niche construction niche construction


The Reference Device Problem
Lewontin (1) drew attention to a problem within evolutionary biology by summarizing standard evolutionary theory as:

      1a     (1a)

     2a         (1b)

Evolutionary change, dO/dt, is assumed to depend on both organisms’ states, O, and environmental states, E(1a), but environmental change, dE/dt, is assumed to depend on environmental states only (1b). With many caveats and complications (2), organisms are not generally regarded as causing significant changes in their environments. The primary exceptions are when the selective environment is provided by other organisms, such as in the case of co-evolution or sexual selection.

For Lewontin, a more accurate general description of how evolution actually works is:

     1b            (2a)  

     2b              (2b)

where environmental change also depends on the environment-modifying activities of organisms. Philosopher Godfrey-Smith (3) drew attention to the same problem by describing standard evolutionary theory as 'externalist', by which he meant that it uses the external environment as its 'explanatory reference device'. It seeks to explain the internal properties of organisms, their adaptations, exclusively in terms of external properties, that is natural selection pressures, in their external environments.

The principal point that the conventional approach obscures is that organisms are active, as well as reactive (1,4). To stay alive organisms must extract resources from their external environments. They are compelled to choose and perturb specific components of their environments, and by doing so, they change some of the selection pressures in their environments. Lewontin’s equation (2b) introduces an additional causal arrow into evolutionary biology, which Odling-Smee labelled 'niche construction' (5).

The 'reference device problem', which Odling-Smee and collaborators (2) regard as the principle obstacle to progress within evolutionary biology, is that the causal arrow in (2) representing niche construction points in the 'wrong' direction, from organisms to environments, and is therefore not compatible with the externalist assumption of standard evolutionary theory. As a result, it is difficult or impossible for evolutionary biologists to describe changes in selection caused by niche-construction as evolutionarily causal. Instead, standard evolutionary theory is forced by its own explanatory reference device to 'explain away' all observed instances of niche construction as phenotypic, or extended phenotypic (6), consequences of prior natural selection. Standard evolutionary theory can recognise niche construction as a consequence of evolution, but it cannot recognise it as a cause.

The Devaluation of Proximate Causes

Responding to structuralist critics, Harvard evolutionary biologist and Synthesis architect Ernst Mayr wrote:Ernst MayrErnst Mayr

All of the directions, controls and constraints of the developmental machinery are laid down in the blueprint of the DNA genotype as instructions or potentialities. (7)

For Mayr, developmental processes cannot be regarded as independent causes of evolutionary events, since their characteristics, including their ability to control and constrain, are themselves fully explained by prior natural selection. If developmental processes direct evolutionary events, this is only the proximate manifestation of the ultimate cause of natural selection; conversely, those aspects of development that have not been shaped by selection play no evolutionary role.

Mayr was extremely influential in bringing this commonly made distinction between 'proximate' and 'ultimate' causes to prominence within biology. In an article published in 1961, Mayr argued that natural selection should be regarded as the ultimate cause of phenotypic characters, thereby effectively devaluing so-called proximate causes as explanatory tools within evolutionary biology. Since niche construction includes developmental processes, this stance also prevented evolutionary biologists’ from recognising niche construction as an evolutionary process in its own right (8, 9).

Instead, niche construction is perceived to have no independent evolutionary significance because, to the extent that it is evolutionarily consequential, it is fully explained by a preceding cause, namely preceding natural selection (10). Niche-construction effects are treated as merely extended phenotypes, and extended phenotypes play the same role in evolutionary biology as ordinary phenotypes, namely to affect the replication potential of the alleles contributing to those phenotypic effects (10). Similar reasoning underlies the treatment of niche construction as an indirect genetic effect, with, for example, a mother’s genes affecting her offspring’s phenotype by modifying its environment (11). While this stance recognizes that modification of the selective environment does occur, it does not view environmental modification as a process with quasi-independent causal significance. Rather it attributes all causal significance associated with niche-constructing activity to genes or genotypes, that are only present because of prior natural selection.

There are two major problems with this line of reasoning (8):

First, not all evolutionarily consequential niche construction, nor all development in general, is under genetic control. This is well illustrated by an example from human biology. Adult humans vary in their ability to consume dairy products without sickness as a result of physiological differences in the activity of the enzyme lactase, which relate to genetic variation (12). A strong correlation exists across cultures between the presence of the genes for lactose absorption and a history of dairy farming (13). This has led to the hypothesis that dairying constructed the selection pressures that led genes for lactose absorption to become common in pastoralist societies, which has received strong theoretical and empirical support (12,14,15).

Dairy farming

Dairy farming is apparently an instance of human cultural niche construction that is mediated by cultural processes. There are no genes for dairy farming (using 'genes for' in the sense of Dawkins) (16). Genes do not constitute the appropriate level of analysis to explain why individuals in some societies farm cattle and others do not—this is a cultural phenomenon. Yet in spite of the fact that dairy farming is not caused by genes and is not a biological adaptation, it has clearly had evolutionary consequences. As described in the niche construction and the human sciences pages, such examples are likely to be very common.


Second, even if it were the case that the niche-constructing activities of organisms were under genetic control, it would still not follow that niche construction was a mere effect of the prior selection of these controlling genes. For in many cases these 'controlling genes' have themselves been selected as a result of prior niche-constructed changes in selective environments. For example, it is tempting to assume that the 'ultimate' explanation for why earthworms modify soils is that prior selection has furnished them with genes for burrowing, tunnelling, exuding mucous, and so forth. However, it is no more than a convention within evolutionary biology that natural selection should be regarded as the ultimate cause of such phenotypic characters. From the perspective of a philosophical consideration of causation, one might equally argue that the ultimate cause of the selection pressures that favoured earthworm soil processing adaptations is the prior niche constructing activities of ancestral worms, without which there would be no soil environment to act as a source of natural selection (2). The niche construction perspctive rejects this kind of reasoning, replacing it with the notion of 'reciprocal causation' (see below).

Misleading Metaphors

Lewontin Evolutionary theory is often viewed as a fusion of Darwinian and Mendelian insights. Notwithstanding the immense progress that these achievements engendered, Lewontin argues that the legacies of Darwin and Mendel include misleading metaphors that:

LewontinRichard Lewontin
are responsible for certain difficulties in biology, difficulties that prevent us from some kinds of further progress and which keep us locked into a rigid framework of thought about the development and evolution of organisms (1, p. 273).

Mendel’s view of organisms as the manifestation of autonomous internal ‘factors’ with their own laws, germinated into a post-Synthesis metaphor in which ontogeny 'is seen as an unfolding of a form, already latent in the genes, requiring only an original triggering at fertilization and an environment adequate to allow ‘normal’ development to continue' (1, p. 276). Darwin’s view of organisms as passive objects moulded by the external force of natural selection encouraged a conception of evolution in which 'the environment "poses the problem"; the organisms posit "solutions", of which the best is finally "chosen"' (1, p. 276). The metaphor of selection, inspired by the efficacy of artificial selection, continues to encourage a view of organisms as passive objects upon which external forces act.

Lewontin argues forcefully that these metaphors are misleading. Similar points have been made by Oyama (17), Griffiths and Gray (18,19), and others.

The Niche Construction Revision
The niche construction perspective differs from the conventional perspective in recognizing two major adaptive processes in evolution, natural selection and niche construction. The perspective also recognizes two general forms of inheritance in evolution, genetic and ecological inheritance. There are two legacies that organisms inherit from their ancestors, genes and modified environments, incorporating modified selection pressures. Ecological inheritance is not a high-fidelity template copying system like genetic inheritance (10,20,21). Instead, organisms transmit to their offspring, and subsequent descendents, physically altered selective environments, both through actions on their biological and non-biological environments, and by their habitat choices.

The solution to the reference device problem was to change the explanatory reference device. Instead of describing the evolution of organisms relative to natural selection pressures in independent external environments, Odling-Smee et al. (2) describe evolution relative to the 'niches', or organism-environment interactive relationships. Because niches are defined by two-way interactions between organisms and their environments (22), this step allowed an 'interactionist' (3) theory of evolution to be substituted for the standard externalist account. The niche is a neutral explanatory reference device for evolutionary theory that can capture reciprocal causation without imposing any bias either in favour of natural selection and against niche construction, or vice versa. Conceptually it is only necessary to recognise that natural selection pressures are not properties of external environments, but properties of niches.

On the basis of this revised explanatory reference device, all developmental processes that modify the organism–environment relationship are potentially evolutionarily causal. That returns us to Mayr’s devaluation of proximate causation in evolution. Niche construction theory replaces both proximate and ultimate causation by 'reciprocal causation' and regards the characteristics of organisms as caused by interacting cascades of selection and construction, (2,8; see also 23) described elsewhere as 'cycles of contingency' (24).

The ultimate/proximate dichotomy, and the convention brought to prominence by Mayr, is clearly not the only way to think about causation in biology (9). The standard view, that niche construction is not a process in evolution because it is determined by prior natural selection, rests on an unstated assumption that the environmental source of prior natural selection is independent of organisms. However, one cannot assume that the ultimate cause of niche construction is prior natural selection by an independent environment if the state of that environment was itself partly caused by prior niche-constructing organisms.

Ultimately, such recursions regress back to the beginning of life. As niche construction is one of the defining features of life (2), there is probably no stage at which we could say natural selection preceded niche construction. From the beginning of life, all organisms have always, in part, modified their selective environments by niche construction, and their ability to do so has always, in part, been a consequence of natural selection. There have probably always been two reciprocal causes in evolution.

This recognition of reciprocal causation in evolution also goes a long way towards addressing Lewontin’s concerns regarding the misleading metaphors of unfolding developmental programs, and the externalism of natural selection. '[G]enes, organisms and environments are in reciprocal interaction with each other in such a way that each is both cause and effect in a quite complex, although perfectly analysable, way (1, p. 276). Hence, the organism is viewed as both a cause of its own development and a cause of its own selective environment.

From the niche-construction perspective 'genes, organisms and environments' are indeed: 'in reciprocal interaction with each other'. First, niche construction is not only itself a developmental process, but it is to some degree obligate. It follows that development must inevitably modify some selection pressures in environments. Second, because they can niche construct, developing organisms adapt to their environments not only by responding to selection pressures or inputs from their environments, but also by actively choosing and perturbing various components of their developmental environments. Third, niche construction is potentially informed by all the processes that inform development, including learning processes in animals, and cultural processes in humans. Thus, acquired characters play a (non-Lamarckian) role in evolution, by modifying natural selection. Fourth, niche-constructing organisms can introduce feedback in developmental processes, as well as in evolutionary processes. Some of the ways in which a developing organism alters its environment are likely to feed back to affect later stages of its own development (25,26). This idea was originally captured by Waddington with his concept of an 'exploitive system' (27).

All of these changes convert developmental processes from the mere unfolding of genetically guided programs to a process of active 'niche regulation'. Niche regulation requires developing organisms partly to respond to inputs from their developmental environments on the basis of their inherited genes, and partly to modify those same developmental environments by their niche-constructing outputs, based on genetically afforded (but not genetically determined) phenotypic plasticity. Hence, from the niche-construction perspective, the key task for any developing organism is the active regulation of its inherited ‘niche’, both by responding to its environment, and by altering its environment, in ways that keep its personal organism–environment relationship continuously adaptive, for the rest of its life.

1. Lewontin 1983 2. Odling-Smee, et al. 2003 3. Godfrey-Smith 1996 4. Waddington 1969
5. Odling-Smee 1988 6. Dawkins 1982 7. Mayr 1984, p. 126 8. Laland & Sterelny 2006 9. Laland, et al. 2011 10. Dawkins 2004 11. Wolf, et al. 1998 12. Durham 1991 13. Ulijaszek & Strickland 1993 14. Feldman & Cavalli-Sforza 1989 15. Holden & Mace 1997 16. Dawkins 1976 17. Oyama 1985 18. Griffiths & Gray 2001 19. Griffiths & Gray 2004 20. Sterelny 2001 21. Sterelny 2005 22. Chase & Leibold 2003
23. West-Eberhard 2003 24. Oyama, et al. 2003 25. Richardson 1999 26. Oyama, et al. 2001
Waddington 1959