Inclusive Fitness

How inclusive fitness works

Classical Darwinian fitness counts personal offspring. A bird that fledges three chicks has fitness 3; the bird next door that fledges five has fitness 5. The allele in the more successful bird wins the long game. This is enough for most cases of evolution — animals that eat well, escape predators, attract mates, and produce viable young pass on their genes. The framework breaks, however, the moment we look at organisms that systematically forgo personal reproduction to help others. A sterile worker bee has Darwinian fitness zero and yet there are 30,000 of them in every colony, so the alleles encoding that worker phenotype are clearly successful at something. The alleles must be propagating somehow.

Hamilton's 1964 insight was to generalise the bookkeeping. From the gene's point of view, what matters is total expected copies in the next generation, regardless of which body produces them. A copy of the worker-caste allele propagates through (a) any personal offspring the worker happens to produce and (b) any relatives whose offspring exist because of the worker's help, weighted by the probability they carry the same allele. Sum those two streams and you get inclusive fitness — the quantity that selection actually maximises. Personal fitness is the special case where part (b) is zero.

The pieces are:

• Direct fitness. Personal offspring, the same quantity as Darwinian fitness.
• Indirect fitness. Extra offspring produced by relatives because of help the focal individual provided, each weighted by the coefficient of relatedness r between focal and relative. The "extra" matters: relatives reproduce on their own; you only get credit for the increment your help adds.
• Inclusive fitness. Direct + indirect. The quantity selection acts on when the population includes interactions between relatives.

If you fledge two chicks of your own (direct = 2) and your help raises a sister's clutch from three to five chicks (extra = 2, r = 0.5, indirect = 1.0), your inclusive fitness is 3.0. A sister who fledged five chicks unaided has direct fitness 5, indirect 0, inclusive 5. Compare strategies, and selection picks the higher inclusive number.

Where Hamilton's rule comes from

Hamilton's rule rB > C is a one-line consequence of inclusive-fitness arithmetic. Suppose an allele encodes an altruistic act that costs the donor C personal offspring and confers B extra offspring on a relative of relatedness r. The donor's direct fitness drops by C; her indirect fitness rises by rB. The change in her inclusive fitness is rB − C. The allele rises in frequency exactly when this is positive — when rB > C. The rule is not a postulate; it falls out of the inclusive-fitness accounting whenever costs and benefits combine linearly and selection is weak enough not to break the assumptions.

The same logic explains why the rule is not always true. When costs and benefits interact non-linearly, when populations are small enough that drift dominates, when the same allele has different fitness effects in different ecological contexts (synergistic effects), the simple inequality must be replaced by more general expressions — but the underlying inclusive-fitness logic still applies. The Price equation (1970) is the most general formal tool, and it has Hamilton's rule as a special case.

Direct, indirect, and inclusive fitness compared

	Direct fitness	Indirect fitness	Inclusive fitness
Definition	Personal offspring count	Σ r · (extra relatives' offspring)	Direct + indirect
Same as classical Darwinian fitness?	Yes	No — extension	Strict generalisation
Captures altruism toward kin?	No — invisible	Yes	Yes
Captures help to non-kin (reciprocity)?	Indirectly through later return	No	No — needs separate framework
What selection maximises	In solitary species	Component, not full quantity	In structured populations with kin interactions
Worker bee value	≈ 0	High (sisters at r = 0.75)	High
Reproductive queen bee value	Very high	Modest (mates' relatives are unrelated to her)	Very high (dominated by direct)
Solitary bird value	Equals offspring count	≈ 0	≈ direct fitness
Helper at the nest value	Low or zero	r × extra siblings raised	Sum, may exceed independent breeding

The numerical example a textbook usually walks through is the helper at the nest. A young Florida scrub jay has two options: leave the natal territory and try to set up a breeding pair (direct fitness path, but most attempts fail because every territory is already held); or stay home and help parents raise siblings (indirect fitness path, with r = 0.5 to siblings). When habitat is saturated and dispersal is dangerous, helping yields higher inclusive fitness even though direct fitness is zero — and that is exactly what young scrub jays do. The numerical match between predicted and observed helping rates (Woolfenden & Fitzpatrick 1984, decades-long study) is one of the cleanest empirical tests of inclusive-fitness theory.

Worked example — meerkat sentinel duty and helpers

Meerkats (Suricata suricatta) live in family groups of 10–30 with one breeding pair and many non-breeding adults of both sexes. The non-breeders take turns standing sentinel — perched on an exposed mound, scanning for predators while the rest of the group forages — and provide pup care, grooming, allomarking, and territorial defence. Each behaviour costs personal foraging time but benefits group members.

The inclusive-fitness accounting goes like this. Most non-breeding meerkats are full siblings or half-siblings of the current pups (r = 0.5 or 0.25) and full siblings of the breeding female herself. Stephens et al. (2005) measured pup survival as a function of helper number and found each additional helper increased the litter's survival probability by about 4%. With litter size of 4 and r = 0.5 between helper and pups, an extra 0.16 surviving pup adds 0.08 to indirect fitness. Multiply across multiple litters per year, sum across the helper's career, and you get a non-trivial indirect-fitness component. Meanwhile, the cost of helping (foraging time lost, predation exposure on sentinel duty) is real but small relative to the inclusive-fitness gain. Helping pays.

The model also predicts when helping should taper off. Older helpers, with shrinking expected lifespans, gain less from indirect investment because their personal reproductive future is also shrinking; the optimal split should shift toward attempting independent breeding. Empirically, Clutton-Brock's long-term Kalahari study found exactly that age-related reduction in helping effort. Inclusive fitness is not just a story; it predicts measurable shifts in behaviour as ecological and demographic parameters change.

Real-world cases of inclusive fitness in action

• Honeybee workers. Direct fitness ≈ 0; indirect fitness via 0.75-related sisters dominates. The textbook eusocial example.
• Florida scrub jays. Helpers at the nest forgo independent breeding to raise siblings; observed helping rates match the inclusive-fitness optimum given habitat saturation.
• Meerkats. Sentinel duty, pup care, and allonursing by full or half-siblings; per-helper survival increase ~4%.
• Naked mole rats. Single breeding queen, hundreds of non-breeding workers, mean r ~0.8 due to inbreeding; inclusive-fitness arithmetic explains the only known eusocial mammal alongside Damaraland mole rats.
• White-fronted bee-eaters. Helpers preferentially help full siblings, then half-siblings, then cousins, in r-weighted order — direct empirical support for relatedness as the deciding variable.
• Long-tailed tits. Failed breeders join relatives' nests as helpers; Hatchwell's long-term study showed helping is conditional on availability of relatives.
• Microbial cooperation. Public goods in Pseudomonas aeruginosa biofilms (siderophore production) follow Hamilton's rule; cheats invade when r drops, cooperation prevails when populations are spatially structured.
• Slime moulds (Dictyostelium). Stalk vs spore-cell allocation in multi-clone fruiting bodies. Cheaters arise; populations resist them through kin-discrimination at the tgr locus.

Variants and refinements

• Neighbour-modulated fitness. Maynard Smith's reformulation: instead of crediting a focal individual with extras conferred on relatives, credit each individual's offspring partly to themselves and partly to neighbours weighted by r. Mathematically equivalent in the linear case.
• Greenbeard models. Inclusive fitness without family-tree relatedness — recognition is via the cooperation allele itself. Explored theoretically by Dawkins, demonstrated empirically in fire-ant Gp-9 and slime-mould csaA.
• Indirect reciprocity. Reputation-based cooperation (Nowak & Sigmund 1998) — separate mechanism, not inclusive fitness, but often filed alongside.
• Group augmentation. Help benefits the helper directly via larger future group size (more sentinels, more foragers). Adds a direct-fitness component to behaviours that look purely altruistic.
• Inclusive fitness in microbes. Without classical sex, relatedness is measured by clonal identity. Same arithmetic, different routes to high r (clonal patches, biofilm spatial structure).

Common pitfalls

• "Inclusive fitness is the same as personal fitness plus offspring." No — only the extra offspring conferred on relatives count, not their total reproduction. Otherwise you'd double-count any reproduction relatives would have done anyway.
• "Inclusive fitness ignores environment." It doesn't. Costs and benefits depend on ecology, demography, and the marginal value of each act of help. The same r yields different inclusive-fitness payoffs in different ecological contexts.
• "Hamilton's rule must hold exactly." The rule is exact only under linear-additive cost/benefit and weak selection. Non-linearities, synergies, and strong selection require generalised forms. The framework still applies; the simple rB > C is an approximation, and a remarkably good one.
• "Inclusive fitness has been replaced by multilevel selection." No — they are equivalent in the standard case. Choose the one that makes the empirical question easier. Most working biologists use inclusive fitness because the parameters (r, B, C) are measurable.
• "Indirect fitness implies organisms make calculations." No more than "maximise expected lifetime offspring" implies that animals do calculus. Selection has tuned heuristics to approximate the optimum on average. The arithmetic happens at the population level, in allele-frequency dynamics.
• "If altruism evolves, organisms will be perfectly cooperative." Not even close. Cheating is always tempting at the individual level; cooperation is sustained by relatedness, recognition, policing, and punishment. Conflict and cooperation coexist within every social group.