Understanding the evolutionary ecology of ethanol production may yield insights into why humans are prone to excessive consumption of ethanol. In particular, Dudley (2000) suggested that human ancestors developed a genetically based attraction to ethanol because they could use its odor plume to locate fruiting trees and because of health benefits from its consumption. If so, ethanol should be common in wild fruits and frugivores should prefer fruits with higher ethanol content. A literature review reveals that ethanol is indeed common in wild fruits but that it typically occurs in very low concentrations. Furthermore, frugivores strongly prefer ripe over rotting fruits, even though the latter may contain more ethanol. (Data on ethanol content of ripe and rotting wild fruit are lacking.) These results cast doubt on Dudley's hypothesis and raise the question of how humans became exposed to sufficiently high concentrations of ethanol to allow its excessive consumption. Because fermentation is an ancient and widespread practice, I suggest that humans “discovered” ethanol while using fermentation as a food preservation technique. They may have been predisposed to consume ethanol from previous and beneficial exposure to much lower doses or they may have become addicted to it at high concentrations because of fortuitous physiological responses. Explanation of alcoholism.
Excessive consumption of alcohol is a major threat to human health in many societies. Research on alcoholism generally focuses on physiological and psychological bases of addiction. This mechanistic approach, which characterizes biomedical research in general, is based on common sense: understanding how the body responds to ethanol consumption allows development of treatments that target key vulnerabilities or that take advantage of existing detoxification mechanisms. Yet despite its success and future potential, this approach is handcuffed.
The handcuffs result from a general failure to incorporate evolutionary theory. Alcoholism did not arise de novo and humans are not unique in their exposure to ethanol, the most common type of naturally produced and consumed alcohol. Indeed, an evolutionary basis for susceptibility to alcoholism is suggested by its heritability and by biogeographic patterns of alcohol dehydrogenase (ADH) occurrence ( Goldman, 1993 , Shen et al., 1997 ). As argued by proponents of Darwinian medicine, consideration of both proximate (mechanistic) and ultimate (evolutionary) factors holds great promise for providing new insights about the origins and treatments of human diseases, including alcoholism ( Nesse and Williams, 1994 , Stearns, 1999 ).
This review addresses a recent evolutionary hypothesis for the evolution of alcoholism. Dudley (2000, 2002 ) posits that the highly frugivorous diet of hominids and their closest relatives chronically exposed them to ethanol within fruits. These primates may have keyed into ethanol because its odor plume can be used to find fruiting trees, and because its ingestion provides calories and can stimulate appetite and yield health benefits via hormetic processes. Just as an evolutionarily-based attraction to highly caloric foods by our ancestors may have predisposed humans to obesity and Type II diabetes in modern societies with an abundance of fat and sugar rich food ( Eaton and Eaton, 1999 ), the susceptibility to alcoholism may be viewed as the result of nutritional excess. A variation of this hypothesis was proposed by ( Singh, 1985 ).
I address three issues related to Dudley's hypothesis. First, I describe the evolutionary ecology of ethanol production in wild fruits. In particular: What accounts for the occurrence of ethanol? How common is it? And, what are typical ethanol concentrations encountered by frugivorous vertebrates? Second, I review the scanty literature on whether vertebrate frugivores prefer fruits with presumably higher levels of ethanol over conspecific fruits with presumably lower levels of ethanol. The goal is to evaluate a central prediction of Dudley's hypothesis: ethanol-rich fruits should be preferred over those with little or no ethanol. My third objective is to propose a modification of the hypothesis. Specifically, I will argue that frequent consumption of ethanol-rich fruits is not a prerequisite for excessive ethanol consumption by humans.
Natural history of fruit rot
By far the most common natural source of ethanol is fermentation of fruit sugars by yeasts. Indeed, yeast is typically envisioned as an anaerobic organism and credited for the following reaction, upon which the beer and wine industries are based: Ecologically, however, yeasts are far more diverse, only a small subset of species specialize on fruits and of those that do, most can ferment carbon-rich substrates other than glucose (e.g., polysaccharides and lipids, Phaff et al., 1966 ). None is reliant on anaerobic metabolism alone and many can shift between fermentative and respiratory (aerobic) pathways, depending on conditions ( Cooke, 1979 ). In aerobic environments, ethanol is oxidized by yeasts or Acetobacter bacteria, often resulting in production of acetic acid (vinegar). Both ethanol and acetic acid can inhibit further growth of yeasts ( Skinner et al., 1980 ). The important point is that ethanol production is self-limiting and occurs only in anaerobic environments. Functionally, this suggests that small fruits, which have relatively high surface-to-volume ratios, will be less likely to harbor large colonies of ethanol-producing yeasts than large fruits, which have a relatively high proportion of pulp shielded from aerobic conditions. Small fruits are additionally unlikely to be an important ethanol source because any ethanol they may contain is more prone to evaporative loss than is the case in larger fruits, which have relatively smaller surface areas. In a similar vein, ethanol content is likely to differ between lipid-rich fruits and sugar-rich fruits, being higher in the latter because simple sugars are more readily fermented than lipids. Unfortunately, predictors of ethanol content in ripe wild fruits remain unexplored.
How common are fruit rot fungi? Tang et al. (2003) report a total of 102 taxa on eighteen species of wild fruits in Hong Kong. In less thorough screenings of fungi from surface-sterilized ericaceous and solanaceous fruits, 14 and 10 taxa, respectively, were isolated from two North American sites ( Cipollini and Stiles, 1993 , Cipollini and Levey, 1997). Often more than one species or genus of yeast is present on a given fruit and different species may replace each other over time—i.e., there is a succession of species as the fruit ripens or rots ( Phaff et al., 1966 , Cooke, 1979 , Spencer and Spencer, 1997 ). Different species can also typify different types of fruits. For example, Candida is most common on lipid-rich fruits and Saccharomyces on sugar-rich fruits ( Skinner et al., 1980 ). In the context of this review, it is noteworthy that Saccharomyces species are abundant, widespread, and champion producers of ethanol, not coincidentally, brewer's yeast (S. cerevisiae) belongs to this genus.
Not all fungi associated with fruit cause rot. Some species are latent or endophytic, infecting flowers or immature fruits and persisting for long periods in an asymptomatic state ( Snowdon, 1990 , Johnson et al., 1992 ). Cipollini and Stiles (1993) hypothesize a mutualistic interaction between endophytic fungi and fruiting plants, whereby the fungi benefit from nutrients within pulp and the plants benefit from the fungi inhibiting growth of other species that would quickly cause fruit rot or be otherwise pathogenic. Because many species of Saccharomyces are endophytic, their presence in fruits (and the production of ethanol that accompanies it) should not be assumed to be detrimental to the fruiting plant.
Although it is tempting to assume that ethanol is common because yeast spores and fruits are omnipresent, it is important to keep in mind that not all yeasts produce ethanol and those that do produce it only under restrictive (anaerobic) conditions. Furthermore, fruits in the temperate zone are typically removed by frugivorous vertebrates before they become damaged or show signs of rot ( Jordano, 2000 , McCarty et al., 2002 ). Of fruits that are not removed, many are impervious to rot because they are protected by secondary compounds, desiccate, or occur when ambient temperatures are too low for microbial growth (Cipollini and Levey, 1997, Jordano, 2000 , McCarty et al., 2002 ). In the tropics, where fruit diversity is higher, fruit abundance typically less seasonal, and abiotic conditions generally more favorable for microbial growth, ethanol production in fruits may be more common than at higher latitudes.
Definition of alcoholism by who
It takes three to tango: evolutionary ecology of fruits and their consumers
A fundamental difference between plants and animals is that animals are mobile and plants are not. Fruits are an evolutionary response to this lack of mobility in plants, they provide a mechanism by which plants disperse seeds to new sites. Frugivorous vertebrates ingest fruits and the seeds within, later defecating the seeds. Frugivores gain nutrients from the fruit pulp, while plants gain various benefits of seed dispersal. Fitness advantages to both participants have been so strong and ubiquitous that mechanisms of seed dispersal by vertebrates have evolved in almost all major lineages of extant seed plants and, conversely, reliance on fruit is evident in all major lineages of vertebrates ( Herrera, 2002 , Labandeira, 2002 ).
Because the interaction between fruiting plants and frugivorous vertebrates described above is so visibly apparent, it has dominated the research agenda of those interested in the chemical composition of fruits. Unfortunately, it is a narrow and restrictive view of fruit-frugivore interactions. A central theme of this review is that fruit-frugivore interactions should be viewed as a triad between fruiting plants, vertebrate frugivores, and invertebrate/microbial frugivores ( Janzen, 1977 , Herrera, 1982 , Cipollini and Levey, 1997) ( Fig. 1 ). Ironically, the most abundant frugivores, fungi, are those that are most rarely considered in studies of frugivory. This oversight is especially pertinent, given that fungi are the primary source of ethanol.
We have already considered the interaction between fruiting plants and vertebrate frugivores. To place this in the context of ethanol production requires consideration of frugivory from a microbial perspective. I now summarize the remaining two sides of the triad: interactions between microbes and fruiting plants and between microbes and vertebrate frugivores.
Unlike the mutualistic interaction between fruiting plants and vertebrate frugivores, the interaction between plants and frugivorous microbes is predatory. Not only do microbial frugivores consume fruits and fail to disperse seeds, but they often render fruits unattractive to vertebrate frugivores that do disperse seeds. Thus, fruiting plants face an evolutionary dilemma: how to make their fruits simultaneously attractive to vertebrate frugivores and unattractive to microbial frugivores. A hypothesized means of solving this dilemma is “directed deterrence,” whereby plants produce compounds in fruits that negatively affect microbial growth and have little effect on vertebrates ( Janzen, 1977 , Herrera, 1982 , Cipollini and Levey, 1997). Directed deterrence provides an adaptive explanation for why all wild fruits contain a rich array of secondary compounds—why they are so much more than packages of calories and proteins for vertebrate dispersers. Ethanol, however, should not be viewed in this framework because it is not produced by the fruiting plant.
The interaction between vertebrate and microbial frugivores (hereafter, fungi) is largely unexplored and more complex than generally appreciated. The two types of frugivores are best viewed as competitors of an ephemeral resource. Vertebrate frugivores consume many fruits and are capable of extracting nutrients from a given fruit in a matter of minutes, whereas fungal frugivores are each restricted to a single fruit and require days or weeks to profit via sporulation. This asymmetry suggests that fruits are a dangerous habitat for fungi, they may be consumed at any moment by a frugivorous vertebrate ( Janzen, 1977 , Herrera, 1989 ). Because the stakes are high for fruit fungi (reproduction vs. death), fungi should defend themselves and “their” fruit as soon as possible after colonizing the fruit ( Janzen, 1977 ). The result is fruit rot. Its efficacy as a defensive mechanism for fruit rot fungi is as apparent as rotten fruit is repulsive.
Three lessons relevant to ethanol production emerge from consideration of the evolutionary triad ( Fig. 1 ). First, if yeast or other frugivorous fungi “win the battle” for fruit pulp, both fruiting plants and frugivorous vertebrates “lose.” All else equal, dual selection pressure by plants and vertebrates against frugivorous fungi and for each other, means that ethanol production and other processes associated with fruit rot should be rare. Second, frugivorous vertebrates are consistently exposed to secondary metabolites from both fruiting plants and frugivorous fungi. Vertebrate frugivores should thus be physiologically adept at detoxifying such compounds, including ethanol. Third, although ethanol is an end product of fermentation, the fungi that produce it are locked in a complex set of interactions with fruiting plants, frugivorous vertebrates, and other microbes. Given that ethanol affects both vertebrates and microbes ( Janzen, 1977 ), it is likely to have at least some adaptive basis. In particular, it may be viewed as a defensive agent, used by yeasts to inhibit growth of competing microbes in much the same way as penicillin is thought to give Penicillium fungi the upper hand in competition with bacteria.
Ethanol concentrations in wild fruits
I am aware of only four studies of ethanol content in wild fruits: Eriksson and Nummi (1982), Dudley (2002), Dominy (2004), and Sanchez et al. (2004). These studies provide data from across a fairly wide range of fruit types: 18 species in 11 families. Ethanol concentrations range from 0.04 to 0.72%. One way to view these data is that ethanol is ubiquitous in ripe fruit, frugivorous vertebrates cannot avoid ethanol ingestion.
Another way to view the same data is that ethanol concentrations are extremely low by anthropomorphic standards, 1–2 orders of magnitude below typical concentrations in beer and wine (3–6% and 8–11%, respectively). Such comparisons are potentially misleading, though. An animal's physiological exposure to ethanol is not only a function of the ethanol concentration of its food, but also of how much it consumes and the dynamics of ethanol absorption and metabolism. Frugivorous birds and bats have particularly high rates of fruit consumption, tied to similarly high fluxes of digesta through the gut ( Witmer and Van Soest, 1998 , Levey and Martínez del Rio, 2001 ). They presumably absorb more ethanol than do less frugivorous species of the same body size because their net absorption rates are higher. Balancing this, at least in frugivorous birds, are unusually high activities of alcohol dehydrogenase ( Eriksson and Nummi, 1982 , Prinzinger and Hakimi, 1996 ). Species that are more highly frugivorous tend to have higher rates of alcohol metabolism than do less frugivorous species ( Eriksson and Nummi, 1982 , Prinzinger and Hakimi, 1996 ).
We have seen that ethanol is common in ripe fruits and that it is rapidly consumed and metabolized by frugivorous vertebrates. Frequent ingestion of low ethanol concentrations likely typifies wild frugivores. At the opposite extreme are rare but consistent accounts of birds and mammals consuming fermented fruits and becoming “drunk” ( Siegel and Brodie, 1984 , Fitzgerald et al., 1990 ). Although such accounts still stir debate about whether the animals' uncoordinated behavior is due to ethanol or to other compound(s) ( Prinzinger and Hakimi, 1996 ), it is safe to conclude that frugivores occasionally consume highly fermented fruits. In the temperate zone, this tends to happen in the winter or early spring, when few other resources are available and fruit cells have ruptured due to freeze-thaw cycles, making them susceptible to fungal attack. Of more interest are situations in which frugivores have the opportunity to select among conspecific fruits with varying amounts of ethanol. This situation is undoubtedly common, probably every fruiting tree presents unripe, ripe, over-ripe, and rotting fruits, which almost certainly differ in ethanol content. In such cases, a key prediction from Dudley's (2000, 2002 ) hypothesis is that frugivores should prefer fruits with higher ethanol content.
Before summarizing the most relevant studies on vertebrate fruit preferences, a disclaimer is necessary. Because no study on fruit selection has considered— let alone measured—ethanol concentration in fruits, one is confronted with the awkward choice of dismissing the entire literature on fruit selection or making a key assumption about how ethanol levels change as fruits ripen. For the purposes of this review, I will assume that ethanol concentrations increase as fruits progress from ripe to over-ripe to rotten. (These stages of maturation are not discrete. Ripe fruits are typically intact, symmetrical, and evenly colored, whereas rotten fruits are characterized by broken skin, asymmetrical shape, and non-uniform coloration. Over-ripe fruits are intermediate.) Although this assumption seems intuitive—the more time yeast have to colonize and grow in a given fruit, the more ethanol will be produced— it may well be false because it does not consider catabolism of ethanol by bacteria or vaporization of ethanol. The only data that address the assumption are scant: in two species of fruits from Panama, Dudley (2002) found ethanol concentrations were higher in “very ripe” than “ripe” fruits.
When given a choice of conspecific fruits that vary in maturation and presumably ethanol content, what do frugivorous vertebrates prefer? Using artificial infructescences of solanaceous fruits placed in a Costa Rican cloud forest, Valburg (1992) found a strong preference for non-rotting over rotting fruits by both birds and bats. Likewise, Levey (1987 and unpublished data) monitored removal of Hamelia patens (Rubiaceae) fruits over an annual cycle and found an extremely strong preference for non-rotted fruits: 76% of unrotted ripe fruits were removed by vertebrates, compared to 0% of rotten fruits.
The studies by Valburg and Levey are unique in that they recorded preferences of wild frugivores for ripe and rotting fruits. Other studies on fruit choice by wild birds and mammals are less informative with respect to ethanol because they focus on preferences between intact fruits and insect-infested fruits that are not rotting in a way suggestive of ethanol production (e.g., Jordano, 1987 ). Nonetheless, these studies overwhelmingly conclude that frugivores prefer ripe, non-rotting fruits over damaged or rotting fruits.
Trials with captive frugivores generally mirror results from field studies: frugivores typically reject rotting fruit, regardless the degree of rot ( Buchholz and Levey, 1990 , Valberg, 1992, Cipollini and Stiles, 1993 ). In a particularly rigorous set of experiments, Cipollini and Stiles (1993) surface sterilized four species of ericaceous fruits and inoculated them individually with one of ten species of fruit-rot fungi. Rotted fruits were paired with conspecific non-rotted fruits and presented to frugivorous birds. Overall, non-rotted fruits were consumed at rates 2–3 times higher than rotted fruits, regardless of fruit species. Fungal species, however, had a major impact on the strength of preference for non-rotted fruits. Some species were “toxic,” severely reducing avian consumption of fruits they infected (e.g.,Alternaria alternata), whereas others had relatively little impact on avian fruit preference (e.g.,Botrytis cinerea). Of particular interest, Saccharomyces cerevisia (a major producer of ethanol), clearly belonged to the “non-toxic” group, suggesting that fruits most likely to contain ethanol are also the rotting fruits most likely to be consumed by birds.
To summarize, rotten fruits are typically discriminated against by frugivores, suggesting that wild frugivores do not prefer ethanol-rich fruits. The only means of adequately addressing frugivore preference for ethanol, however, is to conduct preference trials with experimentally manipulated concentrations of ethanol. Sanchez et al. (2004) provide the first such test. They found that Egyptian fruit bats using olfactory cues selected feeders that contained mango juice over feeders containing dilute solutions of ethanol (0.001% to 1%).
A caveat about interpreting preference trials is that a statistically significant preference for fruits with lower ethanol content does not preclude a biologically significant effect of ethanol consumption. Given that vertebrate frugivores typically have very high ingestion rates ( Levey and Martínez del Rio, 2001 ), that all frugivores tested thus far will consume at least some rotten fruits even when provided with non-rotted fruits (references above), and that ethanol is present in ripe but non-rotting fruits, it seems premature to conclude that a typical frugivore's daily rate of ethanol consumption is negligible. Nonetheless, the prediction from Dudley's (2000) hypothesis that frugivores should prefer ethanol-rich fruits is not generally supported if one accepts the assumption that ethanol content increases as fruits ripen, become over-ripe, and rot. Future studies should examine preferences for ripe versus over-ripe (not rotting) fruit and should experimentally manipulate ethanol concentrations in the range typical of ripe and over-ripe fruit.
HUMANS, ETHANOL, AND ALCOHOLISM
On first appearance, the literature reviewed above does not provide strong support Dudley's hypothesis. Ethanol is ubiquitous in fruits but occurs in relatively low concentrations. And, in situations where its concentrations are presumably the highest (rotting fruits), frugivores strongly prefer other fruits, they do not appear naturally attracted to ethanol (see also Dominy, 2004 , Sanchez et al., 2004 ). Assuming that human ancestors were similar to extant frugivores in their consumption of ethanol, a missing part of the hypothesis becomes apparent: What is a plausible intermediate stage in the evolution of alcoholism? How does one progress from low-level consumption of ethanol in wild fruits to full-fledged alcoholism?
Definition of chronic alcoholism
Anthropologists have long recognized the importance of fermented foods in human nutrition. Practically every culture has developed unique and often elaborate processes of fermentation ( Battcock and Azam-Ali, 1998 ). For example, in the Andean region of South America chicha is traditionally made from pulverized corn ( Steinkraus, 1995 ). Corn meal is prepared by groups of older women, who thoroughly chew it to produce an intermediate product, “muko.” Amylase in their saliva converts starch to sugar in muko, greatly increasing yeast activity and ethanol production in the fermentation process that follows.
In an anthropological context, fermentation can be viewed as controlled spoilage of food. Anaerobic conditions prevent the succession of microbes that would otherwise completely oxidize the foodstuff. The microbes responsible for the later stages of food spoilage generally cannot grow in alcoholic or acidic environments. Thus, by culturing the production of alcohols and in many cases organic acids via limited exposure to oxygen, the food is protected. Long before refrigeration and synthetic additives, fermentation was one of the most important food preservation technologies ( Battcock and Azam-Ali, 1998 ).
Given the ancient and cosmopolitan reliance on fermented foods and the cultural inheritance of their use, I suggest that humans may not have developed their current attraction to ethanol via exposure to wild fruits. Rather, they first encountered high concentrations of ethanol through fermentation processes that were initially fostered for the purpose of food preservation. As they discovered the inebriating qualities of some fermented foods, they focused attention on those fermentative processes, ultimately leading to the beer and wine industries of today.
When viewed in this context, the questions that structured much of this review can be seen as largely irrelevant. It does not matter that ethanol concentrations in wild fruits are typically very low or that wild frugivores seem to avoid fruits with the highest ethanol content. Neither condition is necessary for exposure of humans to physiologically meaningful amounts of ethanol. On one hand, Dudley's hypothesis dodges a bullet—unsupported predictions are not damning. On the other hand, an alternative but less satisfying explanation emerges for alcoholism in humans. Addiction to ethanol may be analogous to addiction to caffeine, nicotine, heroin, or cocaine. All are secondary metabolites that humans have learned to concentrate and that provide a desired physiological response. The response is fortuitous, not adaptive. The principal difference between ethanol and most other addictive substances is that ethanol is produced by fungi, not by plants. However, the evolutionary bases are similar—the adaptive function of such compounds is likely rooted in competitive interactions with other organisms.
I thank Robert Dudley and Michael Dickinson for organizing the symposium, “In Vino Veritas: The Comparative Biology of Ethanol Consumption,” and in particular Robert Dudley for encouraging me to ponder the relevance of the evolutionary triad to alcoholism. Thanks also to participants in the symposium for spirited discussion and to the National Science Foundation for funding.