985a) Group selection, altruism, and structured-deme models. Am Nat 126:212–230
Nunney L (1985b) Female-biased sex ratios: individual or group selection? Evolution 39:349–361
Okasha S (2004a) Multi-level selection, covariance and contextual analysis. Br J Philos Sci 55:481–504
Okasha S (2004b) Multilevel selection and the partitioning of covariance: a comparison of three approaches. Evolution 58:486–494
Okasha S (2005) Altruism, group selection and correlated interactions. Br J Philos Sci 56:703–725
Okasha S (2006) Evolution and the levels of selection. Oxford University Press, Oxford
Okasha S (2011) Reply to sober and waters. Philos Phenom Res 82:241–248
Pedhazur EJ (1997) Multiple regression in behavioral research. Explanation and prediction. Thomson Learning, Boston
Price G (1970) Selection and covariance. Nature 227:520–521
Price G (1972) Extension of covariance selection mathematics. Ann Hum Gen 35:485–490
Richardson R (1996) Critical notice: Robert Brandon, Adaptation and Environment. Philos Sci 63:122–136
Sober E (1984) The nature of selection: evolutionary theory in philosophical focus. MIT Press, Cambridge
Sober E (2011) Realism, conventionalism, and causal decomposition in units of selection: reflections on Samir Okasha’s Evolution and the Levels of Selection. Philos Phenom Res 82:221–231
Sober E (2015) Replies to commentators on Did Darwin Write the Origin Backwards?. Philos Stud 172:829–840
Sober E, Wilson DS (1998) Unto others: the evolution and psychology of unselfish behavior. Harvard University Press, Cambridge
Wade MJ (1985) Soft selection, hard selection, kin selection, and group selection. Am Nat 125:61–73
Wilson DS (1975) A theory of group selection. PNAS 72:143–146
Wilson DS (1979) Structured demes and trait-group variation. Am Nat 113:606–610
Acknowledgements
I am very grateful to Adrian Currie, Bruce Glymour and two anonymous reviewers for comments on earlier versions of the manuscript. This work was supported by a grant of the Romanian National Authority for Scientific Research and Innovation, CNCS—UEFISCDI, project number PN-II-RU-TE-2014-4-2653.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Jeler, C. Multi-level selection and the issue of environmental homogeneity. Biol Philos 32, 651–681 (2017). https://doi.org/10.1007/s10539-017-9578-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10539-017-9578-y