The implications of environmental epigenetics

1 Environmental determinants of health: environment as setting

The first, and dominant, set of geographical approaches seeks to understand the environmental determinants of health. Treating the environment as a setting and the body as a separate object influenced by this setting, the aim is to identify how places shape individual and collective health status. Included here is much of the research within medical geography that demonstrates correlative relations between space and disease and then endeavors to explain particular distributions or clusters. This may take the form of mapping disease data or using spatial epidemiology to relate disease patterns and particular spatial variables, and includes some EJ work that brings in measures or sources of pollutants and correlates them to demographic variables (Gatrell, 2002; Morello-Frosch et al., 2001; Pastor et al., 2002). Scholars in medical geography have noted that these studies make assumptions about the relevant social characteristics of particular spaces while also focusing on a limited number of independent variables with one disease as the dependent variable (Cummins et al., 2005; Curtis and Riva, 2010), and also that they occlude questions of whether the clustering of diseased bodies in space is an effect of place characteristics or the people who inhabit them (the so-called composition or context debate) (Cummins et al., 2007; Smyth, 2008). Others have called the field to task for failing to attend to the contextual specifics of place and embodied experience (treating bodies ‘like dots on a map’) (Kearns and Moon, 2002; Parr, 2002). Our concern is somewhat different: even a place-specific approach takes the environment that presumably causes disease as primarily social. The environment perhaps affects behaviors in ways that manifest biochemically, but does not itself interact materially with human bodies. Environmental justice scholarship is somewhat of an exception; because its objects of investigation are environmental exposures, it assumes a biochemical transmission between air, water, soil, or food and the body. Nevertheless, EJ studies that aim to correlate exposures with demographic and spatial variables continue to treat place as a setting, not an active object.

A somewhat contrasting approach has been to focus on therapeutic landscapes and spaces of care (Gesler, 1992; Gesler and Kearns, 2002; Kearns, 1993; Parr, 2003; Smyth, 2005; Williams, 2007). Early on, scholars argued that some places, i.e. natural settings, are in themselves conducive to health and well-being. More recently, attention has turned to multiple sorts of spaces of care and healing (e.g. clinics and hospitals) and, rather than claiming that place itself shapes health, the focus is how a place is experienced. It is not the setting alone that is causal – it takes human affect to experience it, which then transforms into well-being. Still, for the most part place continues to be a setting, even as the notion of setting is expanded to be experiential and relational. Studies of healing spaces of nature, which posit environment as natural as well as spatial, seem to be the exception. However, as with the EJ approaches cited above, nature is seen less as a set of active processes and more as a characteristic of a setting, such that nature is effectively conflated with place. Further, scholars in this vein do not speculate on the mechanisms by which people are made to feel better other than affectively – and this then raises the related question about what is happening inside bodies.

This brings us to bodies. Although not necessarily associated with health geography, much of the literature on embodiment within geography shares an emphasis on the phenomenological. Taking a discursive approach, earlier research on the body treated it as something that moves through or experiences space as a setting (see, for example, Bell and Valentine, 1997; Longhurst, 1997; Nast and Pile, 1998). While more recent approaches have taken up materiality, materiality here refers primarily to the experiential, not the biological. This is evident in reference to affective responses or viscerality – of touching up against others, occupying space, eating, and other embodied practices (Carolan, 2011; Colls, 2007; Hayes-Conroy and Hayes-Conroy, 2010; Slocum, 2008). Disability studies is the exception to some extent, in that the body is clearly shaped by developmental or biological processes, yet even here the significance is cast largely as discursive or phenomenological (Crooks and Chouinard, 2006; Moss and Dyck, 1996). Work closer to the human-environment tradition, such as recent EJ literature on embodiment, discusses people’s self reports on the experience of being polluted or poisoned, but does not explicitly engage the biophysical processes of that poisoning or pollution (Dillon, 2011). Like the environment, the body remains ontologized as a setting, albeit one that must signal its subjects to communicate their sense of health, pleasure, or disease. While biophysical processes are occasionally acknowledged, neither the mechanisms of that signaling nor the very conditions that give rise to affective responses are addressed. Oddly, then, efforts to deal with the materiality of the body tend to maintain a dualism between the felt body and the known body (see Mol and Law, 2004, on this point).

In short, in all of these spatial approaches to environment and health, ‘environment’ is taken to mean place, so that the environment remains largely inert. Even in EJ, which focuses on active chemicals, the environment is still largely equivalent to place – or a container – in that the question is one of proximity, for example of a community to a hazardous waste incinerator. By seeing space/place as causal, rather than interactive, spatial approaches in effect black-box the environment. As for the body, even though its health status is fundamentally at stake in these questions, much of this literature black-boxes the body, as well. Even when embodiment is taken up explicitly, it is in terms of experience, affect, and viscerality, but not biological function.

2 Political ecology of health

The two other bodies of scholarship we discuss address the environment and the body in different, more ecological ways than do the spatial approaches, opening up both black-boxes – but not simultaneously. The first, offering a different way of approaching the environment, can be found in the relatively small scholarship on political ecology of health (Harper, 2004; King, 2010; King and Crews, 2012; Mayer, 1996; Scott et al., 2012). This scholarship has its roots in disease ecology, which treats people and disease as part of complex ecosystems in order to map disease and its vectors in space. In keeping with the larger framework of political ecology, the focus in political ecology of health is on how human actions, and especially larger-scale political economic processes, change ecological processes in ways that create new health problems. Nature thus arrives as human-influenced ecosystems that shape how diseases emerge and spread; the focus tends to be tropical environments and disease and the narrative is characteristically one of disruption. Crucially for us, what political ecology brings to the table is a much-expanded notion of environment. While environment is still largely a place – a landscape – that has specific characteristics that may harm and/or promote health, environment in political ecology also refers to active processes that are simultaneously political and biophysical – hence socionatural.

However, what political ecology of health has been less good at is unpacking the body. Nature is addressed primarily at the landscape scale, and there is little attention to the nature of the human body, which remains largely black-boxed (cf. Guthman, 2011, 2012a). The environment affects the body, but what happens inside the body – the body as itself socionatural – is something political ecology has yet to address, either conceptually or empirically. In addition, the focus in political ecology of health has mostly been on environmental degradation, telling a declensionist tale of human health effects of human meddling in the natural environment. Within the larger political ecology literature, attention to both scientific narrative (as in ‘critical political ecology’) and socionatures challenges this declensionist narrative (Forsyth, 2003; Robbins, 2004; White, 2006). Such attention is important for this emerging field, because epigenetic mechanisms always change things, yet bodily difference cannot be considered a priori bad: it takes other criteria to adjudicate what is bad and what is just different.

3 Biosocieties: studies of biomedicine

A third body of literature, critical social studies of biomedicine, seems at once parallel to and yet the inverse of political ecology. This work on ‘biosocieties’ has focused on new biomedical and broader life science understandings of life, health, and bodies. Attention has centered on advances at the frontiers of biomedicine, such as genomics, individualized medicine, and other biotechnologies – that is, the widely noted ‘molecularization’ of life (Beck and Niewohner, 2006; Rabinow and Rose, 2006; Rose, 2007). Like political ecology, this scholarship sees scientific knowledge as inherently political, and in so doing it gives prominent attention to political economy (biocapital) and the new ways that populations are constituted, divided, and managed (biopolitics). Parallel to the arguments made by political ecologists about natural landscapes, scholarship of biosocieties draws from new knowledge in the life sciences to argue that nature – biological bodies – is not given, but is quite mutable: biology is something into which people can intervene. Bodies are not only material and experienced, but, like the landscapes of political ecology, they are socionatures in which the line between the biological and social is erased. In addition, this approach rejects declensionist thinking, so much so that it has been criticized for focusing too much on the ability to improve purposefully upon bodies through advances at the frontiers of biomedicine (Braun, 2007; Mansfield, 2012a; Roberts, 2009). Geographers have been at the forefront of focusing not just on biomedical improvement, but on the biopolitics and bioeconomies of public health knowledge and practice, particularly in the area of infectious disease control (Ali and Keil, 2008; Sparke and Anguelov, 2012).

However, in the same way that political ecology neglects the body, in this literature it is the environment that largely goes missing. Molecularization through biomedical intervention has captured attention, such that there is a somewhat surprising lack of scholarship on environmental pathways to bodily transformation (cf. Braun, 2007; Landecker, 2011; Niewohner, 2011; Shostak, 2003). When scholars have looked at wider processes, they have mainly addressed sociopolitical issues such as biological citizenship, biopolitics, and even animal-human relationships (Greenhough, 2010), but not the biophysical environment. An ecological approach would treat the mutable, biological body as being constituted not only through intentional intervention and management but also through interactions with the wider environment – whether that environment is space or nature. Thus, we suggest here that, while this scholarship is quite useful for expanding our notion of the biological body, it misses another important axis of change, which is not just the molecularization of life, but the environmentalization of the chemical molecule.

1 Epigenetics

Epigenetics is a relatively new science studying the mechanisms that affect how genes are expressed (rather than affecting the genes themselves). The term was first used in the 1940s to describe the ‘mechanisms necessary for the unfolding of the genetic programme for development’ (Holliday, 2006: 76). It was not until the 1980s that epigenetics emerged as a field with concrete mechanisms and evidence of cellular processes that activate and deactivate genes, with a focus on cell stability and stem cell differentiation (Holliday, 2006; Jablonka and Lamb, 2002). The central discovery is that genes are not expressed in isolation, but rather the cellular context shapes what genes will be expressed or suppressed, and how that will affect protein synthesis and, hence, phenotype (Crews and McLachlan, 2006). A crucial component of this is that these epigenetic markers can be heritable; changes in the cellular environment can persist and sometimes be passed from parent to offspring in ways that will affect their phenotypical development (Crews and McLachlan, 2006; Jablonka and Lamb, 2002). Epigenetic effects are stochastic, however, and therefore do not determine an outcome (Faulk and Dolinoy, 2011; Heijmans et al., 2009).

The first epigenetic mechanism to be recognized is DNA methylation, which occurs when combinations of carbon and hydrogen atoms attach themselves along the DNA (Francis, 2011; Jirtle and Skinner, 2007; Kuzawa and Sweet, 2009). A normal part of cellular differentiation, this deactivates specific genes, thereby affecting the development of the organism. Because the pattern of methylation is inherited when the cell divides, this represents an inheritance mechanism not based in changes in the DNA sequence. Resulting physiological and morphological changes will persist through the lifetime of the organism and, if germ cells are affected, will be passed to offspring as well. Although DNA methylation was proposed as a concept in 1969, it was a research paper by Robin Holliday in 1987 that associated DNA methylation with epigenetics, and he is associated with the explosion of use of ‘epigenetics’ in the 1990s (Jablonka and Lamb, 2002). Two other possible epigenetic mechanisms, histone modification and regulation by non-coding RNA, have also been identified, but they are apparently not as stable, and are not necessarily heritable (Bollati and Baccarelli, 2010; Dolinoy and Jirtle, 2008; Jablonka and Lamb, 2002; Szyf, 2007; Weaver, 2007).

While identifying these mechanisms in the cellular environment, another research aim has been to understand how these mechanisms link the social and biophysical environment to epigenetic processes. Of primary concern has been the role of the social environment in terms of nutrition and psychosocial stressors (often via maternal behavior), particularly during fetal development and early life (Faulk and Dolinoy, 2011; Landecker, 2011). For instance, researchers have found that genetically identical agouti mice differ in coat color and size when experimental groups are fed folate, a nutrient that spurs methylation (Dolinoy and Jirtle, 2008; Waterland, 2009). In humans, a highly cited epidemiological study involved tracking the descendants of victims of the Nazi-imposed Dutch famine, many of whom have higher BMIs and are more prone to diabetes, hypertension, and cardiovascular disease (Kuzawa and Sweet, 2009). Those exposed to famine in early embryonic stages had lower methylation of the IGF2 gene, which is a key factor in growth and development (Heijmans et al., 2009). In regard to psychosocial stressors, researchers have investigated the behavior of lab animals in terms of caring for offspring, and noted that nurturing behaviors can alter glucocortoroid receptors in ways that diminish stress and have long-term morphological effects (Weaver, 2007). Research on humans has tracked the relationship between familial environment, anti-social behaviors, and the presence of methylated genes (Tremblay, 2008). It has only been very recently that scientists have turned their attention to the biophysical environment, by investigating the role of xenobiotic chemicals in epigenetic processes. We will discuss these findings further below. Here it is worth noting that, while epigenetic pathways have been identified for toxic exposure as well as stress and nutrition, all three are ontologically and chemically distinct pathways to body transformation (Landecker, 2011).

Even as research on epigenetics proliferates, the range of thought regarding the character and significance of epigenetics still remains quite wide, varying along a number of axes: (1) the degree of stability of epigenetically induced changes (Bird, 2007); (2) the degree to which an epigenetic change is a disturbance or non-normal event (Jablonka and Lamb, 2002) versus a regular event in an open system that explains natural variation; (3) the range of influences – from the cellular environment to the organism’s social environment to the biophysical environment; (4) the strength of the epigenetic mechanisms, i.e. genomic activity; and (5) the breadth of phenomena which the field can explain, from very specific medical syndromes to the evolution of life itself. Definitions at the more stringent ends of these axes (i.e. unstable disturbances emanating from the cellular environment that only alter genetic expression at the edges in a small number of cases) have limited implications. But those at the other ends (stable changes within open systems at multiple scales that continually overwrite genetic codes) have the potential completely to discredit the programming metaphor for DNA and the related ontological separation between bodies and environments. For some, the environment has such a formidable influence on genes that ‘genomic activity is as much effect as cause during cellular differentiation, both normal and pathological’ (Francis, 2011: 159). As this quote indicates, lying across all of the axes is the normative question about whether epigenetic change is normal, pathological, or indifferent.

2 Environmental toxicology

Whereas epigenetics has only recently come to focus on the role of xenobiotic chemicals in shaping phenotype, by definition modern toxicology seeks to understand the adverse biological effects of chemicals. A key focus for toxicology has been to identify dose-response relationships, which identify the specific effects of exposure to specific doses of individual chemicals. The basic model is one in which exposure to chemicals is safe at low doses, while adverse effects are caused by abnormal, high-dose exposures. Toxicology has aimed to identify for each chemical the threshold, or dose below which exposures are considered safe (Hayes, 2001; Wargo, 2009). There is presumed to be a linear relationship between dose and effect: above the threshold, higher doses will lead to more pronounced effects (Calabrese and Baldwin, 2003). Until recently, this ‘dose-makes-the-poison’ model has been the underlying tenet of toxicological research (see below). Originally, the responses studied by toxicologists were defined in terms of acute toxicity: obvious impairment, illness, and death. Later study of carcinogenesis drew attention to new, synthetic chemicals and shifted the focus to delayed effects drawn out over time, but it did not fundamentally change the basic tenets of thresholds and dose-response relationships. Toxicologists still assumed linear or at least curvilinear relationships between dose and manifestation (Krimsky, 2000).

A series of tragedies and accidental discoveries from the 1950s onwards challenged these tenets. For example, accidental exposure to methylmercury in Minamata, Japan, caused not only acute symptoms associated with central nervous system poisoning, but also retardation and developmental deficits in children whose mothers were exposed when pregnant – even if those women showed no adverse symptoms themselves (Grandjean et al., 1994). Tragedies were also associated with DES and thalidomide, both pharmaceuticals purposefully given to pregnant women. Their offspring later suffered from gross malformation and severe reproductive disorders, many of which manifested only when the children were themselves adults (Langston, 2010; Steingraber, 1997). Another accidental discovery resulted from the presence of wildly proliferating cells in breast cancer assays before scientists introduced the activating estrogens to the cultures, which brought attention to the previously unknown biological activity of heretofore neutral substances such as BPA (Colborn et al., 1996).

By the 1990s, these tragedies and accidents provoked a major rethinking of toxicology. The findings undermined the notion that low doses are safe, that dose-response curves are linear, and that the critical effects are either acute toxicity or cancer in adults. Adverse effects were not only found at much lower doses than previously expected, but the effects themselves were neurological and reproductive changes that presented themselves – and persisted – well after the time of exposure. These discoveries also highlighted the timing of exposure, with particular attention to the period of fetal development. Low doses at a crucial moment could have greater effects than higher doses at different times (Gore et al., 2006). To make sense of many of these findings, scientists developed a controversial paradigm regarding ‘endocrine disrupting chemicals’ or EDCs. They theorized that certain xenobiotic chemicals could interfere with the action of bodily produced hormones by mimicking, enhancing, or inhibiting them. Since hormones signal cells to make proteins, and proteins are the building blocks of bodily function, endocrine-disruption could alter function or phenotype (Krimsky, 2000: 116). This paradigm thus catalyzed new ways of thinking about the wide-ranging, long-lasting effects of chemicals.

3 Environmental epigenetics: convergence and key insights at the intersection

The convergence of epigenetics and environmental toxicology in environmental epigenetics occurred only in the last decade, particularly as investigators searched for biological mechanisms that might explain the long-lasting effects of even low-dose exposure to some chemicals. In 2006, this was still a novel idea: ‘It is well known that EDCs, and very likely endogenous hormones, can act on a gene’s developmental mechanisms, altering phenotype expression. We are now seeing that the mechanism of these phenotypic changes is probably epigenetic’ (Crews and McLachlan, 2006: S4). To be sure, scholars had hinted at such connections between environment, phenotype, and cellular mechanisms of gene expression before this. A paper in 1979 by Holliday first suggested that epigenetic processes may be involved in cancer (Jablonka and Lamb, 2002). In Our Stolen Future, Colburn et al. (1996: 40) noted that the endocrine system affects development permanently – as least as much as genetic coding – since hormones can control the expression of inherited genes by, for example, silencing them. Papers in a 1997 special issue of Mutation Research point to the unexplained etiology of birth defects, only 20% of which were estimated to be caused by mutations, and suggested that epigenetic mechanisms might be at work (Bishop et al., 1997; Dellarco and Kimmel, 1997; Ferguson and Ford, 1997). But it was only by the late 2000s that environmental toxins were arrayed beside psychosocial and nutritional stress as one of the established sources of epigenetic change (Bollati and Baccarelli, 2010; Dolinoy and Jirtle, 2008; Thayer and Kuzawa, 2011).

Here, we summarize seven key findings and concerns raised by the new environmental epigenetics. Together, they challenge conventional explanations of dosages, inheritance, and the origins of disease, while also raising questions about normal versus pathological outcomes.

Environmental epigenetics challenges the standard toxicological understanding of dose-response relationships. Rather than a linear (or curvilinear) relationship, dose-responses to EDCs and some other xenobiotic chemicals are represented by non-monotonic curves, in which effects of low and moderate doses not only may be greater than those of high doses, but may even be opposite (Krimsky, 2000; Vogel, 2008). In addition, the mechanisms are highly complex and interactive. For example, an EDC can influence multiple targets, acting as agonists on one hormone receptor and antagonists on another; they can act at different dosages from one system to another, affect different tissues differently, and be acted upon as well as act on (Gore et al., 2006: S2). Because of these multiple ways of interacting with the body, these chemicals are not predictable, and their effects may be subtle. For example, at low doses, the neurotoxin methylmercury causes small changes in development, neurological response, and cognitive ability – changes subtle enough that even their existence has been the subject of scientific debate (Grandjean et al., 2010).

1 For the human-environment tradition

One of the most salient points to come out of these new scientific discoveries is the fundamental permeability of the body, a topic that has also been of recent interest to geographers (see Braun, 2008; Greenhough, 2012; Martin, 1998). In the case of psychosocial factors, epigenetic research suggests that bodies interact with the environment by themselves producing biologically active chemicals. In the case of xenobiotic chemicals, the environment enters bodies through multiple points of contact, including inhalation, ingestion, dermal contact, and the placenta. Insofar as the chemical environment comes directly into the body to remake it, we see that there is nothing about the body that forms a solid boundary – or threshold – between it and the external environment (Alaimo, 2010). As such, environmental epigenetic processes blur boundaries between what have often been seen as ontologically separate (if interacting) objects.

This interchange of environmental and bodily molecules suggests a transformation in what we mean by ‘nature’ and ‘nurture’ such that the lines between them are being erased, even obliterated. ‘Nature’ has largely been taken to mean that which is given – the unchangeable – and in recent decades has meant given in the genes. Clearly, epigenetics undermines this idea, showing that genes do not act alone, but always do their work in a specific biosociochemical environment that influences how those genes act. Yet, in erasing the idea of pre-given nature, epigenetics does not adopt ‘nurture’ in any simple way either. Certainly, there are ways in which epigenetics seems to imply new ways actively to intervene in (i.e. nurture) biochemical processes. This is so when epigenetic research is linked to the possibility of new medical interventions, and also when it is linked to the idea that managing individual lifestyle is a way to control epigenetic action. The more popular research on environmental epigenetics in particular imparts the idea that exposure can be limited through lifestyle changes, especially for the pregnant woman (Mansfield, 2012a, 2012b). It seems to us, however, that the larger lesson of this research is that most of these processes remain beyond the control of intentional human activity. The permeability of the body to ubiquitous and subtly acting chemicals should put to rest the idea of the body as a citadel that can be protected by intentional individual behaviors. In an epigenetic understanding of environments, both nature and nurture are abandoned for an understanding of the body as always changeable, influenced by biosociochemical pathways acting at multiple sites, from the cell nucleus (e.g. DNA methylation) to the global atmosphere (e.g. transport of chemicals).

Part of what is exciting about this notion of epigenetic permeability is that it brings health into abiding debates about nature-society dualisms, as captured in notions such as hybridity and socionatures, which have been of interest in nature-society geography over the past quarter-century (e.g. Bennett, 2009; Castree, 2002; FitzSimmons, 1989; Swyngedouw, 1999; Whatmore, 2002). It further contributes to this scholarship by describing some of the material mechanisms by which socionatural environmental bodies are produced. At the same time, this new understanding of permeability challenges some current geographical approaches to environments and bodies which treat the environment largely as a space or setting. Environmental epigenetics highlights the activity of natural, material processes, through which the supposedly external environment actively enters, shapes, and becomes part of the body. At the same time, epigenetics encourages us to open up the black-box of the body. The body is not a passive recipient of various exposures but is active in its own remaking: an exposure to an environmental molecule triggers a molecular hormonal response, which then triggers cellular processes that affect the structure and function of the body, which then responds anew to the environment in an ongoing, iterative relationship. These, then, are dynamic ecological interactions between the body and the environment that suggest a need to attend to biochemical flows rather than spatial containers, bodies as receptors, or bodies as mutable only via intentional human action. Through this perspective, affect alone is not enough to understand bodily materiality, and molecularization cannot be treated as separate from environmental flows (Braun, 2007, 2008). A political ecology of the body must therefore pay attention to ecological processes both within and around the body.

In addition, while the proliferation of environmental chemicals in the 20th century may have generated problems that this new science is trying to understand, there is nothing to suggest that the processes themselves are in any way new. Epigenetic processes presumably have been happening throughout the history of life, such that the environment, body, and genome have always been co-produced (Hird, 2009; Margulis, 1999). This therefore negates the idea that there is a perfect ‘natural’ state to which we could return, which means there is no basis to a priori posit epigenetic changes as bad, as a form of degradation. It is probable that some of the changes that we are currently seeing are more pathological, but some are likely just different, while others are adaptive and thus could be construed as healthy (Lock and Nguyen, 2010). Thus, as with other forms of environmental change, an important aim of a new political ecology of the body would be to evaluate what is bad and what is good for the body, for whom, and in what ways.

2 For the place space tradition

Several aspects of the new environmental epigenetics pose challenges to the place and space tradition of medical geography, as well as environmental justice scholarship. Not only geographers but others who rely on spatial epidemiology see exposure as a problem of space: people in places that ‘contain’ dangerous environmental toxins will be more susceptible to illnesses and other conditions associated with those chemicals. Spatial epidemiology necessarily reduces a very complex environment to a set of variables, which tend to be those that can easily be measured and tracked, and therefore tend to be static features of the environment rather than the substances that flow through it (Cummins et al., 2007; Curtis and Riva, 2010). Furthermore, to the degree that spatial epidemiology addresses toxins, studies are generally designed to test for one chemical and one effect (often self-reported), even though people are regularly exposed to a multiplicity of chemicals and it is difficult to isolate the separate effect of any one chemical (Krimsky, 2000; Steingraber, 1997; Wargo, 2009). If, however, the environment and the body are both biochemically active and dynamic, as the previous section just emphasized, methodologies that use a set of variables and see space as determinative are not adequate to the task.

A particular challenge to spatial epidemiology stems from discoveries in environmental epigenetics about the multiple ways in which time matters in terms of the effects that xenobiotic chemicals may have, when they may manifest, and for how long they may persist.

Methodologically, spatial epidemiology assumes a static relationship between space and population. Snapshots are taken to establish correlative relationships, and the existence of a spatial cluster of a disease or condition demonstrates an exposure. Yet, just as chemicals move in and out of bodies, they move in and out of particular spaces. People are mobile as well; those exposed elsewhere may reveal diseases not from those spaces, while those in those spaces might not be manifesting the disease. These sorts of challenges to multivariate studies have already been noted, with scholars calling for more attention to complexity (Curtis and Riva, 2010). Yet the added complexity often accounts for other spatial variables rather than temporality. With substantial, even intergenerational, lag times between exposure and manifestation, these relationships may be particularly difficult to detect. (That said, a major study in the Salinas Valley of California is doing just that by longitudinal tracking of children born to farmworkers.) That effects of exposures are heritable over multiple generations attenuates the spatial connection even more. In other words, methods and their respective data that rely on spatial clustering to prove causality become untenable in a context of high mobility of both chemicals and bodies over time.

Another challenge involves the plasticity of bodies and the issue of already-produced vulnerability. Recalling that, although the ubiquity and level of environmental exposure has certainly deepened in the chemical-industrial age, epigenetic theories suggest that human bodies have been always in flux and always historically constituted by their environments. Bodies already made vulnerable by other phenomena may be more susceptible to adverse effects, but those can also be attributed to other causes. As Nash (2006: 198) has written, the problems in identifying cancer clusters is in part because of the radical contingency of exposure, but also because studies often control for race, which itself is already entangled in access to medical care, long-term and cumulative exposures, and presence in certain environments. Epigenetic mechanisms may thus compound the difficulty in health geography of disaggregating what is an effect of space itself and what is an effect of the clustering of historically made bodies in space. This ‘context or composition’ question generally turns on whether health outcomes are best predicted by the attributes of places or the characteristics of people who happen to inhabit them by virtue of social practices that have produced race and class segregation (see Curtis and Riva, 2010; Smyth, 2008; Tunstall et al., 2004). While some have argued that composition and context are inseparable because bodies make places and places make bodies (Cummins et al., 2007; Tunstall et al., 2004), with epigenetic mechanisms the place comes into the body in ways that dissolve the distinction between composition and context altogether. Put differently, health disparities may be epigenetic outcomes of social relations rather than context or composition in any simple way.

Overall, environmental epigenetics suggests that health disparities are historical/temporal and not only spatial. This poses a fundamental challenge to methods that take place and space as causal and determinative. Space must therefore be de-privileged if geographers are going to engage with environmental epigenetics.