Paper #2i:
This one is about the evolution of mutualism in snakes (specifically a special kind of mutualism called "Pseudo-Reciprocity"). PR and Reciprocal Altruism really changed the way I view our social reality. I didn't think 3 years ago when I started reading & thinking about all this kind of stuff that I'd ever end up doing a Master's on it! Anyhow, this paper is under review for a journal called "Oecologia." Hope you like it:
A Den of Pit vipers
The pit viper Gloydius shedaoensis is found on the island of Shedao, in the Bohai Sea, off the north-eastern coastline of China (Li 1995). Shedao pit vipers are one of the few snake species worldwide to feed on birds from birth (Shine 1983). Pit viper venom is highly toxic and passerines die within seconds of a strike (Zhao et al. 1979, citation in Shine et al. 2002a). Pit vipers expend only a small portion of their venom reserves to realize a kill, thus many prey can be killed and ingested by pit vipers during prey migrations.
Twice a year, thousands of birds layover on the island of Shedao during their migration to and from Siberian breeding sites (Li 1995). The vast majority of the birds are passerines, a group of migratory songbirds that range in size from small finches to large ravens. During these migrations, birds are so abundant that pit vipers have the opportunity to kill far more than they need to survive to the next migration. As with most animals living in seasonal migratory zones, pit vipers experience two boom-bust cycles of prey abundance every year. During fasting periods between the migration feasts, pit vipers hibernate in subterranean burrows.
Three behaviors of the Shedao pit viper have been labeled "accidental altruism." First, despite the high density of Shedao pit vipers--approximately one snake per square meter--aggression is nonexistent. Pit vipers ignore each other's presence, even when the snakes are in physical contact on the same branch from which they ambush passerines (Shine et al. 2002b). Second, juveniles are one-tenth the size of adults. Consequently, they are gape-limited and can swallow only the smallest passerines. Regardless of their small size, juveniles strike at any prey, often killing passerines that are too large to ingest. Large passerine carcasses are abandoned by the juveniles, but not wasted, as they are quickly scavenged by nearby adult vipers. Third, adult pit vipers viciously attack and kill the sparrowhawk, Accipiter nisus, which preys on juvenile pit vipers, but not the adults (Shine et al. 2002a). Adults kill hawks at no immediate benefit to them as they do not ingest hawk carcasses.
Such "altruistic" behaviors do not fit easily into current models of social evolution. For example, kin selection cannot explain the Shedao pit viper's accidental altruism as littermates typically disperse long distances; thus, nearby pit vipers are unlikely to be siblings (Shine et al. 2002a). Therein lies the pit viper's enigma. How do we explain the fact that juveniles enhance the survival of non-kin adults by providing them free food; and that adults enhance the survival of non-kin juveniles by providing them free protection?
In this paper, we simulated a 4-D environment (Cassill 2006, Cassill and Watkins 2009) to compare the lifetime fitness, over ten breeding seasons, of pit viper breeders with altruistic-only genotypes, selfish-only genotypes or a combination of altruistic and selfish genotypes. The computer simulated experiment showed the ease with which a 4-D environment favors individuals with flexible behavioral repertoires--altruistic when the opportunity arises and selfish when the opportunity arises (see also Alexander 1974, Ghiselin 1974, Cassill 2006).
A Den of Selection Models
In an "eat or be eaten" world, organisms must be ever-vigilant, looking over their shoulders for predators at the same time as they are looking ahead and searching for prey. A four dimensional (4-D) natural selection model (Cassill 2006) is needed to account for the three spatial dimensions involving populations of breeders, predators and prey, and a time dimension measured over two generations (Fig. 1). Our 4-D natural selection model predicts that the lifetime fitness of breeders (F1 generation) will be greater if able-bodied offspring (F2 generation) help themselves when the opportunity presents itself and also help others in need when the opportunity presents itself. In this paper, we tested our 4-D model experimentally using computer simulations to compare the lifetime fitness of pit viper breeders with altruistic genotypes, selfish genotypes or altruistic and selfish genotypes.
Figure 1: Fitness calculations in a 4-D environment. Pit vipers, their prey and their predators comprise the three spatial dimensions; time in the form of generations comprises the fourth dimension. Direct fitness is a 4-D retro-generation construct. Offspring survival depends on interactions with predators, prey and conspecifics. The number of surviving pit viper offspring in generation F2 determines the fitness of breeders in generation F1. Group fitness is a 1-D intra-generation construct. The number of surviving members within the same generation determines group fitness. How fitness genes are passed into the next generation has not been resolved. Inclusive fitness is a 2-D "probability" construct involving one spatial dimension and a future-generation time dimension. Follow carefully. The ideal number of relatives (Hamilton's "r" coefficient) that must be produced in a future, F3 generation by an ideal number (Hamilton's "r" coefficient) of surviving relatives of an altruistic organism in the F2 generation is determined by the probability (Hamilton's "r" coefficient) that the altruist and its surviving relatives in the F2 generation share genes by decent from a common ancestor in the F1 generation. Inclusive fitness is a highly constrained estimate of future reproduction by offspring who, in actuality, might not live to reproduce after all.
A Simulated Den of Pit vipers, Prey and Predators over Ten Breeding Seasons
Our 4-D simulation consisted of interactions among pit viper offspring, their prey and their predators on an island of 400 cells. Conditions and behaviors were as follows:
Pit viper breeders and offspring: Seven pit viper breeders were randomly located on the island; each occupied one of the 400 cells (males were assumed, and not included in the simulation). Each pit viper breeder produced four juvenile offspring per breeding season. Offspring randomly dispersed to unoccupied cells on the island. Once they entered a cell, they stayed in that cell for the remaining breeding seasons or until they died, whichever came first. In ten years, if all 40 offspring survived, 70% of the island's 400 cells would be occupied.
Pit viper nutritional needs: Juvenile pit vipers must ingest two small passerines to survive until the next migration. Adult pit vipers must ingest two large passerines to survive until the next migration.
Pit viper fitness: With ten breeding seasons and four offspring per season, the maximum lifetime fitness for each pit viper breeder was 40 offspring. If offspring survived predation and starvation the first breeding season, they then transformed into non-breeding adult offspring. The actual lifetime fitness per breeder was the total number of surviving juvenile and non-breeding adult offspring over the ten breeding seasons.
Table 1: Altruistic and selfish behavior algorithms for pit vipers.
Ratio of altruistic:selfish Juveniles Adults
1:0 (altruistic only) Kill 0% of small prey and 100% of large prey when opportunity arises. Kill 0% of prey.
Kill 100% of hawks when opportunity arises.
0:1 (selfish only) Kill 100% of small prey and 0% of large prey when opportunity arises. Kill 100% of prey, small and large, when opportunity arises.
Kill 0% of hawks when opportunity arises.
1:1 (both altruistic and selfish) Kill 100% of prey, small and large, when opportunity arises. Kill 100% of prey, small and large, when opportunity arises.
Kill 100% of hawks when opportunity arises.
0.75:1 (75% altruistic and always selfish) Kill 100% of small prey and 75% of large prey when opportunity arises. Kill all prey, small and large, when opportunity arises.
Kill 75% of hawks when opportunity arises.
0.50:1 (50% altruistic and always selfish) Kill 100% of small prey and 50% of large prey when opportunity arises. Kill all prey, small and large, when opportunity arises.
Kill 50% of hawks when opportunity arises.
0.25:1 (25% altruistic and always selfish) Kill 100% of small prey and 25% of large prey when opportunity arises. Kill all prey, small and large, when opportunity arises.
Kill 25% of hawks when opportunity arises.
Pit viper parental genotype: Simulated pit viper breeders were programmed with one of six genotypes for "altruism" and "selfishness". Each genotype represented the ratio of altruistic to selfish behavior (Table 1). Pit vipers with a 0:1 altruistic:selfish ratio were 100% selfish. For example, juvenile pit vipers struck and killed only small birds that they could swallow; adults did not kill hawks because they preyed only on juveniles, not adults. Pit vipers with a 1:1 ratio acted altruistically when the opportunity arose and acted selfishly when the opportunity arose. For example, juveniles struck 100% of prey regardless of size when prey flew nearby; adults struck 100% of prey and 100% of predatory hawks when the opportunity arose. Pit vipers with a 1:0 ratio were 100% altruistic, and never selfish. For example, juveniles struck only large prey, not small prey; adults struck only hawks, and no prey at all. A 0.5:1 ratio meant that juveniles killed and ingested 100% small prey when the opportunity presented itself, but killed only 50% of large prey (without ingesting them) when the opportunity presented itself. The other 50% of the time, juveniles ignored large prey. Adult pit vipers killed and ingested 100% of small and large prey when the opportunity arose, but killed only 50% of hawks when the opportunity presented itself. The other 50% of the time, adults ignored hawks.
Two controls were added: "no hawks present" for the 1:1 altruistic and selfish genotype and "no hawks present" for the 0:1 selfish only genotype. In the figures and text, data for "no hawk" controls were labeled with an asterisk (1.1* and 0.1*).
Pit viper kinship: To test the influence of kinship among neighbors on breeder lifetime fitness, offspring were surrounded by kin or non-kin neighbors. In the non-kin population, offspring dispersed randomly throughout the island. In the kin population, offspring moved to cells that were adjacent to their littermates.
Pit viper diversity: To test the influence of genotypic diversity among neighbors, an additional treatment included a mix of breeder genotypes on the island. Pit viper breeders were homogeneous for each of the six genotypes. And, to create a heterogeneous population, the six genotypes were randomly distributed among the seven breeders and their offspring for each of 50 replicates.
Prey migration: Each breeding season, after juvenile pit vipers hatched from the natal cell and disbursed, 800 passerines migrated to the island over two days. During Day I, 400 prey (50% large, 50% small) flew onto the island and landed randomly, one bird per cell, until the island cells were saturated. If a cell contained a pit viper, the bird might be struck and killed depending on the behavioral algorithm (Table 1). Surviving passerines flew from the island to the mainland. During Day II, a second wave of 400 passerines flew to the island (50% large, 50% small), saturating the island's cells as before. During the two-day passerine migration, each pit viper was guaranteed an opportunity to kill two birds--2 small birds, 2 large birds or 1 small and 1 large bird.
Predator migration: The first year, after two waves of prey had migrated to and from the island, seven hawks (one for each pit viper breeder on the island) migrated onto the island at randomly selected cells and then flew to the nearest pit viper offspring. If the pit viper was a juvenile, the hawk killed and ate it and then flew off the island, but returned the next year (Table 1). If the pit viper was an adult, there were two outcomes, depending on the pit viper's genotype. If the adult pit viper's genotype was altruistic, the hawk was killed, but not eaten. If the adult pit viper's genotype was selfish, the hawk lived and left the island to return the next year.
Experimental design and data analysis: The independent variables were: pit viper breeder genotype (6 genotypes in 6 homogeneous populations; 6 genotypes in 1 heterogeneous population), population relatedness (kin or non-kin neighbors) and replicates (50 per treatment). The dependent variable was the lifetime fitness for each pit viper breeder over 10 breeding seasons (7 x 2 x 50 = 700 breeder lifetime fitness outcomes). Controls were analyzed independently. Lifetime fitness data per genotype were normally distributed (Pearson's skew index), thus data were analyzed with multi-factor ANOVA, t-tests or regression using JMP IN statistical software (Sall et al. 2002).
A Den of Altruistic Offspring Wins
Among populations of non-kin dispersing offspring, genotype had a significant effect on the lifetime fitness of breeders (multi-factor ANOVA: F5,295 = 100.48; p<0.0001; Fig. 2a). On average, 23.1% of 1:1 altruistic and selfish offspring survived; 7.3% of 0:1 selfish only offspring survived; and 0.0% of altruistic only offspring survived. Thus, breeders with both altruistic and selfish genotypes realized more than three times the lifetime fitness relative to breeders with selfish-only genotypes and twenty-three times the lifetime fitness of breeders with altruistic-only genotypes. When the opportunity arose, juveniles provided meals for some adults; when the opportunity arose, adults provided protection for some juveniles. Taken altogether, the "sometime" altruistic behavior combined with the "sometime" selfish behavior resulted in increased survival for both juveniles and adult offspring, thus greatly enhancing their parent's lifetime fitness.
Accidental prey sharing among offspring accounted for 78.5% of the observed variance in breeder fitness (0:1* vs 1:1*: t98 = -26.9; p<0.0001; Fig. 2a). Hawk killing accounted for 21.2% of the observed variance in breeder fitness (0.1* vs 1:1: t98 = -7.31; p<0.0001; Fig. 2b). When juveniles provided extra prey carcasses, the presence/absence of predators did not affect breeder lifetime fitness (1.0* vs 1.0: t98 = -2.60; p = 0.995; Fig. 2a) because more adult offspring survived; when more adults survived, hawks were killed significantly sooner (Fig. 2b), eliminating predation on future litters of juvenile offspring.
Figure 2: (a) Genotype and lifetime fitness (see Table 1). In homogenous populations, breeders with a combination of altruistic genotypes had significantly greater lifetime fitness than breeders with selfish-only genotypes or altruistic-only genotypes. (b) Genotype and hawk survival. Hawks survived longer when pit viper breeders were selfish rather than altruistic. Hawks increased offspring mortality. (* = treatments which excluded hawks from the simulation; black bars = homogenous populations; gray bar = heterogeneous populations).
When breeder populations were homogeneous, kinship of neighbors did not affect a breeder's lifetime fitness (multi-factor ANOVA: F1,598 = 1.20; p = 0.274). In hindsight, this is not surprising as the alleles for these behaviors were the same or monomorphic for kin and non-kin neighbors. However, when breeder populations were heterogeneous, kinship had a significant affect on a breeder's lifetime fitness (multi-factor ANOVA: F1,48 = 48.30; p<0.001; Fig. 2a). In heterogeneous populations, when neighbors were non-kin, an average of 15% of breeder offspring survived regardless of their genotype. Thus, breeders in heterogeneous populations with kin neighbors and 100% selfish genotypes (0:1) doubled their mean lifetime fitness relative to breeders in non-kin neighbors. In contrast, breeders in heterogeneous populations with kin neighbors and combined altruistic and selfish genotypes lost half their fitness relative to those with non-kin neighbors.
The relationship between the timing of a hawk's death and breeder lifetime fitness (number of surviving offspring) was significant (Regression: breeder fitness = 8.84 - 0.53 hawk death; R2 = 0.435; p<0.001; Fig. 3). When adults killed hawks, more juveniles survived, and breeder fitness increased.
Figure 3: Predator death and breeder fitness. When adult offspring behaved selfishly, hawks lived longer and killed a larger number of juvenile offspring. As a result, the mean lifetime fitness of breeders declined.
All together, our simulation revealed three significant findings: (1) in a 4-D environment, altruism is often self-interested mutualism; (2) in a 4-D environment, expressing altruistic and selfish behavior, depending on the opportunities, provided a significant increase in offspring survival and thus, parental lifetime fitness relative to selfish-only or altruistic-only behaviors; and (3) the significant increase in offspring survival was so large, we would expect a mix of behaviors, altruistic and selfish, to evolve, hand in hand, in many species of animals exposed to a 4-D environment. Indeed, our simulation on pit vipers might enlighten us on our own dual nature; that of a fiercely competitive and, at the same time deeply compassionate, being.
The Dark Lining of Altruism's Silver Cloud is Self-Interest
The unifying theme of kin selection (Hamilton 1964) and group selection (Sober & Wilson 1999) is that altruism and self-interest are antagonists, requiring extraordinary conditions to pressure organisms into cooperative behavior (Corning 1996, Boyd 2006). Findings from our 4-D simulated experiment falsified this assumption (see also Seger 1991, Corning 1996, Landa 1998, 1999, Gifford 2000, Cassill 2006 Cassill et al. 2007). Over ten breeding seasons, pit viper breeders with a combination of altruistic and selfish genotypes increased their lifetime fitness three-fold relative to breeders with selfish-only genotypes and twenty-fold relative to breeders with altruistic-only genotypes (23% versus 7% versus 0% offspring survival rate respectively).
We suggest that the Shedao pit viper's accidental altruism is a form of pseudo-reciprocity (+/+)_you kill my predator for me, I kill your prey for you_rather than a strict form of altruism (-/+) (Trivers 1971, Connor 1986, 1995). The antecedent "accidental" is indeed correct. Juvenile pit vipers were not intentionally helping adults. They were practicing to acquire ambush skills by striking at any passerine that flew near. Practice is not that costly as venom is highly toxic and little is needed to kill prey. Likewise, non-breeding adults were perhaps practicing parental skills by striking at hawks that might someday attack their newly-hatched offspring. Regardless of their intent, adults were better fed when they reduced juvenile death rates by killing hawks. In turn, juveniles were better protected by killing large prey for adults to feed on. In both cases, offspring survival increased because neighbors helped each other. The result was a significant increase in the lifetime fitness of their parents.
Our findings demonstrate that altruistic and selfish behaviors can readily coevolve within the same genome, hand in hand, through natural selection processes. Each behavior was expressed depending on the opportunities presented to each pit viper offspring. Because opportunities varied among offspring, the expression of altruistic and selfish behavior varied among offspring. Some were more frequently altruistic than others; some were more frequently selfish than others. The important point is that their behaviors were reactive, dictated by opportunities that the environment provided. Let us be very clear about this distinction. In the homogenous populations, offspring had equal "potential" for altruistic and selfish behavior. However, the expression of that potential varied based on the environmental opportunities that crossed the path of each offspring.
The implications of these findings are several. First, most animal species live in "eat or be eaten" environments (Fig. 1). As a consequence, most animal species should have both altruistic and selfish genes within their genomes. Second, if we accept the idea that altruism evolves as easily as selfishness, we can then accept the possibility that altruism (i.e. cooperation) drives evolution as much as, or perhaps more than, selfishness or competition (see Simberloff 1982, Lewin 1983, Lewin and Lewin 1983).
Third, our findings lend credence to the notion that gene possession does not mean gene expression. The gene possession-expression dislinkage is well documented in developmental biology where organisms with the same genotype diverge in phenotype as they mature. We propose that the gene possession-expression dislinkage applies to behavior as well. Just because an organism possesses a gene for altruism, does not mean that gene will ever be expressed. Only when an opportunity presents itself might the gene be expressed. Organisms might cooperate in one circumstance, but not in another, depending on the environment (i.e. season, time of day, temperature, etc.) and traits of the other individual (i.e. health, status, age, sex, etc.). In addition, the degree of cooperation can vary as well.
Fourth, we demonstrated that once altruism and selfishness (i.e. cooperation and competition) have become fixed in the gene pool, it matters not at all to parental fitness whether help for offspring comes from kin or non-kin. Donors and recipients alike benefit whether they are kin or not. Thus, the determinants of whether neighbors are kin or non-kin are likely a result of environmental factors such as offspring dispersal opportunities (see also Zahavi and Zahavi 1997). For some species, dispersal might be more successful when offspring are juveniles; for other species, dispersal might be more successful when offspring are adults.
By understanding the dark lining of self-interest that surrounds altruism's silver cloud, and by understanding that individual organisms are capable of both altruism and selfishness, depending on their immediate circumstances, we are able to integrate both behaviors into ordinary Darwinian selection theory (Cassill 2006). We no longer need to rely so much on alternative models such as kin selection (Hamilton 1964) or group selection (Sober and Wilson 1998) to explain every instance of altruism. It should be noted that group selection came close to solving the enigma of altruism, but missed the mark by predicting mixed populations of individuals capable of altruistic or selfish behaviors rather than homogenous populations of individuals capable of altruistic and selfish behaviors. In other words, group selection predicts inter-organismal behavioral complexity whereas skew selection predicts intra-organismal behavioral complexity (as did Trivers 1974 in a different context).
In the final analysis, our 4-D natural selection experiment solved the enigma of accidental altruism in the Shedao pit-viper (see also Alexander 1974, Ghiselin 1974, Cassill 2006). We anticipate that, if we look for it, we will find that altruistic behaviors among conspecifics are as ubiquitous as selfish behaviors (see also Simberloff 1982, Lewin 1983, Lewin and Lewin 1983). The goal now is to improve on variations of existing natural selection models to further explain nuances of behavior. For example, Hopper (1999) provides an excellent review of bet-hedging (Cohen 1966, Philippi & Seger 1989, Seger & Brockmann 1987), detailing recent studies that confirm or disprove the model. Essentially, bet-hedging posits that in variable environments, breeders might reduce or forego reproduction during the current breeding season, in the hope that conditions will improve in future breeding seasons. By reducing annual fitness, a breeder might reproduce for more years and hence increase its overall lifetime fitness (Cassill 2002, Grafen 2006). Thus, there may be many evolutionary pathways to cooperation (Hirshleifer 1999), and the more open biologists are to finding them, the sooner we can return to ordinary Darwinian fitness to make sense of their benefits.
This one is about the evolution of mutualism in snakes (specifically a special kind of mutualism called "Pseudo-Reciprocity"). PR and Reciprocal Altruism really changed the way I view our social reality. I didn't think 3 years ago when I started reading & thinking about all this kind of stuff that I'd ever end up doing a Master's on it! Anyhow, this paper is under review for a journal called "Oecologia." Hope you like it:
A Den of Pit vipers
The pit viper Gloydius shedaoensis is found on the island of Shedao, in the Bohai Sea, off the north-eastern coastline of China (Li 1995). Shedao pit vipers are one of the few snake species worldwide to feed on birds from birth (Shine 1983). Pit viper venom is highly toxic and passerines die within seconds of a strike (Zhao et al. 1979, citation in Shine et al. 2002a). Pit vipers expend only a small portion of their venom reserves to realize a kill, thus many prey can be killed and ingested by pit vipers during prey migrations.
Twice a year, thousands of birds layover on the island of Shedao during their migration to and from Siberian breeding sites (Li 1995). The vast majority of the birds are passerines, a group of migratory songbirds that range in size from small finches to large ravens. During these migrations, birds are so abundant that pit vipers have the opportunity to kill far more than they need to survive to the next migration. As with most animals living in seasonal migratory zones, pit vipers experience two boom-bust cycles of prey abundance every year. During fasting periods between the migration feasts, pit vipers hibernate in subterranean burrows.
Three behaviors of the Shedao pit viper have been labeled "accidental altruism." First, despite the high density of Shedao pit vipers--approximately one snake per square meter--aggression is nonexistent. Pit vipers ignore each other's presence, even when the snakes are in physical contact on the same branch from which they ambush passerines (Shine et al. 2002b). Second, juveniles are one-tenth the size of adults. Consequently, they are gape-limited and can swallow only the smallest passerines. Regardless of their small size, juveniles strike at any prey, often killing passerines that are too large to ingest. Large passerine carcasses are abandoned by the juveniles, but not wasted, as they are quickly scavenged by nearby adult vipers. Third, adult pit vipers viciously attack and kill the sparrowhawk, Accipiter nisus, which preys on juvenile pit vipers, but not the adults (Shine et al. 2002a). Adults kill hawks at no immediate benefit to them as they do not ingest hawk carcasses.
Such "altruistic" behaviors do not fit easily into current models of social evolution. For example, kin selection cannot explain the Shedao pit viper's accidental altruism as littermates typically disperse long distances; thus, nearby pit vipers are unlikely to be siblings (Shine et al. 2002a). Therein lies the pit viper's enigma. How do we explain the fact that juveniles enhance the survival of non-kin adults by providing them free food; and that adults enhance the survival of non-kin juveniles by providing them free protection?
In this paper, we simulated a 4-D environment (Cassill 2006, Cassill and Watkins 2009) to compare the lifetime fitness, over ten breeding seasons, of pit viper breeders with altruistic-only genotypes, selfish-only genotypes or a combination of altruistic and selfish genotypes. The computer simulated experiment showed the ease with which a 4-D environment favors individuals with flexible behavioral repertoires--altruistic when the opportunity arises and selfish when the opportunity arises (see also Alexander 1974, Ghiselin 1974, Cassill 2006).
A Den of Selection Models
In an "eat or be eaten" world, organisms must be ever-vigilant, looking over their shoulders for predators at the same time as they are looking ahead and searching for prey. A four dimensional (4-D) natural selection model (Cassill 2006) is needed to account for the three spatial dimensions involving populations of breeders, predators and prey, and a time dimension measured over two generations (Fig. 1). Our 4-D natural selection model predicts that the lifetime fitness of breeders (F1 generation) will be greater if able-bodied offspring (F2 generation) help themselves when the opportunity presents itself and also help others in need when the opportunity presents itself. In this paper, we tested our 4-D model experimentally using computer simulations to compare the lifetime fitness of pit viper breeders with altruistic genotypes, selfish genotypes or altruistic and selfish genotypes.
Figure 1: Fitness calculations in a 4-D environment. Pit vipers, their prey and their predators comprise the three spatial dimensions; time in the form of generations comprises the fourth dimension. Direct fitness is a 4-D retro-generation construct. Offspring survival depends on interactions with predators, prey and conspecifics. The number of surviving pit viper offspring in generation F2 determines the fitness of breeders in generation F1. Group fitness is a 1-D intra-generation construct. The number of surviving members within the same generation determines group fitness. How fitness genes are passed into the next generation has not been resolved. Inclusive fitness is a 2-D "probability" construct involving one spatial dimension and a future-generation time dimension. Follow carefully. The ideal number of relatives (Hamilton's "r" coefficient) that must be produced in a future, F3 generation by an ideal number (Hamilton's "r" coefficient) of surviving relatives of an altruistic organism in the F2 generation is determined by the probability (Hamilton's "r" coefficient) that the altruist and its surviving relatives in the F2 generation share genes by decent from a common ancestor in the F1 generation. Inclusive fitness is a highly constrained estimate of future reproduction by offspring who, in actuality, might not live to reproduce after all.
A Simulated Den of Pit vipers, Prey and Predators over Ten Breeding Seasons
Our 4-D simulation consisted of interactions among pit viper offspring, their prey and their predators on an island of 400 cells. Conditions and behaviors were as follows:
Pit viper breeders and offspring: Seven pit viper breeders were randomly located on the island; each occupied one of the 400 cells (males were assumed, and not included in the simulation). Each pit viper breeder produced four juvenile offspring per breeding season. Offspring randomly dispersed to unoccupied cells on the island. Once they entered a cell, they stayed in that cell for the remaining breeding seasons or until they died, whichever came first. In ten years, if all 40 offspring survived, 70% of the island's 400 cells would be occupied.
Pit viper nutritional needs: Juvenile pit vipers must ingest two small passerines to survive until the next migration. Adult pit vipers must ingest two large passerines to survive until the next migration.
Pit viper fitness: With ten breeding seasons and four offspring per season, the maximum lifetime fitness for each pit viper breeder was 40 offspring. If offspring survived predation and starvation the first breeding season, they then transformed into non-breeding adult offspring. The actual lifetime fitness per breeder was the total number of surviving juvenile and non-breeding adult offspring over the ten breeding seasons.
Table 1: Altruistic and selfish behavior algorithms for pit vipers.
Ratio of altruistic:selfish Juveniles Adults
1:0 (altruistic only) Kill 0% of small prey and 100% of large prey when opportunity arises. Kill 0% of prey.
Kill 100% of hawks when opportunity arises.
0:1 (selfish only) Kill 100% of small prey and 0% of large prey when opportunity arises. Kill 100% of prey, small and large, when opportunity arises.
Kill 0% of hawks when opportunity arises.
1:1 (both altruistic and selfish) Kill 100% of prey, small and large, when opportunity arises. Kill 100% of prey, small and large, when opportunity arises.
Kill 100% of hawks when opportunity arises.
0.75:1 (75% altruistic and always selfish) Kill 100% of small prey and 75% of large prey when opportunity arises. Kill all prey, small and large, when opportunity arises.
Kill 75% of hawks when opportunity arises.
0.50:1 (50% altruistic and always selfish) Kill 100% of small prey and 50% of large prey when opportunity arises. Kill all prey, small and large, when opportunity arises.
Kill 50% of hawks when opportunity arises.
0.25:1 (25% altruistic and always selfish) Kill 100% of small prey and 25% of large prey when opportunity arises. Kill all prey, small and large, when opportunity arises.
Kill 25% of hawks when opportunity arises.
Pit viper parental genotype: Simulated pit viper breeders were programmed with one of six genotypes for "altruism" and "selfishness". Each genotype represented the ratio of altruistic to selfish behavior (Table 1). Pit vipers with a 0:1 altruistic:selfish ratio were 100% selfish. For example, juvenile pit vipers struck and killed only small birds that they could swallow; adults did not kill hawks because they preyed only on juveniles, not adults. Pit vipers with a 1:1 ratio acted altruistically when the opportunity arose and acted selfishly when the opportunity arose. For example, juveniles struck 100% of prey regardless of size when prey flew nearby; adults struck 100% of prey and 100% of predatory hawks when the opportunity arose. Pit vipers with a 1:0 ratio were 100% altruistic, and never selfish. For example, juveniles struck only large prey, not small prey; adults struck only hawks, and no prey at all. A 0.5:1 ratio meant that juveniles killed and ingested 100% small prey when the opportunity presented itself, but killed only 50% of large prey (without ingesting them) when the opportunity presented itself. The other 50% of the time, juveniles ignored large prey. Adult pit vipers killed and ingested 100% of small and large prey when the opportunity arose, but killed only 50% of hawks when the opportunity presented itself. The other 50% of the time, adults ignored hawks.
Two controls were added: "no hawks present" for the 1:1 altruistic and selfish genotype and "no hawks present" for the 0:1 selfish only genotype. In the figures and text, data for "no hawk" controls were labeled with an asterisk (1.1* and 0.1*).
Pit viper kinship: To test the influence of kinship among neighbors on breeder lifetime fitness, offspring were surrounded by kin or non-kin neighbors. In the non-kin population, offspring dispersed randomly throughout the island. In the kin population, offspring moved to cells that were adjacent to their littermates.
Pit viper diversity: To test the influence of genotypic diversity among neighbors, an additional treatment included a mix of breeder genotypes on the island. Pit viper breeders were homogeneous for each of the six genotypes. And, to create a heterogeneous population, the six genotypes were randomly distributed among the seven breeders and their offspring for each of 50 replicates.
Prey migration: Each breeding season, after juvenile pit vipers hatched from the natal cell and disbursed, 800 passerines migrated to the island over two days. During Day I, 400 prey (50% large, 50% small) flew onto the island and landed randomly, one bird per cell, until the island cells were saturated. If a cell contained a pit viper, the bird might be struck and killed depending on the behavioral algorithm (Table 1). Surviving passerines flew from the island to the mainland. During Day II, a second wave of 400 passerines flew to the island (50% large, 50% small), saturating the island's cells as before. During the two-day passerine migration, each pit viper was guaranteed an opportunity to kill two birds--2 small birds, 2 large birds or 1 small and 1 large bird.
Predator migration: The first year, after two waves of prey had migrated to and from the island, seven hawks (one for each pit viper breeder on the island) migrated onto the island at randomly selected cells and then flew to the nearest pit viper offspring. If the pit viper was a juvenile, the hawk killed and ate it and then flew off the island, but returned the next year (Table 1). If the pit viper was an adult, there were two outcomes, depending on the pit viper's genotype. If the adult pit viper's genotype was altruistic, the hawk was killed, but not eaten. If the adult pit viper's genotype was selfish, the hawk lived and left the island to return the next year.
Experimental design and data analysis: The independent variables were: pit viper breeder genotype (6 genotypes in 6 homogeneous populations; 6 genotypes in 1 heterogeneous population), population relatedness (kin or non-kin neighbors) and replicates (50 per treatment). The dependent variable was the lifetime fitness for each pit viper breeder over 10 breeding seasons (7 x 2 x 50 = 700 breeder lifetime fitness outcomes). Controls were analyzed independently. Lifetime fitness data per genotype were normally distributed (Pearson's skew index), thus data were analyzed with multi-factor ANOVA, t-tests or regression using JMP IN statistical software (Sall et al. 2002).
A Den of Altruistic Offspring Wins
Among populations of non-kin dispersing offspring, genotype had a significant effect on the lifetime fitness of breeders (multi-factor ANOVA: F5,295 = 100.48; p<0.0001; Fig. 2a). On average, 23.1% of 1:1 altruistic and selfish offspring survived; 7.3% of 0:1 selfish only offspring survived; and 0.0% of altruistic only offspring survived. Thus, breeders with both altruistic and selfish genotypes realized more than three times the lifetime fitness relative to breeders with selfish-only genotypes and twenty-three times the lifetime fitness of breeders with altruistic-only genotypes. When the opportunity arose, juveniles provided meals for some adults; when the opportunity arose, adults provided protection for some juveniles. Taken altogether, the "sometime" altruistic behavior combined with the "sometime" selfish behavior resulted in increased survival for both juveniles and adult offspring, thus greatly enhancing their parent's lifetime fitness.
Accidental prey sharing among offspring accounted for 78.5% of the observed variance in breeder fitness (0:1* vs 1:1*: t98 = -26.9; p<0.0001; Fig. 2a). Hawk killing accounted for 21.2% of the observed variance in breeder fitness (0.1* vs 1:1: t98 = -7.31; p<0.0001; Fig. 2b). When juveniles provided extra prey carcasses, the presence/absence of predators did not affect breeder lifetime fitness (1.0* vs 1.0: t98 = -2.60; p = 0.995; Fig. 2a) because more adult offspring survived; when more adults survived, hawks were killed significantly sooner (Fig. 2b), eliminating predation on future litters of juvenile offspring.
Figure 2: (a) Genotype and lifetime fitness (see Table 1). In homogenous populations, breeders with a combination of altruistic genotypes had significantly greater lifetime fitness than breeders with selfish-only genotypes or altruistic-only genotypes. (b) Genotype and hawk survival. Hawks survived longer when pit viper breeders were selfish rather than altruistic. Hawks increased offspring mortality. (* = treatments which excluded hawks from the simulation; black bars = homogenous populations; gray bar = heterogeneous populations).
When breeder populations were homogeneous, kinship of neighbors did not affect a breeder's lifetime fitness (multi-factor ANOVA: F1,598 = 1.20; p = 0.274). In hindsight, this is not surprising as the alleles for these behaviors were the same or monomorphic for kin and non-kin neighbors. However, when breeder populations were heterogeneous, kinship had a significant affect on a breeder's lifetime fitness (multi-factor ANOVA: F1,48 = 48.30; p<0.001; Fig. 2a). In heterogeneous populations, when neighbors were non-kin, an average of 15% of breeder offspring survived regardless of their genotype. Thus, breeders in heterogeneous populations with kin neighbors and 100% selfish genotypes (0:1) doubled their mean lifetime fitness relative to breeders in non-kin neighbors. In contrast, breeders in heterogeneous populations with kin neighbors and combined altruistic and selfish genotypes lost half their fitness relative to those with non-kin neighbors.
The relationship between the timing of a hawk's death and breeder lifetime fitness (number of surviving offspring) was significant (Regression: breeder fitness = 8.84 - 0.53 hawk death; R2 = 0.435; p<0.001; Fig. 3). When adults killed hawks, more juveniles survived, and breeder fitness increased.
Figure 3: Predator death and breeder fitness. When adult offspring behaved selfishly, hawks lived longer and killed a larger number of juvenile offspring. As a result, the mean lifetime fitness of breeders declined.
All together, our simulation revealed three significant findings: (1) in a 4-D environment, altruism is often self-interested mutualism; (2) in a 4-D environment, expressing altruistic and selfish behavior, depending on the opportunities, provided a significant increase in offspring survival and thus, parental lifetime fitness relative to selfish-only or altruistic-only behaviors; and (3) the significant increase in offspring survival was so large, we would expect a mix of behaviors, altruistic and selfish, to evolve, hand in hand, in many species of animals exposed to a 4-D environment. Indeed, our simulation on pit vipers might enlighten us on our own dual nature; that of a fiercely competitive and, at the same time deeply compassionate, being.
The Dark Lining of Altruism's Silver Cloud is Self-Interest
The unifying theme of kin selection (Hamilton 1964) and group selection (Sober & Wilson 1999) is that altruism and self-interest are antagonists, requiring extraordinary conditions to pressure organisms into cooperative behavior (Corning 1996, Boyd 2006). Findings from our 4-D simulated experiment falsified this assumption (see also Seger 1991, Corning 1996, Landa 1998, 1999, Gifford 2000, Cassill 2006 Cassill et al. 2007). Over ten breeding seasons, pit viper breeders with a combination of altruistic and selfish genotypes increased their lifetime fitness three-fold relative to breeders with selfish-only genotypes and twenty-fold relative to breeders with altruistic-only genotypes (23% versus 7% versus 0% offspring survival rate respectively).
We suggest that the Shedao pit viper's accidental altruism is a form of pseudo-reciprocity (+/+)_you kill my predator for me, I kill your prey for you_rather than a strict form of altruism (-/+) (Trivers 1971, Connor 1986, 1995). The antecedent "accidental" is indeed correct. Juvenile pit vipers were not intentionally helping adults. They were practicing to acquire ambush skills by striking at any passerine that flew near. Practice is not that costly as venom is highly toxic and little is needed to kill prey. Likewise, non-breeding adults were perhaps practicing parental skills by striking at hawks that might someday attack their newly-hatched offspring. Regardless of their intent, adults were better fed when they reduced juvenile death rates by killing hawks. In turn, juveniles were better protected by killing large prey for adults to feed on. In both cases, offspring survival increased because neighbors helped each other. The result was a significant increase in the lifetime fitness of their parents.
Our findings demonstrate that altruistic and selfish behaviors can readily coevolve within the same genome, hand in hand, through natural selection processes. Each behavior was expressed depending on the opportunities presented to each pit viper offspring. Because opportunities varied among offspring, the expression of altruistic and selfish behavior varied among offspring. Some were more frequently altruistic than others; some were more frequently selfish than others. The important point is that their behaviors were reactive, dictated by opportunities that the environment provided. Let us be very clear about this distinction. In the homogenous populations, offspring had equal "potential" for altruistic and selfish behavior. However, the expression of that potential varied based on the environmental opportunities that crossed the path of each offspring.
The implications of these findings are several. First, most animal species live in "eat or be eaten" environments (Fig. 1). As a consequence, most animal species should have both altruistic and selfish genes within their genomes. Second, if we accept the idea that altruism evolves as easily as selfishness, we can then accept the possibility that altruism (i.e. cooperation) drives evolution as much as, or perhaps more than, selfishness or competition (see Simberloff 1982, Lewin 1983, Lewin and Lewin 1983).
Third, our findings lend credence to the notion that gene possession does not mean gene expression. The gene possession-expression dislinkage is well documented in developmental biology where organisms with the same genotype diverge in phenotype as they mature. We propose that the gene possession-expression dislinkage applies to behavior as well. Just because an organism possesses a gene for altruism, does not mean that gene will ever be expressed. Only when an opportunity presents itself might the gene be expressed. Organisms might cooperate in one circumstance, but not in another, depending on the environment (i.e. season, time of day, temperature, etc.) and traits of the other individual (i.e. health, status, age, sex, etc.). In addition, the degree of cooperation can vary as well.
Fourth, we demonstrated that once altruism and selfishness (i.e. cooperation and competition) have become fixed in the gene pool, it matters not at all to parental fitness whether help for offspring comes from kin or non-kin. Donors and recipients alike benefit whether they are kin or not. Thus, the determinants of whether neighbors are kin or non-kin are likely a result of environmental factors such as offspring dispersal opportunities (see also Zahavi and Zahavi 1997). For some species, dispersal might be more successful when offspring are juveniles; for other species, dispersal might be more successful when offspring are adults.
By understanding the dark lining of self-interest that surrounds altruism's silver cloud, and by understanding that individual organisms are capable of both altruism and selfishness, depending on their immediate circumstances, we are able to integrate both behaviors into ordinary Darwinian selection theory (Cassill 2006). We no longer need to rely so much on alternative models such as kin selection (Hamilton 1964) or group selection (Sober and Wilson 1998) to explain every instance of altruism. It should be noted that group selection came close to solving the enigma of altruism, but missed the mark by predicting mixed populations of individuals capable of altruistic or selfish behaviors rather than homogenous populations of individuals capable of altruistic and selfish behaviors. In other words, group selection predicts inter-organismal behavioral complexity whereas skew selection predicts intra-organismal behavioral complexity (as did Trivers 1974 in a different context).
In the final analysis, our 4-D natural selection experiment solved the enigma of accidental altruism in the Shedao pit-viper (see also Alexander 1974, Ghiselin 1974, Cassill 2006). We anticipate that, if we look for it, we will find that altruistic behaviors among conspecifics are as ubiquitous as selfish behaviors (see also Simberloff 1982, Lewin 1983, Lewin and Lewin 1983). The goal now is to improve on variations of existing natural selection models to further explain nuances of behavior. For example, Hopper (1999) provides an excellent review of bet-hedging (Cohen 1966, Philippi & Seger 1989, Seger & Brockmann 1987), detailing recent studies that confirm or disprove the model. Essentially, bet-hedging posits that in variable environments, breeders might reduce or forego reproduction during the current breeding season, in the hope that conditions will improve in future breeding seasons. By reducing annual fitness, a breeder might reproduce for more years and hence increase its overall lifetime fitness (Cassill 2002, Grafen 2006). Thus, there may be many evolutionary pathways to cooperation (Hirshleifer 1999), and the more open biologists are to finding them, the sooner we can return to ordinary Darwinian fitness to make sense of their benefits.
Alot of hard work went into this!