CHAPTER 11 Kin Selection and Social Behavior nt workers carrying leaves back to their colony/s nest. Inside the nest, other workers will chew the leaves and feed the colony’s fungal garden. OCIAL INTERACTIONS CREATE THE POSSIBILITY FOR CONFLICT AND COOPERATION, Consider two American crows (Conus bracltyrhynchos) patrolling the edge of their adjacent nesting territories. If one moves across the established boundary, its action may trigger a sive calls, a flight chase, or even physical combat. But ifa hawk flies by, the two antagonists will cooperate in chasing the predator away. Later in the day, these same individuals may spend con- siderable time and effort feeding the youn; birds in the nests in their respective territories, even though the nestlings are the crows’ siblings or half-siblings and not their own offspring. When and why do these individuals cooperate with each other, and why do they help their parents raise their siblings instead of leaving home to rear their own offipring? What conditions lead to conflicts with each other and with their parents, and how are these conflicts resolved? These are the types of questions addressed in this chapter In fitness terms, an interaction between individuals has four possible outcomes (Table 11.1). Cooperation (or mutualism) is the both participants. Altruism represents cases in which the individual instigating the action pays a fitness cost and the individual on the receiving end benefits. In other words, altru- erm for actions that result in fitness gains for ism entails a sacrifice on behalf of another. Selfishness is the opposite: The actor gains and420 PARTIII Adaptation Explaining altruistic behavior Js a challenge for the Theory of Evolution by Natural Selection. Table 11.1 Types of social interactions ‘The “actor” in any social interaction affects the recipient of the action as well as itself The costs and benefits of interactions are measured in units of surviving ofispring (fitness). Actor benefits Actor is harmed Recipient benefits Cooperative Altruistie Recipient is harmed Selfish Spitetul the recipient loses. Spite is the term for behavior that results in fitness loses both participants. Understanding the evolution of these four interactions is made simplerb cause there are no clear-cut examples of spite in nature (Keller et al. 199 but see Foster et al. 2000). It is straightforward to understand why spitel not evolved: An allele that results in fitness losses for both actor and recip would quickly be eliminated by natural selection. Intuition might sugg that a spiteful actor could come out ahead if the cost he imposes on 2.0 petitor is greater than cost he suffers himself. Consider, however, the fitnes rd party who avoids involvement in spiteful acts and pays neither the ¢ of doling them out nor the cost of receiving them, Altruism would seem equally difficult to explain, because the actor st a fitness loss. Altruistic behavior appears to be common, however, Exam range from the crows that help at their parents’ nests to a human who df into a river and saves a drowning child. This is the first question we neg address: Why does altruism exist in nature? 11.1 Kin Selection and the Evolution of Altruism Altruism is a central paradox of Darwinism. It would seem impossible form al selection to favor an allele that results in behavior benefiting other indivi at the expense of the individual bearing the allele. For Darwin (1859: 236) apparent existence of altruism presented a “special difficulty, which at fint peared to me insuperable, and actually fatal to my whole theory.” Fortunate was able to hint at a resolution to the paradox: Selection could favor traits that sult in decreased personal fitness if they increase the survival and reprodui success of close relatives. Over a hundred years passed, however, before this was formalized and widely applied. Inclusive Fitness In 1964, William Hamilton developed a genetic model showing that an allele favors altruistic behavior could spread under certain conditions (Hamill 1964a). The key parameter in Hamilton's formulation is the coefficient of latedness, r. This is the probability that the homologous alleles in two indi als are identical by descent (Box 11.1). The parameter is closely related ta the coefficient of inbreeding, which we introduced in Chapter 6. F is the pm bility that homologous alleles in the same individual are identical by descent.Chapter 11 Kin Selection and Social Behavior 421 Box 11.1 Calculating coefficients of relatedness S lating r, the coefficient of relatedness, re- quires a pedigree that includes the actor (the in- Mividual dispensing the behavior) and the recipient (ite individual receiving the behavior). Starting with the actor, all paths of descent are traced through the Pedigree to the recipient. For example, half-siblings hare one parent and have two genealogical connec- fons as indicated in Figure 11.1a. Parents contribute Ihlfthcir genes to each offiprin Hi genes are identical by descent (ibd) in each step the path is $0 the probability 2, Put another way, the probabil ata particular allele was transmitted from parent to Hicior is 1/2. The probability that the same allele was Hinsmitted from parent to recipient is 1/2. The Gpobubility that this same allele was transmitted to both the actor and the recipient (meaning that the lees in actor and recipient are ibd) is the product of Hfese two independent probabilities, or 1/4. Fill siblings, on the other hand, share genes in- ited from both parents. To calculate r when actor and recipient are full-siblings, we have to add the Shsbilicies that genes are ibd through each path in Mhe pedigree. In this case, we add the probability fiat genes are ibd through the mother to the prob- pity that they are ibd through the father (see Bare 11.1b).This is } + 1 = 4 Using this protocol results in the following co. Hiicients 9) Hat slings (©) Ful-sibings Shared parent Mother Bohs dt |e Recipient Actor Recipient Given 7, the coef scott Ha Parearte ofBpeing:} Grandparent to grandchild, $ unite eee reed The analyses we have just performed work for autosomal loci in sexual organisms and assume that no inbreeding has occurred. If the population is inbred when studying populations in the field, investig then coefficients will be higher. But tors usually have no data on inbreeding and have to assume that individuals are completely outbred. On this basis, coefficients of relationship that are reported in the literature should be considered minimum estimates. Another uncertainty in calcu lating values of r comes in assigning paternity in pedigrees. As we indicated in Chapter 10, extra- pair copulations are common in many species. If paternity is assigned on the basis of male-female pairing relationships and extra-pair copulations go undetected, estimates of r may be inflated. When constructing genealogies is impractical, be estimated dire, ly from genetic data (Queller and Goodnii 1989) proving to be extremely useful for calculating r in coefficients of relatedness ca Microsatellites and other marker loci are a wide variety of social insects (e.g 1999). Peters et al (6) Cousins Actors parent ‘Actor's aunt orunce Figure 11.1. Computing ) relatedness with pedigrees 1. Actor Recipient ‘ent of relatedness between the actor and the recipient, iton’s rule states that an allele for altruistic behavior will spread if Br-C>0 Bis the benefit to the recipient and Cis the cost to the actor, Both B GC are measured in units of surviving offspring. This simple law means that422 PARTI Adaptation Inclusive fitness consieta of direct fitness due to personal reproduction and indirect fitness due to additional reproduction by relatives. Behavior that results in indirect fitness gains fe favored by kin selection. altruism is more likely to spread when the benefits to the recipient are gf the cost to the actor is low, and the participants are closely related. To generalize this result, Hamilton offered the concept of inclusive fit pointed out that an individual's fitness can be partitioned into two conf nts, Direct fitness results from personal reproduction. Indirect fitness re from additional reproduction, by relatives that is made possible by an indivi actions. Indirect fitness"Kécrues when relatives achieve reproductive success ab and beyond what they would have achieved on their own—that is, without Natural selection favoring the spread of alleles that increase the indirect 0 nent of fitness is called kin selection. As we will see, most instances of altuis nature are the result of kin selection. Robert Trivers (1985:47) called Hamilton’s rule and the concept of indl fitness “the most important advance in evolutionary theory since the Wo Charles Darwin and Gregor Mendel.”"To see why, we will apply the theany venturing to the Sierra Nevada of California and observing a social mami Belding’s ground squirrel (Spermophilus belding). Alarm Calling in Belding’s Ground Squirrels Explaining alarm calling in birds and mammals is a classical application of sive fitness theory. When flocks or herds are stalked by a predator, prey ind als that notice the intruder sometimes give loud, high-pitched calls. warnings alert nearby individuals and allow them to flee o dive for covet may also expose the calling individual to danger. Belding’: ground squinrd two kinds of alarm calls (Sherman 1985): They trill in response to pred mammals and whistle in response to flying hawks. The first question to ask about the alarm calls of Belding’s ground squit whether they are genuinely altruistic. During 14 years of observation, Paul rman (1983), his colleagues, and his students observed 256 natural pred tacks, including 30 in which ground squirrels were captured and Kile. Ie ut that when squirrels spot an attacking hawk and whistle, the whisting rel is captured only 2% of the time while non-whistling squirrels are cap 28% of the time. The squirrel raising the alarm apparently reduces its chances of dying, perhaps by informing the hawk that the caller has seen squirrels spot an attacking mammal and trill, on the other hand, the tilling rel is killed 8% of the time while non-trilling squirrels are killed only 4%0f time. The squirrel raising the alarm apparently increases its own fieril to the efit of other squirrels nearby. Whistles are selfish, but tills ar Paul Sherman (1977, 1980) studied patterns in the altruistic alarm eal gi by Belding’s ground squirrels to determine why this behavior evolved. Bell ground squirrels breed in colonies established in alpine meadows. Males dsp far from the natal burrow, while female offipring tend to rem: by. As a result, females in proximity tend to be closely related. Because Sh had individually marked many individuals over the course of the study. able to construct pedigrees and calculate coefficients of relatedness fori members of the colony When Sherman compiled data on which squirrels called at the approach +, two outstanding patterns emerged: Females werepient are gre: Chapter 11 o) i oe eS Bn wee 10 Ll, >—— Expected ‘ Observed 11.2 In ground squirrels, most alarm calling is done by females (a) This female Beld: und squirrel is giving an alarm call. (b) This bar chart reports the observed and expected fr ‘of alarm calling by different sex and age classes of Belding’s ground squirrels, based on 102 fers with predatory mammals. The expected values are indicated by the blue line and are cal by assuming that individuals call randomly—that is, in proportion to the number of times present when the predator approaches. The observed and expected values are significantly (P= 0.001). Reprinted with permission from Sherman (1977). Wikely to call than males (Figure 11.2), and females were much more likely When they had close relatives within earshot (Figure 11.3). These data _"4ividuals are more likely to Support the hypothesis that kin selection is responsible for the evolution gm calling. Sherman (1981) has also been able to show that mothers, daugh- sisters are much more likely to cooperate when chasing trespassing squir- their territory than are more distant kin or nonrelatives (Figure 11.4) ata show that altruistic behavior is not dispensed randomly. It is nepotis- Psacrficing behavior is directed at close relatives and should result in indi- ‘ans. ‘ive alarm calls when close relatives are nearby. Fronted Bee-Eaters ‘lassial system for studying kin sclection in vertebrates is helping behav- (Gee Brown 1987; Stacey and Koenig 1990). In species from a wide va~ Figure 11.3. Female ground squirels are ss S aay an ae cannes ’ qj tre nearby This bar cert summarizes data , 7 eiiectasr epee ee PA approached aground sel colony Each OY AT os puted comparson inthe igure represents the OSS FP SFE end (Randa ns oexatons whan at least rw foray each coe o Fe (9>ry was present, but no others. The expected i . = values are computed as in Figure 11.2. The Gval- Les reported here represent a modified form of _ eee ae the 4’ statistic introduced in Chapter 5. WS. stands for not significant. From Sherman (1980). Observed Kin Selection and Social Behavior 423424 PARTI Adaptation ‘more likely to cooperate than distant kin This bar chart reports how frequently teritory-owning females were joined by different categories of relatives and nonrelatives in chas- ing away trespassing ground squirrels. The number in paren: ‘theses indicates the number of chases that occu different dyads of each kind that were observed. The vertical ‘bars give the standard deviations of the frequencies. There is a significant difference in the frequency of cooperation be- ‘ween the three categories on the left and the seven cate- 0.08). 020-159950) 69) 6 4) GO) 48) GH os} oso] Io Fare 134. Cooperate bell ghaiees digas oe T os Lt el “ear ne sos N 2 8 7 1% 4 0 10 both types of individuals were present. N gives the number of Bg g§ BB geEg z yo G82 582 4 a6 a3 5 toe ia2 om 3 ea. ERS ee! poe TEE #85685 € BE eg a gee 4 dct cua ab = 2 ae * From Sherman (1981) Helpers assist with nest building, nest defense, or food delivery to incubating cents and the nestlings. Helping at the nest is usually found in species where breeding opportuni are extremely restricted, either because habitats are saturated with esabli breeding pairs or because suitable nest sites are difficult to obtain, In these @ gaining direct fitness is almost impossible for young adults. Gaining indirect ness by helping becomes a best-of-a-bad-job strategy. Steve Emlen, Peter Wrege, and Natalie Demong have completed an inte study of helping behavior in the white-fronted bee-eater (Merops bullocka This colonial species, native to East and central Africa, breeds in nesting cham excavated in sandy riverbanks (Figure 11.5). The 40-450 individuals in a ca are subdivided into groups of 3 to 17, th of which defends a feeding teri up to 7 km away. The c clans may include several sets of parents and offspring Many year-old bee-eaters stay to help at the nest during what would o wise be their first breeding season. Clan members are related, so helpers ti have a choice of nestlings with different degrees of kinship as recipients of helping behavior (Hegner et al. 1982; Emle point: Be Figure 11.5 White-fronted bee-eaters The individual in the middle is performing a wing waving display, and may be soliciting a feeding et al. 1995 ise kinship varies among the potential recipients of altruistic This choice is@Chapter 11 Kin Selection and Social Behavior 425 white-fronted bee-eaters are an excellent species for researchers to use in @ ing theories about kin selection. gO] (158 Nata Aer marking large numbers of individuals and working out genealogies over Biniaw fight-year study period, c (1988, 1991) found that bee-caters — B form to predictions made by Hamilton’ rule. They determined, for example, 4p the coefficient of relatedness with recipients has a strong effect on whethera. : Bibrecding member of the clan helps (Figure 11.6a). Further, nonbreeders ac~ 220 au decide to help the most closely related individuals available (Figure 11.66) & is, when young with different coefficients of relatedness are being reared fihin their clan, helpers almost always chose to help those with the highest r x 11.2). Their assistance is an enormous benefit to pai ater young die of starvation before lea nts. More than half of ng the nest. On average, the pres- fe ofeach helper results in an additional 0.47 offspring being reared to fledg- (Figure 11.7). For young birds, helping at the nest results in clear benefits for ive fitness. 10 : . Observed Selection in Other Contexts Brent research has rev Figure 11.6 In bee-eaters, helpers assist close relatives (@) Bee-eater clans often con tain nonbreeders that have paired with members of the clan. Their r with the offspring being raised that season is 0. This bar chart shows that they are much les likely to help than are clan members (P < 0.01). (6) In this bar chart, the expected probability of helping is calculated by assum- ing that helpers assist clan ‘members randomly, in propor: tion to the rs of nestlings in clan nests. A G test rejects the then null hypothesis that helping is at the non-kin tad- directed randomly with respect In fact, only 6 of 28 cannibals ate their siblings, a significantly lower per- ‘© kinship (P < 0.01). From than expected by chance (Figure 11.8b). Se SPiennig and colleagues (1999) tested whether the discriminating behavior of fibiistic tadpoles satisfies the requirements of Hamilton's inequality. The re- led that kin selection is responsible for the evolution of fustc behavior in a great variety of other situations beyond the class Bples we have reviewed. We briefly consider three such cases here. ibaistic Tadpoles tadpoles of spadefoot toads (Spea bombiffons) can develop into either of two Biphs. The typical morph is omnivorous and feeds mostly on decaying plant ‘The alternative morph has enlarged jaw nuseles that allow it to eat big f including other spadefoot toad tadpoles (Figure 11.8a). David Pfennig BY) tested whether cannibalistic tadpoles distinguish between kin and no He put 28 cannibalistic tadpoles into separate containers. Each tadpole its container with two omnivorous tadpoles it had never seen before: One fill sibling; the other was non-kin. Pfennig then waited until each cannibal eaten one of its companions. If cannibalistic tadpoles are indiscrimin Wwould be as likely to eat their siblings as they are to z 1 thers used tiger salamanders (Ambystoma tigrinum), another species in which E ls can develop either into a typical morph or a cannibal morph. Pfennig & 9] 1 galleagues kept 18 cannibal tadpoles in separate cages placed ina natural 3 | Each cannibal shared its cage with 6 of its typical-morph siblings, and 18 © 9 | i in ypical-morph tadpoles. As in spadefoot toads, some of the cannibalistic 8 | 2 discriminated between kin versus non-kin—eating kin les often than 3 | 59.7) libe expected by chance—and others did not. 3 es by Hamilton's rule discrimination is favored if B(1/2) — C > 0, where B eee Be benefit to kin of avoiding eating them and C is the cost to the cannibal of Creare inga chance for a meal. Pfen g and colleagues estimated the benefit by count ich cannibals siblings that survived. They found that the sib- ig cannibals were twice as likely to survive as the siblings of the number of of discrimin: Figure 11.7. Fitness gains due to helping From Emlen ‘and Wrege (1991),426 PARTI Adaptation Box 11.2 Kin recognition Gear individuals coproteins coded for by MHC loci are released have accurate mechanisms for assessing their de- the urine of mice, and that mice can distinguish th molecules by smell. Mice are, for example, abe distinguish full siblings from half-siblings on the h gree of kinship with members of their own species, or conspecifics. This phenomenon, called kin recogni- tion, has been divided into two broad categories: di- of their MHC genotypes. But do mice dispense rect and indirect (Pfennig and Sherman 1995). truistic behavior accordingly? Indirect kin recognition is based on cues like the im- Hou: mice form communal nests and nurse & ing or location of interactions. Many species of adult others’ pups. Because individuals could take advant birds rely on indirect kin recognition when their chicks are young and will feed any young bird that ap- pears in thei of this cooperative system by contributing les th their fair share of milk, Manning et al. predicted & mothers would prefer to place their young in nd containing close relatives. The logic here is that el st. Direct kin recognition, in contrast, is based on specific chemical, vocal, or other cues, ‘There is currently a great deal of interest in deter-__kin should be less likely to cheat on one another cause of the cost to their indirect fitness. Througl program of controlled breeding in which wild-cug mice were crossed with laboratory strains, Manning al. created a population of mice with known MI genotypes. This population was allowed to establish self in a large barn. The researchers then recon mining whether loci in the major histocompatibility complex (MHC) function in direct kin recognition, In Chapter 6, we introduced the MHC and its role in self nonself recognition by cells in the immune system. Although loci in the MHC clearly evolved to function in disease prevention, polymorphism is so extensive that non-kin share very few alle sult, the genes can serve as reliable markers of kin- born pups after birth. The null hypothesis was ti ship (Gee Brown and Eklund 1994). Could similarity in MHC provide a reliable cue of kinship and offer a criterion for dispensing altruistic behavior? Jo Manning and colleagues (1992) addressed this question in a population of house mice (Mus muscu- lus domesticus). A kin recognition system. requires three components: production of the signal, recogni. tion of the signal by conspecific, and action based on where mothers in this population placed their 1 mothers would choose to rear their offspring randa ly with respect to the MHC genotypes presenting communal nests available at the time. Contrary tof null expectation, mothers showed a strong prefers for rearing their young in nests containing off with similar MHC genotypes. This result confirm role for MHC as a signal used in direct kin recog tion and shows that mice are capable of dispensing that recognition. Previous work had shown that gly- truistic behavior on the > genotypes nondiscriminating cannibals (Figure 11.8c). Thus the benefit in this st roughly 2. The researchers assessed the cost to a discriminating cannibal of mil a chance for a meal by comparing the growth rate and age at metamorphosis of criminating versus non-discriminating cannibals. In neither case was there a sig cant difference. Thus the cost in this system is nearly zero, Putting these ther, we have B(1/2) — C = approximately 1. Pfennig and colleagues @ cluded that not eating siblings is favored by kin selection in tiger salamander tad Altrnistic Sperm Harry Moore and colleagues (2002) documented altruistic behavior in thes of the common European wood mouse (Apodemus sylvutiats). Female wood are highly promiscuous (Short 2002). When females mate with more thin male in short succession, the sperm from the two males are in a race to the see Cl apter 10). This appears to explain why male wood mice have suchChapter 11 Kin Selection and Social Behavior 427 Expected Percent that ate kin N=28 Observed Camivores lion's ls, avoidance of sibling cannibalism will be selectively favored when B{1/2) -C > 0. Benefit 8 = 2 Cost: C0 {Change in SVL (men) Age at metamorphosis () U8 Kin-selected discrimination in cannibalistic tadpoles (2) In spadefoot toads develop into carnivores and become cannibals. (b) Given a choice between eating Sus non-kin, most cannibals are discriminating and preferentially eat non-kin. From Pfen- (i) The same phenomenon occu in tiger salamanders. Under semi-natural conditions, fing cannibals see twice as many siblings survive as nondiscriminating cannibals left) but ible cost (center and right). The costs and benefits of discriminating tastes thus satisfy fequation for kin selection. From Pfennig et a. (1999). Bure 11.9a). Large testes are an adaptation for sperm competition, because sperm a male enters in the race the greater his chances of winning, feeshow other adaptations for sperm competition too, as Moore and col- red when they examined wood mouse sperm under a microscope. Miouse sperm have hooks on their heads (Figure 11.9b). Using these hold other sperm by their heads or their flagella, wood mouse sperm iS composed of tens or thousands of individual cells (Figure 11.9). the sperm in a train can move about twice as fast as sperm igure 11.94). This cooperation would seem to improve their but there isa catch. Before many of the sperm in the train undergo acrosome reaction, an Feleases enzymes that usually help a sperm fertilize an egg. By releasing ;,many of the sperm in the train sacrifice428 PART IN| Adaptation 3 ‘Average path velocty ums") 1 10 Viscosity og scale) Figure 11.9 Altruistic sperm (2) Male wood mice have large testes, which suggests that sperm competition iso From Short (2002). (b) Wood mouse sperm have hooks on their heads. (c) The sperm use these hooks to join together in trains. () Trains of sperm swim about twice as fast as lone sperm. Before fertilization, however, the sperm ina tain mul ‘engage. They accomplish this via the self sacrifice of many of the train's members. (b) (c) and (d) from Moore etal. Selection can favor making sacrifices for kin; it should also favor avoiding sacrifices for non-kin themselves on behalf of their siblings. Given that sibling sperm share half alleles, this altruistic sacrifice appears to be consistent with kin selection, Clever Coots The flip side of incurring costs on behalf of kin is avoiding paying costs on of non-kin. Bruce Lyon (2003) documented a defense against parasitized atl in an aquatic bird called the American coot (Fulica americana). Female cootsdl try to increase their reproduce success by laying eggs in other females’ n phenomenon known as conspecific nest parasitism. Accepting a parasitic gg rearing the chick that hatches from it comes at a substantial cost. About the chicks in a typical nest die of starvation and the number of chicks that su is the same for parasitized and non-parasitized nests. These facts suggest hata ter parent loses one of its biological offspring for every parasitic offspring it ‘Assuming that the coefficient of relatedness is zero between the foster pare the parasitic chick, Hamilton's rule predicts that coots should have evolved fenses against parasitism, During four field seasons, Lyon monitored over 400 nests in a wild coot ulation. Coot eggs show considerable variation in appearance (Figure 11.1 two new eggs appeared in a nest within a 24 hour period, Lyon could infer the nest had been parasitized. And if one of the eggs did not match the oth Lyon could infer that it was the impostor. The coots themselves apparently pay attention to similar clues. Among I hosts, 43% rejected one or more parasitic eggs. The rejected eggs differed the host's eggs significantly more than did the accepted eggs (Figure 11.10b).Chapter 11 429 Kin Selection and Social Behavior © 2{ 02 Acceptors ro aee. Go] ss Behl ois aoe ace = 8] o's Giaplor Se? dl Biba imming 5 a ine y; L arving : ve is a . 7 Be Lp ack eameae 7 3 é <1 s é = 8 . — " Accepted Rejected 3 ‘00 ol fd 3 i z 3 Number of parasitic eggs by day 3 erin large mp figure 11.10 Clever coots protect themselves against parasitized altruism (a) Coot eggs vary in both background color h must dis: fe sedstribution of spots, As a result, when an extra egg appeared in a nest Lyon (2003) was able to tell which egg had been laid (2002), Iy2paraste. lon developed a scale, color rank, that allowed him to quantify the difference in appearance between host eggs andparastic eggs. (b) Many hosts reject parasitic eggs. These box plots show that rejected eggs ctfer from the host's eggs more than do accepted eggs. The large horizontal nes indicate the median rank color difference, the tops and bottoms of the boxes inate the 25th and 75th percentiles, and the thin ticks indicate the 10th and 90th percentiles, () Females that accepted para Sicean aid fewer of their own eggs. The dashed horizontal line indicates the mean clutch size for unparasitized females, the Slline isthe best-fit ine through the data, and the numbers next to data points indicate multiple females. () Females that e- ested parasitic eggs laid the same number oftheir own eggs as unparasitized females, even though they decided how many eggs s on behalf foley before they cisposed of the eggs they would reject. From Lyon (2003). aif of theif ed altruism coots often les’ nests, a tic egg and sour half of that survive fadicates chat coots can distinguish their own eggs from those of other females hhsed on appearance. Females that accepted parasitic eggs laid one fewer egg of their own for every IPasitic egg they accepted (Figure 11.10¢). This meant that the acceptors reared s total, the same as the average clutch size for non-parasitized i chat a fos imal. Females that rejected parasitic eggs laid an average of 8 of their own vce it rearalfp gecven though they decided how many of their own eggs to lay before they Msposed of the eggs they would reject (Figure 11.10d). This gave them a final hich size of 8 egys. And it shows that coots can count. By counting their eggs and rejecting the extras that do not look right, coots Prevent themselves from being duped into unwitting maladaptive altruism. parent and evolved des d coot pops ¢ 11.103). 1 ld infer that + the others, Itthe examples of kin selection we have discussed so far, the alleles responsible for Greenbeard Alleles aliustic behavior have risen to high frequency by playing the odds. Some of the Among 134) fin that the alleles influence their carriers to help ca alleles; lffered frompp hers do not. But so long as the alleles for altruism induce their carriers to obey 11.10b). This f Hamilton's rule—helping kin only when the cost/benefit ratio is sufficiently low tain copies of th432 PARTI Adaptation In haplodiploid species, females are more closely 1 to their sisters than they are to their own offepring Eusociality (true sociality) is used to describe social systems with three chan istics (Michener 1969; Wilson 1971;Alexander et al. 1991): (1) ove between parents and their offipring, (2) cooperative brood care, a castes of nonreproductive individuals. Eusocial species are found in a varie orders (Table 11.2), snapping shrimp (Duffy 1996; Duffy et al. 2000; Duffy 2002), and one family of rodents (the mammal family Bathyergidae, or mole As an entree to the extensive literature on eusociality in this section we sider how reproductive altruism evolved in two very different groups: the menoptera (ants, bees, and wasps) and mole-rats. Haplodiploidy and Eusocial Hymenoptera The Hymenoptera represent the pinnacle of social evolution. A single ant eal may number millions of individuals, each appearing to function more like at in a superorganism than an individual pursuing its own reproductive inten Worker, soldier, and reproductive castes, which seem analogous to tissues if body, can be identified on the basis of their morphology and the tasks they form. But unlike cells and tissues, individuals in the colony are not genetil identical. What factors laid the groundwork for such extensive altruism? eusociality so widespread in Hymenoptera? William Hamilton (1972) proposed that the unusual genetic system ants, wasps, and bees predisposes them to eusociality. Hymenoptera hive Table 11.2 Sociality in insects ‘This table summarizes the taxonomic distribution of eusociality in insects Species are called “primitively eusocial if queens are not morphologically differentiated from other individuals. Order Family Subfamily Eusocial species Hymenoptera Anthophoridae In seven genera (carpenter bees) Apidae Apinae (honeybees) highly cusocial species Bombinae (bumble bees) 300 primitively eusocial Euglossinae (orchid bees) None Meliponinae (stingless bees) 200 eusocial species Halictidae (sweat bees) In six genera Sphecidae (sphecoid wasps) In one genus Vespidac (paper wasps, _Polistinae ‘Over 500 species all soe yellow jackets) Stenogastrinae ‘Some primitively eusocial species Vespinae ‘About 80 species all ets Formicidae (ants) 11 subfamilies ‘Over 8,800 described: Many all eusocial or descended ‘eusocial species ‘other families None Isoptera (termites) Nine families All species (over 2288) Homoptera (plant bugs) Pemphigidae Sterile soldiers found ini Coleoptera (beetles) Curculionidae Austroplatypus incomp Thysanoptera (thrips) _Phlaeothripidae Subfertile soldier are fa Source: From Crozier and Panto (1996). in OncorhripsChapter 11 WBitice charac nusual fo: n of sex determination: Males are haploid and females are tap in gen {iploid. Males develop from unfertilized eggs; females de egu5.As a result of this system, called haplodiploidy, ants, bees, and re more closely related to their sisters than they are to their own off- This follows because sisters share all of the iol lop from fertilized al nd (@) specialized [a variety of insect 2000; Duty et al spring es they inherited from their father, which is half their genome, and half the genes they inherited fiom their mother (the colony's queen), which is the other half of their {section we con= groups: the Hy= genome. Thus, the probability that homologous alleles in hymenopteran sis fers are identical by descent is (1 x 1/2) + (1/2 X 1/2) = 3/4. To their vn offspring, how females are related by the usual r of 1/2. This unique system favors the production of reproductive sisters over daughters, sons, or | brothers. (Females are related to their brothers by r = 1/4; see Figure 11.12) Imore like a cell ftctive interests, Thus, females will maximize their inclusive fitness by acting as workers rather than as reproductives (Hamilton 1972). Specifically, their alleles will increase in the population Tica: a faster when they invest in the production of sisters rather thin producing their own offspring. This is the haplodiploidy hypothesis for tasks they per= 7 the evolution of eusociality in Hymenoptera, not genetically Kewism? Why is WIN) Ting she Haplodiplotdy Hypothesis In addition to offering an explanation for why workers prefer to invest in sisters, Iather than their own offipring, the haplodiploidy hypothesis predicts that workers Prefer to invest in sisters over brothers. Because their r with sisters is 3/4 and only W/4 with brothers, workers should favor a 3:1 female-biased sex ratio in reproduc~ fic system of Jptera have an fit offipring (meaning, offspring that are not destined to become sterile workers, Br soldiers; Trivers and Hare 1976; but see Alexander and Sherman 1977). eeens, in contrast, are equally related to their sons and daughters and should favor BU: sex ratio in the reproductives produced (see Box 11.3). The fitness interests of Horkers and queens are not the same. The question is: Who wins the conflict? Do Gueens or workers control the sex ratio of reproductive offiprin Liselotte Sundstrm and coworkers (1996) set out ¢o answer this question by species nec Gktermining the sex ratio of reproductive offspring in wood ant (Formica exsecta) Holonies. They found that queens laid a roughly equal number of male and fe Mule eggs, but that sex ratios were heavily female-biased at hatching. To make Setse ofthis result, the researchers hypothesize that workers are able to determine Hie sex of eggs and that they selectively destroy male offspring, Based on results from similar studies, most researchers acknowledge that fe- fle-biased sex allocation is widespread an asocial hymenoptera. (For re- Bless, sce: Bourke and Franks 1995; Crozier and Pamilo 1996; Chapuisat and Reler 1999; Sundstrém and Boomsma 2001.) In the tug-of-war over the fitness Biisests, workers appear to have the upper hand over queens. Pethaps the more important general message of this work, however, is that Hilonies of ants, bees, Bymmecry in relationship between queens and offipring versus workers and off- Bpiing produces a sharp conflict of interest Dues the Haplodiploidy Hypothesis Explain Eusociality? The prediction and affirmation of 3:1 sex ratios in reproductive offipring, at least Brome hymenopterans, confirms that the haplodiploid system of sex determina- Hon has a strong effect on how workers behave. But is haplodiploidy the reason Kin Selection and Social Behavior 433 Mother Father (diploid) (hapiois) ©. 10 | >< Sister Brother Figure 11.12 Haplodiploidy produces un- usual coefficients of relation- ship The arrows describe the paths by which genes can be identical by descent in hy: menopterans. Note that there is ‘no path of shared descent be- tween sisters and brothers through their father, because males have no father.434 PART Ill Adaptation Box 11.3. The evolution of the sex ratio n many species, the sex ratio at hatching, tion, or birth is 1:1, Ronald Fisher (1930) explained why this should be so. Fisher pointed out that if one sex is in short supply in a population, then an allele that leads to the production of the rarer sex will be favored. This is because individuals of the rarer sex will have more than one mate on average when they mature, simply because in a sexual species every indi vidual has one mother and one father. Members of the rarer sex will experience increased reproductive ative to individuals of the more common sex. Indeed, whenever the sex ratio var from 1:1 selection favoring the rarer sex will exist until the ratio returns to unity. Fisher’ explanation is a classic example of frequency-dependent selection—a con- cept we introduced in Chapter 5 with the example of yellow and purple Elderflower orchids Fisher's argument is based on an mportant as- sumption: that parents invest equally in each sex When one sex is more costly than the other, parents should adjust the sex ratio to even out the invest- ment in each. For this reason evolutic ary biologists distinguish the numerical sex ratio from the invest- ment sex ratio and speak, in general terms, about the issue of sex allocation (Charnov 1982) that so many hym ter ing tha the female workers in the colo true, Multiple mating is com Honey Haplodiploidy affects the behavior of eusocial are to their own jortant fa ading t efficient of relatedness of 0, Third, many on the phylogeny of the females that produce only enou; fertilization of their sisters, resulting in a sem with a stron, the answer is no. There are several reasons for this. First, the prediction that workers favor the production of sisters ove duction of their own offspring is based on an important assumption fl .¢ queens, for example, mate an average of 17.25 times before a colony (Pag: and Metcalf 1982). As a result, it is common to find thal age coefficient of relatedness among honeybee workers is under 1/3 ( pecies, 31.1997, 1 998). In these colonies, workers are not more closely related (0) ters than th Second, in many species more than one queen is active in founding If they have neither parent in common, then workers in thes al species are not haplodiploid, and many hapl species are not eusocial, Nonreproductive castes are found in all termite for example, even though termites are diploid and have a normal eh system of sex determination (Thorne 1997} And although eusociality mon among hymenopterans, it is by no means universal. Reviewing re hymenoptera will help drive this last point hom Robert Trivers and Dan Willard (1973) with an important extension to Fisher's mo, suggested that when females are in good ph cal condition and are bett young, and when dif able to care in the cond young are sustained into adulthood, then they preferentially invest in male offipring. This isk differences in condition affect male reprot success RS more than female RS (see Chapi This predictior called condition-dependent § location, has been confirmed in a wide varie mammals, including humans. (For examples, Clutton-Brock et al. 1984; Betzig and Turke 198 A third prominent result in sex-ratio theory to William Hamilton (1967). In insects that a eggs in fruit or other insects, the young offen develop, and mate inside the host. Frequently are parasitized by a single female. Given this tion, Hamilton realized that selection should males to female bias. This phenomenon as local mate competition, ‘nas been observed i riety of parasitic inse 1967; Werren 1984 s. (For examples, see Hi species are cusocial? Most researchers ate ony have the same father. In many species hi pups of eusocial Hym offspring, colonies436 PARTIII Adaptation In paper wasps, reproductive altruiom ia facultative. Females can choose between helping at a nest or breeding on their own. importantly, Hunt noted that eusociality only evolved in groups that build plex nests and that care for their larvae for extended periods. ‘The association between nest building, care of larvae, and because it suggests that the primary agent favoring reproductive altruism in inse ecological in nature—not genetic as proposed by the haplodiploidy hypothesis logic here is similar to the “best-of-a-bad-job” explanation for helping behav birds reviewed in Section 11.1. Nest building and the need to supply larvae wi continuous supply of food make it difficult or impossible for a female to bree her own (ee Alexander et al. 1991). Also, when predation rates are high but yo are dependent on parental care for a long period, then individuals who breed al are unlikely to survive long enough to bring their young to adulthood (Q 1989; Queller and Strassmann 1998). In short, to exphin the evolution of eusod ty we clearly need to consider ecological factors that affect Band C in Hani inequality, as well as genetic factors that dictate t Facultative Strategies in Paper Wasps Paper wasps in the genus Polistes have been an especially productive group fo search into the costs and benefits of reproductive altruism. Unlike workers soldiers in ants and termites, paper wasp workers are not sterile. Instead of obligate helpers, Polistes females are capable of reproducing on their own contrast is important. To achieve reproductive success, worker and soldi and termites have no choice but to assist relatives—the nutrition they reed larvae guarantees that they are sterile. But in paper wasps, females have the of helping relatives or breeding on their own. In Polistes dominulus, Peter Nonacs and Hudson Reeve (1995) found thi males pursue one of three distinct strategies: They either initiate their own join a nest as a helper, o wait for a breeding opportunity. Each option isa ated with costs and benefit. What are the costs and benefits of founding a nest? In the population Nonacs and Reeve studied, nests were founded by single females or by mill ‘male groups. Earlier studies had shown that single foundresses are ata distng advantage compared to multifemale coalitions. Adult mortality is high, add with multiple foundresses are less likely to fail because surviving females keg nest going, Nonacs and Reeve also found that multifoundress coalitions ae likely to renest after a nest is destroyed. When they analyzed 106 instances failure due to predation or experimental removal, they determined that onl 4 single foundresses rebuilt while 21 of 51 multifoundress groups did Although the success rate of multifemale nests is hi compared i gle-foundress nests, multifemale coalitions are not free of conflicts. Fight tween wasps are decided by body size. As the graph in Figure 11.14 multifoundress nests grew fastest when there was a large difference in the size of the dominant female and her subordinate helpers. To interpret this Nonacs and Reeve suggest that productivity is low in coalitions where bod is similar because subordinate females frequently challenge for control ofthe and the right to lay most of the eggs. Why would females join a coalition and help rear ofpring that ate not own? Subordinates gain indirect fitness benefits because they are usuChapter 11 Kin Selection and Social Behavior 437 Hups that build cot t PBocialty is impor . rere altruism in insect . loidy hypothesis.T Si helping behavion E supply larvae witht Bes : female to breed @ aie / Be ich buc sa Figure 11.14 In paper wasps, the success of female Bint coalitions varies This graph plots the growth rate of Polistes ool? dlominulus nests as a function ofthe size difference between 003005 O10 ons 020 distinguish his is coopa The critical Pry are wary, however, and hag between dividuals in the population, either kin selection or reciprocal altruism could op- tte in this system. Wilkinson was able to show that both occur. Over the Dthe study, he witnessed 110 episodes of regurgitation. Seventy-seven of these sto docua biologists nine one off In this sys= bism can be b 2 lowered) piyitation (Figure 1 impire bats Ibasic social { offspring during the ‘total of 14 ture, many Brecks, a Tkely to v ives who are frequent roostmates Because the degree of relatedness and degree of asso. any given night. By studying weight loss in captive bats when food wa Wilkinson was able to show that bats who go three consecutive teil are likely to starve to death. Here between mother and child and are simply examples of parental care. In 21 lfthe remaining 33 cases, Wilkinson knew both the rand the degree of associa fion between the actor and beneficiary and could examine the effect of both Saribles. Wilkinson discovered that both degree of relatedness and degre Heciation have a statistically significant effect in predicting the probability of re~ 1.22) Chapter 11. Kin Selection and Social Behavior 447 search for large mammals—primarily horses and cattle—that can provide a meal ‘of young bats and 7% of adults fail to feed on withheld, without a ation varied among in- Vampire bats sharing blood meale. They usually share with clo are roostmates and may late Bats do not regurgitate blood meals to one another ih captivity, withheld food from a different individual inndomly. They are much more likely to regurgitate to relatives and to nonrela- To confirm that bats actually do reciprocate, Wilkinson held nine individuals ch night for sever tated to whom over the course of the experi- the null hypothesis that hungry individuals received Fampire bats are reciprocal altruists. Territory Defense in Lions Allpossible donors ©, r «4 alll i Regurgitators al oh alla 0125456789 ‘Assocation @ oups. These groups, called prides, consist of three to six related females, their ® @ bal “ll ] | 2 ot ti tt OTESTS Regrsiatore Hood randomly from cagemates. Instead, hungry individuals were much more ive blood from an individual they had fed before. This confirms that lions (Panthers leo; Figure 11.23) are the only species of cat that lives in social pring, and a coalition of males, Males are often related to one another but are Figure 11.22 Association, relatedness, and altruism in vampire bats These histograms plot the total number of bat-bat pairs at Gerald Wilkinson's study site versus their (a) degree of association forall potential blood donors in the roost, (b) degree of association for blood: sharing pairs who are not related at mother-offspring, (©) coefficient of relatedness for all potential blood donors in the roost, and (d) coefficient of relatedness for blood-shar- ing pairs who were not related as mother-offspring. The visual impression in these figures is that regurgitators are more likely to be related and more lik than the general population. This is confirmed by a stats tical procedure called a stepwise logistic regression. Both relatedness and association affect the probability of blood:sharing. From Wilkinson (1984).448 PART Ill Adaptation Figure 11.23 African lions When female lions hear roars made by un- familiar females, they move across thei territory toward the source. Females that lead the movement frequently glance back at female pride members that are lagging behind, unrelated to pride females (Packer and Pusey 1982; Packer et al. 1991). Males operate in defending themselves against other coalitions of males that attack a try to take over the pride. If a pride takeover occurs, the incoming males o kill the young cubs present (see Chapter 10). Females cooperate in defending their young against infanticidal attacks, nursing young, in hunting prey that are difficult to capture (females do the ¥ majority of hunting in lions), and in defending the pride’ territory against cursions by females in neighboring prides (Packer and Pusey 1983, 1997).Bal with intruding females are dangerous, especially if pride females do not coop ate in defense, Solitary lions are often killed in same-sex encounters. To study cooperation during incursions by strange females who threaten th territory, Robert Heinsohn and Craig Packer (1995) placed speakers near th edge of pride territories and played tape recordings of roars given by unfuml females. Pride females respond to these roars by approaching the speakers even attack if a stuffed female lion is placed near the speaker. They also do not bituate to this stimulus—they continue responding when the experiment is peated. As a result, Heinsohn and Packer were able to quantify how fe responded to threats over multiple trials. In addition to recording how long took an individual to reach the midpoint between the pride’ original posit and the speaker, they also calculated the difference between each individ time-to-midpoint and the leader’ time, documented the order within the that each female reached the midpoint, and counted the number of glancesa male made back at lagging pride members. They collected data on female & sponses in eight different prides. Statistical analyses showed significant differences in the strength of respon among females within prides. Some females in each pride always led; others ways lagged behind. Other females adopted conditional strategies: They woul assist the lead female(s) more frequently when Heinsohn and Packer pl tapes of more than one female roaring (that is, when the threat to the pride’Chapter 11_kin Selection and Social Behavior Fister) Still others would lag behind more when multiple roars were played. Bese differences were not correlated with the age or body size of females or btheir coefficient of relationship with other pride members. These results are paradoxical in the context of both inclusive fitness and reci- Bity theory Why do the leaders tolerate the laggards? According to the theo Ine developed earlier, they should be punished. But leaders were never Benved to threaten laggards or to withhold benefits. For example, they did not rds caught up. They glanced back and appeared to recog- The dynam up defense in lions have yet to be Bp leading until la fie that laggards were IAlack of dominance and coercion among females is a general feature of society. Packer, Anne Pusey, and Lynn Eberly (2001) analyzed an extensive Basct on female reproductive success. They found no evidence that the vari explained by kin selection or ing, but continued approaching the speaker altruiom, fe in reproductive success among females within prides was due to any other than chance. This is in contrast to many other cooperatively ding animals, in which dominant individuals do most of the breeding and mlinates provide assistance The simple answer is that we do not know how a mixture of altruist and Breiprocator strategies can persist among lions. There are a number of pos ies. A leading hypothesis, stil to be tested, is that females who contribute Wp territory defense reciprocate in other ways—through exceptional ng prowess or milk production, for example. Another idea is that even ly lions are so well armed in tooth and claw that it is just too d Punish cheaters. Finally, having even a laggard for an ally may be better than ig no ally at all. If so, tolerating a certain amount of social parasitism may superior strategy to chasing the offender away. Clearly, social interactions Bing lions are complex. Individuals interact, four outcomes are possible fh respect to fitness: cooperation, altruism, selfish- and spite. The evolution of altruism was one of igrat paradoxes of evolutionary biology until it frsolved by two important advances filam Hamilton showed mathematically that a for altruism will spread when Br— C > 0 ere B is the benefit to the recipient in units of Biviving offipring, ris the coefficient of relatedness, een actor and recipient, and C is the cost to actor. When Hamilton's rule holds, kin selection Bills in altruistic behavior bert Trivers developed the theory of reciprocal diuism. Altruism phe if the Bpient are large and the cost co the actor is small, mong unrelated individuals can benefits of an altruistic act to the re- and the benefits are later returned to the actor by the recipient Kin selection explains a great variety of phenomena such as alarm calling in ground squirrels and helping, behavior in birds. In Belding’ ground squirrels, indi- viduals are more willing to risk giving an alarm call if close relatives are nearby. In white-fronted bee-eaters, 1-year-old individuals preferentially help at the nests of closely related individuals (often their parents). A variety of other behaviors are likewise illuminat- ed by kin selection. It explains why cannibalistic tad- poles spit out their siblings, why wood mouse sperm sacrifice themselves to help their brothers win the big and why coots count their eggs. Kin selection has been important in explaining the evolution of eusociality in Hymenoptera and in