When does natural selection take place?

Abstract

Although many studies of form and function find a correlation between performance and adaptive specialization, others fail to discern such a tight link despite careful monitoring and observation. This inconsistency among studies raises the question of when, how often, and how effectively natural selection and the organism’s own activities operate to maintain or improve the adapted state. I suggest here that most organisms operate well within the limits of their capacities (safety factors) most of the time and that interactions and circumstances that cause natural selection and test the body’s limits come in discrete, intermittent events rather than as continuously present or chronic conditions. Everyday life without such events does not test performance limits and therefore does not usually result in natural selection. This perspective on selection as rare, intermittent testing by ecological agencies suggests that studies of selective processes and activity in the wild should focus on observing and measuring the intensity and frequency of selective events and responses, intense challenges stemming from agencies such as predators, competitors, mating-related rituals, and extreme weather.

Introduction

The fact that organisms are well adapted to their surroundings is a cornerstone of evolutionary biology. Nearly as universally accepted is the inference that natural selection, based on the differential representation of adaptive inherited traits over time, is responsible for adaptation, although agency—activity, behavior, resistance, and the effect organisms have on their environment—also contributes fundamentally to the good fit between living things and their circumstances (Diogo, 2017; Kull, 2014; Levis & Pfennig, 2019; Lister, 2014; Turner, 2007; Vermeij, 2013, 2019, 2023). A vast literature has accumulated on the causes (or agencies) and outcomes of natural selection, but much less attention has been devoted to questions about when, how often, and how effectively these agencies bring about selection or under which conditions selection reinforces the adaptive status quo or propels a population to a new adapted state. Causal links exist among adaptation, traits, functions, genes, agencies, and natural selection, but the operation of selection and activity in the lives of organisms remains elusive (Oudman & Piersma, 2018).

In an important and provocative paper on the lack of correspondence between body form and field-based average swimming behavior of tropical Indian Ocean reef fishes, Satterfield et al. (2023) have injected some much-needed reconsideration of how form and adaptive function are causally linked. These authors show that the everyday demands of life on reefs are adequately met by fishes regardless of whether the body was evidently specialized for maneuvering in tight spaces, fast swimming in more open habitats, or resting for long periods between bouts of movement. Their findings strongly imply that optimal design is not necessary for success under the conditions of routine existence and that natural selection might not operate under the circumstances of uneventful life. In other words, these fishes are operating far below the safety factors enabled by their construction.

Other studies have likewise failed to find a tight correlation between morphological specialization and measures of functional performance. R. M. Alexander (1991), for example, pointed out that swimming performance in fishes (measured as speed relative to body length) and jumping performance in insects (measured as distance relative to leg length) cannot reliably be predicted from the degree of body streamlining and leg length, respectively. Sustained maximum swimming speed in streamlined scombrid fishes (endothermic tuna and ectothermic mackerel) do not correlate with body temperature, as might have been expected (Sepulveda & Dickson, 2000). Average crawling speeds of marine gastropods widely overlap among species that differ in the degree of shell streamlining and aperture (and foot) size (Palmer, 1980).

Moreover, many organisms survive and reproduce despite incurring injury or occupying suboptimal habitats. This is indicated by abundant examples or repaired shell damage in slow gastropods and bivalves (Dietl & Alexander, 1998, 2005; R. R. Alexander & Dietl, 2005; Vermeij & Zipser, 1986a; Zipser & Vermeij, 1980), claw damage and loss in crabs (Juanes & Smith, 1995), injured horns in male beetles (McCullough, 2014), and imperfectly healed legs in wading birds (Reichert et al., 2017) among many other cases. Hermit crabs typically occupy gastropod shells that are too small or sometimes too large for effective protection or egg-laying capacity (Bertness, 1981). Giant pandas (Ailuropoda melanoleuca) are morphologically but not physiologically well adapted to a diet of bamboo (Nie et al., 2019). Mussels (species of Mytilus) and other marine animals grow faster and to a larger size below the tide line, but they survive and reproduce well in the intertidal zone (Dayton, 1973). Finally, almost every species shows a high incidence of parasitism that compromises but often does not prevent the survival and fecundity of individuals.

Functions, and the adaptations enabling them, operate under a wide variety of conditions, and as a rule are not finely tuned to particular circumstances (Janzen, 1985; Vermeij, 2013). Successful biological design is therefore flexible and sufficient rather than optimal (Dudley & Gans, 1991; Garland and Huey, 1987; Vermeij, 1987). Adaptive evolution, in fact, depends on the universal condition that there is always room for improvement (Darwin, 1859; Vermeij, 2004). Organisms succeed evolutionarily when they survive long enough and leave offspring.

And yet, adaptive specialization and high-performance levels do go hand in hand in a vast number of cases. The rate of burrowing into sand is higher among bivalves and gastropods with streamlined shells and a large foot than in burrowing species without features that either reduce resistance to digging or produce more power (Stanley, 1970; Vermeij & Zipser, 1986b). Wulff (2021) showed that Caribbean sponges have specialized defenses targeting particular sponge-feeding predators. Arms races between plants and herbivorous or pollinating insects, though often not species specific, demonstrate specialization and evolution by both parties (Becerra, 2015; Ehrlich & Raven, 1964). These examples could be multiplied by the thousands.

Selection and safety factors

How can it be that some studies demonstrate a close connection between form and functional performance whereas others do not? One possibility can be easily dismissed: flaws in study design. The measurements made in these studies and the inferences drawn from them are credible and reasonable. Instead, the apparent contradiction raises a more consequential issue: How does natural selection, together with the organism’s own targeted actions, yield adaptive function?

To answer this question, consider an alternative way of expressing the findings that form and function are often poorly matched. In the examples cited above, the organisms were operating well within the limits of the safety factors of their morphological and physiological capacities. Performance was measured as an average (Palmer, 1980; Satterfield et al., 2023) or as a maximum sustainable speed (Sepulveda & Dickson, 2000). In the case of the reef fishes studied by Satterfield et al. in the Indian Ocean, there were no filmed interactions between individual fish and predators, interactions that might have called for exceptional speed, maneuvering, or cryptic behavior (Satterfield, personal communication, 2023). Everyday life there therefore does not test or reflect functional limits. Measures of central tendency (mean, median, or mode) likewise fail to capture the performance levels of which individuals are capable during an occasional crisis. In other words, most of the time individuals are not subjected to the circumstances that might lead to natural selection. Adaptation is a universal (if imperfect) condition of life, but the activity and selective processes fundamental to it do not operate except during episodic (or discrete) events or circumstances.

What are these circumstances? Traits that confer high functional performance become critical when there is a threat to survival or to the ability to leave offspring. Locomotor performance, for example, might be tested during escape from a predator or, as in hummingbirds, during energetically intense courting flights (Wilcox & Clark, 2022). During snowstorms, house sparrows (Passer domesticus) are tested for their ability to locate and fight for food, placing a premium on traits and actions related to aggression (Johnston et al., 1972). Failure to procure food under less stressful weather conditions would be a temporary competitive setback, but would not normally result in death because the birds are operating well within their adaptive limits. Chronic parasitism or infection might interfere with effective resistance or response to challenges, but these challenges remain discrete. Tests of functional performance will be less frequent than the day-to-day or average conditions of life. They must be frequent and severe enough that individuals are exposed at least once during their lifetimes, but challenges cannot be so severe that exposure results in death or the inability to reproduce. Extremely rare and excessively traumatic events are by their nature unpredictable and therefore exceed the safety factors of the adapted state (Vermeij, 2008).

Most studies of safety factors in organisms have emphasized mechanical aspects of failure of part or all of the body. Classic examples are studies of the strength of vertebrate limb bones (R. M. Alexander, 1981, 1984; Brandwood et al., 1986), limpet shells (Lowell, 1985, 1987), and crab claws (Palmer et al., 1999), tree branches and other plant stems (Higham et al., 2021), and the resistance of stems of sapling trees and lianas to cavitation (van der Sande et al., 2019). Physiological safety factors were explored by Dillon et al. (2010), who showed that tropical land-dwelling species live closer to their thermal maxima than temperate ones. Except for Lowell’s work on temperate and tropical limpets, these studies generally did not link biomechanical performance and safety factors to natural selection and evolution, although Higham et al. (2021) did call for exploration of this link. Lowell (1985, 1987) showed, and Dillon et al. (2010) implied, that safety factors are higher when the environmental factors causing selection are more variable. Maximum temperatures in the tropics are both more common and less variable than those in cooler climates. Safety factors are themselves subject to adaptive evolution. A greater power budget (time and energy), enabled by a favorable thermal environment and abundant and accessible food, both result from and enable the evolution of greater capacities to deal with increasingly powerful and common challenges.

Although calculations of safety factors vary among studies, it is clear that some species operate far below their safety factors, whereas others live under conditions close to maximum performance levels. R. M. Alexander (1981, 1984), for example, reported safety factors of between 1.6 and greater than 5. Safety factors are higher for shell breakage in mollusks than for limb-bone injuries in vertebrates (Brandwood et al., 1986). Functional redundancy, as occurs among the limbs of vertebrates and arthropods, permits injury and subsequent healing even if competitiveness and the risk of predation are temporarily increased. Among mollusks, damage to the growing margin of the shells of bivalves entails greater risk than damage to the shell edge in gastropods because gastropods gain extra protection by being able to withdraw the vulnerable soft parts whereas bivalves mostly cannot (Vermeij, 1983). Safety factors must therefore be considered in the context of the entire organism, not only of individual parts.

Almost nothing is known about the distribution of the potency and frequency of selective agencies in the wild. A power-law distribution, with many common and relatively weak agencies greatly outnumbering rare and potent ones, has been demonstrated by Hoekstra et al. (2001) in their review of field studies of selection. The frequency and effectiveness of selective events should correlate with the frequency and power of interactions that call for performance levels close to the safety limits of the organism.

An example is provided by the Taiwanese rhinoceros beetle, Trypoxylus dichotomus. In this species, horns are used in male–male combat, with the longest horns providing a competitive advantage but also incurring the most frequent injury (McCullough, 2014). Sexual selection pushes males closer to the safety limits of their horns in part because the competitive stakes are high and because the agent of selection (another male, and perhaps a female observing the male) is predictably common and potent. Hoekstra et al. (2001), in fact, showed that the frequency and intensity of sexual selection over the short term is higher than of what they call viability selection.

Organisms must perform many, potentially conflicting functions well enough to survive and reproduce. Some functions, such as crawling in marine snails, may rarely be tested to their limits unless the gastropods in question have a well-developed running response to slow predators. In such cases, the primary defense (and test) may lie elsewhere. In the case of many gastropods defending against predators, this might entail chemical defense or various aspects of shell armor. In more active animals, such as some slow-swimming fishes, aggression might be the most vigorously tested function in defense. The important point is that only some of the many functions an organism must execute will be tested, or if all are tested, the frequency and intensity of selection on their performance likely differ among functions. A thick-shelled or narrow-apertured gastropod may be more frequently tested for its armor than for its speed, whereas for a fast-swimming fish or cephalopod, many functions will be subordinate to speed (Vermeij, 1982). A close correspondence between form and performance should therefore be expected primarily for those functions for which selection is more frequent and most intense.

Maximum power is reduced under cold, nutrient-poor, dysoxic, and mobility-impairing conditions. Under such conditions, the number of selective emergencies may also be reduced, weakening the intensity of selection and keeping maximum levels of performance low.

Several interesting possibilities might explain the results of the study by Satterfield et al. (2023). Given that many would-be predators of reef fishes have been severely depleted on most reefs—think sharks, for example—selection due to predators might be much weaker now than in the past. As a result, there could be much less premium on burst swimming when compared with routine movement. Data on average speed would be too insensitive to evaluate such a hypothesis. High-performance levels, together with a high correlation between form and function, might still apply in settings where predators remain abundant and powerful.

Another possibility is that, as in other studies, the obvious performance metric—speed or maneuverability in this case—is not the capacity that matters most to an organism faced with a threat. For example, movement of the body with as little associated water turbulence (and therefore noise detectable by enemies) as possible might be more important to survival for a swimming animal, and this could be achieved not just by a streamlined body moving at high speed, but also by a slow-moving swimmer. Stealth, potentially complemented by cryptic coloration or behavior might be the criterion being tested, whereas acrobatics and rapid escape might be the more crucial criterion in other species. Function and selection are highly context dependent and can be inferred only by the kind of careful observation and monitoring undertaken by Satterfield et al. (2023), Wulff (2021), and others.

Conclusions

My perspective on adaptive evolution follows from the premise that natural selection acts on phenotypes of whole bodies and that these whole bodies affect their selective environment (Vermeij, 2023). Selection in this view is essentially an ecological process, caused by identifiable agents (other organisms or agencies not directly attributed to organisms). Traits contributing to the whole-body phenotype are ultimately heritable, but there is a long and often indirect path from genes to an organism’s phenotype and agency (Oudman & Piersma, 2018; Vermeij, 2019).

Another important implication of my perspective on selection and safety factors is that selective events are discrete phenomena, not chronic conditions. This contention is indirectly supported by Hoekstra et al. (2001), who found that the frequency of directional selection in the wild decreases as the interval of time over which selection is measured increases. This results from the fact that as the time interval increases more time is included when no selection takes place, implying that selection is discontinuous or at least highly variable. An analogous situation occurs in the calculation of sediment accumulation rates. The longer the interval over which sedimentation rate is evaluated, the lower the rate will appear to be, because many intervals are included when no sedimentation takes place (Sadler, 1981). Like natural selection, sedimentation is highly episodic, in this case being markedly higher during floods, landslides, or other discrete events (see also Stallard, 1992).

In sum, there are many reasons to expect the results found by Satterfield et al. (2023) and other authors that average performance is often not correlated with morphology and its biomechanical properties. Correlations between form and function do exist, but they arise from intermittent threats and responses that test the limits of the body, not the average activity. Theories built on principles of optimality—the best possible performance given the structure and physiology of the organism—might tempt investigators looking for an appropriate null hypothesis against which to test alternatives (Seger & Stubblefield, 1996), but they are at odds with adaptive reality.

If this perspective on selection and adaptation gains greater acceptance, studies of selection in the wild should focus on careful observation, monitoring, and measurement of interactions and circumstances that test the performance limits of organisms. Safety factors, the duration and frequency of selective episodes, and the role of agency (or behavior) in establishing and maintaining the good fit between organism and environment must take center stage in future work on what Hutchinson (1965) famously called the ecological theater and the evolutionary play.

Data availability

There are no new data in this article.

Conflict of interest: The author declares no conflict of interest.

Acknowledgment

I thank Tracy Thomson for assistance with preparing and submitting the manuscript.

References