Developmental Plasticity as Information Processing

Several different models of developmental plasticity as information processing have been proposed previously. A general approach was offered by Bateson (2001); in this “weather forecast” model, the developing organism receives cues of impending environmental conditions and selects an appropriate developmental trajectory accordingly. Since physiological plasticity cannot be maintained indefinitely, a specific strategy must be selected early in life during a critical window of development. An accurate weather forecast is assumed to enable an appropriate future strategy, whereas an inaccurate forecast results in the organism being poorly prepared for its long-term environment. Key questions arising from this perspective are, first, what specific cues about the environment are obtained and, second, what broader ecological parameters do those cues index?

This “forecasting” framework has been extended by Gluckman and Hanson (2004a, b) in the form of the “predictive adaptive response” (PAR) hypothesis. This model assumes that developing offspring receive cues during pregnancy about the state of the environment and use them to predict the adult environment in which reproduction is likely to take place (Gluckman and Hanson 2004a, b; Gluckman et al. 2007). For example, offspring experiencing famine in early life are assumed to prepare for persisting famine in adulthood through enhancing traits such as insulin resistance and central adiposity. The challenge of this approach for a long-lived species like humans is that predictions about the environment must be accurate for several decades in the future and remain accurate, given that reproduction does not even begin until late in the second decade after birth.

This PAR hypothesis has been extensively criticized on several grounds, all related to the idea that such long-term forecasting is implausible (Jones 2005; Bogin et al. 2007; Rickard and Lummaa 2007; Wells 2007a, 2010, 2012c). First, spectral analyses of simulated or historically stochastic environments fail to support the hypothesis that current or recent-past conditions can predict future conditions (Wells 2007a; Baig et al. 2011). Second, mortality is highest in human foragers in early life, raising questions as to how “long-term anticipatory adaptation” could develop in traits already strongly exposed to selection earlier in the life course (Wells 2007a). Third, empirical data often contradict the predictions of the PAR hypothesis (Wells 2012c): for example, Gambians under seasonal energy scarcity do not develop insulin resistance following low birth weight (Moore et al.

2001). An alternative “silver spoon” hypothesis predicts that offspring receiving more early-life investment have higher reproductive htness in all types of adult environments (Monaghan 2008). This hypothesis is supported in the comparative literature for a variety of vertebrate animal species, including humans (Monaghan 2008; Hayward et al. 2013).

An alternative “maternal capital hypothesis” emphasizes that the information processed by offspring during placental nutrition and lactation derives from the maternal phenotype, rather than directly from the external environment (Wells 2003, 2010, 2012c). Notably, human birth weight is only moderately depressed during maternal famine and only moderately increased following maternal supplementation, indicating that maternal physiology buffers the fetus from shortterm fluctuations (Wells 2003). Exposure of the fetus to maternal phenotype, representing the cumulative effect of the nutritional environment experienced during development (Emanuel et al. 2004; Hypponen et al. 2004; Jasienska

2009), as well as any previous reproductive experience for the mother, means that “short-term fluctuations [are] smoothed out to provide a more reliable rating of environmental quality” (Wells 2003). In this approach, adaptation through developmental plasticity is considered to be not to long-term future conditions, but to “maternal capital” (Wells 2003, 2010, 2012c). This approach also emphasizes that offspring plasticity makes possible “maternal effects” that benefit maternal as well as offspring fitness (Wells 2003, 2007b).

Elements of both the PAR hypothesis and maternal buffering have been presented by Kuzawa (2005) in his model of “intergenerational inertia.” As with the maternal capital hypothesis, Kuzawa argued that maternal phenotype buffers the offspring from short-term ecological perturbations and provides a smoothed signal of ecological conditions deriving from matrilineal experience. As in the PAR hypothesis, however, Kuzawa assumes that this smoothed signal early in life aids the long-term prediction of ecological conditions, by providing “a ‘best guess’ of conditions likely to be experienced in the future” (Kuzawa and Bragg 2012).

These three models, therefore, while they all treat developmental plasticity as adaptive, have significant differences (Box 1). One difference involves the timescale

Box 1: Different Models of the Adaptive Nature of Developmental Plasticity


Authors and reference

Source of ecological cues

Target of adaptation

Maternal capital

Wells (2003, 2010)

Maternal phenotype, integrating effects of maternal development (and hence grandmaternal phenotype), and life history strategy

Matching early offspring growth to maternal nutritional resources

Predictive adaptive response

Gluckman and Hanson (2004a, b)

External environment

Matching adult phenotype to adult reproductive environment





Maternal and matrilineal phenotype

Matching adult phenotype to adult reproductive environment

We aimed to develop a simple mathematical model that enables the evaluation of different strategies by which offspring can obtain information early in life on environmental conditions relevant to fitness. We addressed three different types of ecological variability, as depicted schematically in Fig. 3.1. First, the environment may be subject to clear annual cycles, with peaks and troughs in ecological productivity. Second, the environment may be subject to systematic trends, such that ecological productivity may rise or fall over lengthy time periods, as might occur through larger climate trends. Third, the environment may be subject to irregular, unpredictable “extreme events,” which superimpose major perturbations on other, more consistent patterns. Each of these three types of variability can be detected in the segment of the climate record relevant to human evolution, as well as in recent decades. For example, India experiences annual climate cycles, local systematic trends in rainfall, and irregular ENSO events that provoke monsoon failure (Glantz 2001; Guhathakurta and Rajeevan 2007).

To operationalize this approach, we used rainfall data from India as an index of ecological productivity and considered the kind of information that could be extracted from this record and processed adaptively by humans early in life, during the period of greatest developmental plasticity. We first considered the conditions under which it pays offspring to process information at all; then, having demonstrated that information processing can indeed be adaptive, we considered how different kinds of ecological variability can be adaptively translated and processed.


Box 1 (continued)

A schematic diagram of variation in ecological conditions over time, illustrating three components of variability

Fig. 3.1 A schematic diagram of variation in ecological conditions over time, illustrating three components of variability: (a) regular cycles, such as seasonality, (b) long-term systematic trends such as climate change, and (c) extreme perturbations, such as El Nino- Southern Oscillation (ENSO) events

over which the information is acquired. Another difference concerns the stage of the life course at which the adaptive response is assumed to be targeted. To date, debate over these contrasting approaches has been conducted through verbal arguments, with little systematic testing of competing hypotheses.

< Prev   CONTENTS   Source   Next >