Undoubtedly, multidisciplinary approaches are required to explain differential susceptibility to obesity, but this article suffers from lack of reference to several key literatures, and I find the conclusions seriously flawed. In my own work, I have proposed and tested several adaptive hypotheses relating to adiposity (Wells Reference Wells2006; Reference Wells2009; Reference Wells2010a; Reference Wells2011a; Reference Wells2012a; Reference Wells2012d; Reference Wells2012e; Reference Wells2012f). Paleo-climate data indicate that hominins were exposed to growing ecological stochasticity in comparison with non-human apes and evolved many components of physiological flexibility in response (Wells Reference Wells2012d), of which adipose tissue was particularly important. There is little evidence that starvation was the primary selective pressure; rather there are several “fitness functions” of adipose tissue (Wells Reference Wells2010a) broadly connected to tolerating short-term or seasonal changes in energy balance. I have provided empirical support for body composition variability enabling adaptation to diverse stresses including climate, pathogen burden, and seasonality, while buffering health, growth, and reproduction (e.g., Wells Reference Wells2012c; Reference Wells2012g; Wells & Cortina-Borja Reference Wells and Cortina-Borja2013; Wells et al. Reference Wells, Griffin and Treleaven2010). Crucially, I restricted these analyses to populations in non-Western settings, and I have repeatedly emphasized that the adaptive biology of adipose tissue must be clearly differentiated from the pathological scenario of obesity and the impact of obesogenic environments (Wells Reference Wells2010a; Reference Wells2012e; Reference Wells2012f). Nettle et al. are aware of this literature, having cited it elsewhere (Nettle et al. Reference Nettle, Frankenhuis and Rickard2013), but have adopted the opposite approach: proposing that adaptive models explain obesity itself.

Nettle et al.’s approach posits that psychological mechanisms underlie “decisions” about “how much to eat,” though they do not specify whether this involves conscious deliberation or not. They then argue that it is adaptive to assess food security (again, without specifying how) and select an appropriate level of fat-insurance to acquire. Regardless of how these cost-benefit decisions are made, their “appropriate level” model explicitly assumes that individuals must decide both when to acquire fat and when to stop acquiring it, as their model proposes rising fitness costs. But the fundamental problem of obesity, well known to all in the field, is that fat individuals become even fatter, and far from supporting their adaptive arguments, their meta-analysis directly contradicts them.

According to adaptive principles, food insecurity should predict the greatest weight gain in thin individuals. This is very evident in the bird literature they reviewed – for example, the experimental study of greenfinches where food insecurity promoted weight gain inversely in association with baseline weight (Ekman & Hake Reference Ekman and Hake1990). Paradoxically, however, food insecurity showed significantly stronger associations with excess weight in the meta-analysis for obesity than for overweight. From an allometric perspective, birds have adequate fat stores to buffer only short periods of starvation, whereas men and women of average U.K. height with a body mass index (BMI) of 30 kg/m² have sufficient energy reserves to endure total starvation for ~75 and ~88 days and to endure half-rations for 150 and 175 days, respectively. Larger individuals with BMI of 40 kg/m² could tolerate half-rations for over nine months simply by oxidizing fat reserves (though in practice for even longer, as lean mass also declines during extended weight loss). That food insecurity should predict excess weight more strongly among those already with substantial energy stores than among those with lower reserves contradicts their adaptive model. Their “positive” findings thus derive from high-income populations analogous to “fat greenfinches,” whereas what we really need to know is whether food insecurity promotes weight gain in thin humans.

If their findings fit poorly with their energy insurance hypothesis for obesity, what other explanations are more likely? It is baffling that they make no reference to the neurobiology of appetite regulation; the role of key hormones such as insulin, leptin, and ghrelin; or the role of insulin resistance in driving hyperphagia among those obese. High circulating insulin levels in obese individuals make them highly susceptible to energy-dense diets, particularly those high in sucrose (Lustig Reference Lustig2006; Reference Lustig2008; Wells & Siervo Reference Wells and Siervo2011). This means that certain foods can themselves drive hunger and lethargy, helping explain why those who are already fat keep getting fatter (Lustig Reference Lustig2006; Reference Lustig2008; Wells & Siervo Reference Wells and Siervo2011).

A much simpler explanation for Nettle et al.’s finding, therefore, is reverse causation. Food-insecure individuals are also poorer, which shapes their dietary choices, and poorer groups are well established to have greater geographical exposure to junk food and other obesogenic factors (Block et al. Reference Block, Scribner and DeSalvo2004; Drewnowski & Specter Reference Drewnowski and Specter2004; Larson et al. Reference Larson, Story and Nelson2009). Indeed, I have argued that poor and food-insecure groups have the least agency to resist commercial interests, and that this lack of agency is itself promoted by corporate manipulation of dietary quality and food availability (Wells Reference Wells2016). This looks less like adaptive energy insurance acquisition and more like the interaction of perturbed metabolism with material deprivation and disempowerment – thus, from obesity to morbid obesity.

I have similar concerns over Nettle et al.’s perspective on the developmental origins of obesity. Their proposal that offspring acquire energy reserves in anticipation of future food insecurity reiterates the predictive adaptive response hypothesis (Gluckman & Hanson Reference Gluckman and Hanson2004), yet this has been extensively criticized specifically in relation to growth and metabolism (Wells Reference Wells2010b; Reference Wells2012b). Early growth patterns primarily predict later size and lean mass (Wells Reference Wells2011a; Reference Wells2011b) rather than adiposity, and associations of birth weight or infant weight gain with later obesity appear restricted to obesogenic settings (Wells et al. Reference Wells, Chomtho and Fewtrell2007). In Peru, early-life exposure to food insecurity following the 1998 El Nino event reduced childhood height and lean mass but had no effect on fat mass (Danysh et al. Reference Danysh, Gilman, Wells, Pan, Zaitchik, Gonzalez, Alvarez and Checkley2014).

Obesity is a serious public health issue because even when people strongly desire to lose weight, their metabolism overrides their intentions. This is most powerfully demonstrated by the effects of bariatric surgery: Long before any change has occurred in energy stores, profound alterations in hormone levels lead to reduced appetite and improved metabolic health (Rubino et al. Reference Rubino, Gagner, Gentileschi, Kini, Fukuyama, Feng and Diamond2004).

It would be interesting to test the food insecurity–energy insurance hypothesis in those of poorer nutritional status, but the authors have failed to provide evidence that it drives human obesity.