TREE vol. 14, no. 2 February 1999 0169-5347/99/$ ? see front matter ? 1999 Elsevier Science. All rights reserved. PII: S0169-5347(98)01529-8 49
From the algae that helppower reef-building corals,to the diverse array of pol-linators that mediate sex-
ual reproduction in many plant
species, to the myriad nutritional
symbionts that fix nitrogen and
aid digestion, and even down to
the mitochondria found in nearly
all eukaryotes, mutualisms are
ubiquitous, often ecologically
dominant, and profoundly influen-
tial at all levels of biological organ-
ization1?6. Although mutualisms
can be simply defined as recipro-
cally beneficial relationships be-
tween organisms, they range from
diffuse and indirect interactions
to highly integrated and coevolved
associations between pairs of spe-
cies. Such mutualisms usually in-
volve the direct exchange of goods
and services (e.g. food defense and
transport) and typically result in
the acquisition of novel capabil-
ities by at least one partner2,3.
Current theory5?8 suggests that mutualisms are best
viewed as reciprocal exploitations that nonetheless pro-
vide net benefits to each partner. This view stresses the
disruptive potential of conflicts of interests among the
erstwhile partners. Consequently, identifying factors that
influence the costs and benefits to each partner and quan-
tifying their influence constitute primary research objec-
tives9. In particular, inquiry centers on the description of
conflicts of interest between partners and the attempt to
understand what mediates them10. This requires a clear
appreciation of the spatial, temporal and taxonomic con-
text in which these systems operate. Breakthroughs in
understanding have, and will, come precisely because of
the increased attention paid to the different ecological and
evolutionary scales within which the mutualisms function.
The expanding availability of a wide range of molecular
data has produced qualitative leaps in the types of infor-
mation available to researchers. This information can be
usefully combined with the results from field and laboratory
studies. For example, genetic characterization of mutualists
has facilitated the unambiguous determination of the num-
ber and identity of interactants (e.g. genotypes and spe-
cies), the degree and scale of their specificities and their
patterns of ecological transmission11?15. Similar approaches
can also reveal the phylogenetic patterns of relationships
both between and within taxa of mutualists, and thus the
extent to which speciation in hosts is tracked by speciation
in symbionts16?19, as well as the number of origins of particu-
lar types of relationship11,16. Results from these studies
have a direct bearing on one of the
central evolutionary questions
concerning mutualism: what fac-
tors align the interests of part-
ners so that the relationships re-
main mutually beneficial and
evolutionarily stable?
Current theory of conflict,
cooperation and constraint
The potential for conflicts of
interest to shape or destabilize
mutualistic associations will de-
pend on the extent to which the
survival and reproductive inter-
ests of the symbiont align with
those of the host. Given that con-
flicts of interest can occur even
within the genomes of single indi-
viduals5,6,20, it seems unlikely that
the interests of mutualists will
ever be completely concordant.
Although there is no general
theory of mutualism, several fac-
tors that can help align mutual-
ists? interests have been identified.
The passage of symbionts from parent to offspring (verti-
cal transmission), genotypic uniformity of symbionts within
individual hosts, spatial structure of populations leading
to repeated interactions between would-be mutualists, and
restricted options outside the relationship for both part-
ners are thought to align interests and promote long-term
stability. Conversely, movement of symbionts between un-
related hosts (horizontal transmission), multiple symbi-
ont genotypes and varied options are thought to unravel
them5?8,21?23. This framework is logically appealing, and many
cases appear to conform well with its predictions24,25.
However, it is worth scrutinizing why these factors are
thought to reduce the potential for conflict among would-be
mutualists and noting that those factors are often not inde-
pendent. First, in the case of vertical transmission, both
symbiont and host benefit from successful reproduction by
the host. Second, vertical transmission over many gener-
ations will tend to reduce the genetic diversity of symbionts
by eliminating novel inputs to the symbiont community and
by providing a potential bottleneck at each generation11.
The resulting genetic homogeneity of symbionts within a
host reduces selection for traits that increase between-
symbiont competitive ability to the detriment of the host?s
wellbeing and reproductive success5,6,23,25. Finally, vertical
transmission implies a continual interaction between host
and symbiont lineages. The absence of an independent
phase in a symbiont?s life cycle facilitates the evolution of
complete dependence, which reduces the evolutionary
viability of nonsymbiotic alternatives over the long term.
REVIEWS
The evolution of mutualisms: 
exploring the paths between conflict
and cooperation
E.A. Herre, N. Knowlton, U.G. Mueller and S.A. Rehner
Mutualisms are of fundamental importance
in all ecosystems but their very existence
poses a series of challenging evolutionary
questions. Recently, the application of
molecular analyses combined with
theoretical advances have transformed our
understanding of many specific systems,
thereby contributing to the possibility of a
more general understanding of the factors
that influence mutualisms.
E.A. Herre and N. Knowlton are at the Smithsonian
Tropical Research Institute, Apartado 2072, Balboa,
Republic of Panama (herrea@gamboa.si.edu); 
N. Knowlton is also at the Scripps Institute of
Oceanography, University of California, San Diego, 
La Jolla, CA 92093-0202, USA (nknowlton@ucsd.edu); 
U.G. Mueller is at the Dept of Zoology, University of
Maryland, College Park, MD 20742, USA
(um3@umail.umd.edu); S.A. Rehner is at the Dept of
Biology, PO Box 23360, University of Puerto Rico,
Rio Piedras, San Juan, PR 00931, USA
(attaboy@hotmail.com).
50 TREE vol. 14, no. 2 February 1999
Nonetheless, not all mutualisms follow this pattern of
vertical transmission with its proposed benefits. For exam-
ple, many marine symbionts (Box 1) and mutualist associ-
ates of plants [e.g. pollinators (Box 2) and mycorrhizae]
are horizontally transmitted, yet they are usually clearly
beneficial. Moreover, vertical transmission does not guar-
antee benevolence (Box 3). Given these exceptions, it is
important to determine the extent to which real systems
conform to these patterns, and what factors are most
responsible for determining conformity where it exists.
Identifying the players
Determining the number and identities of the partici-
pants in mutualistic associations is a necessary first step
for any evolutionary analysis, but it can be a surprisingly
nontrivial task. Hosts and symbionts often lose characters
found in their closest free-living relatives, or gain novel
characters, making them difficult to distinguish and char-
acterize taxonomically. The traditional solution for bac-
terial and fungal symbionts has been culturing. However,
in some symbioses, what is successfully cultured does not
necessarily reflect the actual community present in intact
associations; and in other systems, symbionts cannot pres-
ently be cultured11,26?28. For these reasons, molecular analy-
ses have played a critical role both in genetically charac-
terizing isolated mutualists and in screening assemblages
directly to assess the nature of symbiont communities.
The resulting discoveries of stunning and unexpected di-
versity have transformed our understanding of mutualisms
involving corals (Box 1), leaf-cutter ants (Box 4), and root
symbionts26,27,29, among others.
It is important to appreciate that symbiont diversity,
cryptic and otherwise, can occur at different levels. At the
level of different host species, different hosts can contain
morphologically indistinguishable symbionts that are never-
theless quite distinctive both genetically and functionally.
At the level of different individual hosts within a species,
genetically different symbionts can be found in association
with different host individuals (or populations). Even within
individual host organisms, several distinct symbionts can
be found12,26?28. The recognition that individual hosts can
act as landscapes for communities of potentially competing
symbionts (Box 1) raises the question of why competition
among symbionts does not destabilize the mutualism, much
as competition among parasites is believed to result in se-
lection for increased virulence23,25. The ecological flexibility
provided by symbiont diversity28,30 might play an important
counterbalancing role.
Patterns of ecological transmission and evolutionary
association
For patterns of transmission, it is useful to distinguish be-
tween transmission over ecological (generation to gener-
ation) and longer evolutionary (lineage to lineage) time-
scales. For example, systems dominated by strict vertical
ecological transmission might be expected to produce con-
cordant phylogenies between host and symbiont at all taxo-
nomic scales, whereas in systems dominated by horizontal
transmission, this outcome might be thought to be less
likely.
The explosion of systematic analyses using molecular
techniques has generated phylogenetic reconstructions for
one or both members of several speciose groups of mutual-
ists. These studies show that patterns of transmission over
ecological timescales do not necessarily translate into simi-
lar patterns at evolutionary timescales; available evidence
suggests that all combinations of different patterns of eco-
logical transmission and different degrees of phylogenetic
concordance are found. Specifically, there are cases in
which both evolutionary and ecological transmission appear
to be predominantly vertical18. However, vertical evolution-
ary transmission (between lineages) is also found in cases in
which ecological transmission is predominately horizontal
(e.g. fig-pollinating wasps19, luminescent bacteria associated
with deep-sea fish31 and sulfur oxidizing bacteria and some of
their bivalve hosts14,32), apparently because vertical trans-
mission is not the only mechanism that promotes cospeci-
ation. Moreover, many intracellular bacteria (e.g. Wolbachia,
REVIEWS
Box 1. Marine invertebrates and photosynthetic algae: 
the ecological significance of symbiont diversity
Throughout the shallow tropical oceans, sessile animals often have symbiotic
associations with photosynthetic, single-celled algae. Among the most spectacu-
lar and ecologically important are the associations formed between reef-building
corals and dinoflagellates of the genus Symbiodinium. For many years, these
symbionts were considered to be a single species, but physiological and genetic
studies11,28 have revealed enormous, previously unsuspected, diversity. What
was once considered a single species is now recognized as a group with at least
three clades that, by extrapolation to free-living forms, are distinct at the family or
ordinal level. These studies also revealed that there was no obvious concor-
dance between host and symbiont phylogenies.
Despite the growing appreciation of this cryptic diversity, it remained widely
assumed that any single host formed an association with only one type of sym-
biont. However, in several ecologically dominant corals, it is now known that a
single coral species and even single colonies are capable of hosting two or more
types of symbiont28. Zonation of symbionts across the reef and within colonies
appears to be related to levels of light. During adverse conditions, such as
unusually high temperature, the mutualism between corals and algae can break
down (?coral bleaching?) in complex patterns that reflect this zonation. Thus, from
the alga?s perspective, the host is more like a landscape composed of more and
less suitable conditions than a uniformly hospitable environment28.
From the coral?s perspective, horizontal transmission and complex mixtures of
symbionts might provide short-term ecological flexibility to cope with fluctuating
physical conditions that outweighs the possible costs of evolutionary conflicts
among symbionts28. Many of the themes emerging from these studies of corals
characterize other symbiotic systems as well11,26,27,29,35.
Box 2. Figs and yuccas: model systems for understanding
evolutionary conflicts
There are over 700 species of figs (Ficus) described worldwide. The figs depend
on minute pollinator wasps (Agaonidae) for continued sexual reproduction, and
the wasps depend on the figs to complete their life cycle. Fossil evidence indi-
cates that this relationship dates back at least 40 million years. In most cases,
the relationship is overwhelmingly species-specific. In addition, recent molecular
work suggests that the long evolutionary history of figs and their pollinators has
been dominated by cospeciation between the two taxa19.
Although in the long term the two mutualists depend completely upon one
another, their reproductive interests are not identical. The fig benefits both from
the production of viable seeds and from the production of female pollinator
wasps that will potentially transfer the tree?s pollen to produce seeds in other
trees. The wasps benefit only from the production of offspring (that necessarily
come at the expense of approximately 50% of the potentially viable seeds). 
What prevents the shorter lived and much more numerous wasps from exploiting 
an ever greater number of seeds is still unanswered9,45. However, for most
aspects that have been studied, the tree appears to be largely in control of 
the system9,45.
It is interesting to compare the fig-wasp system with the yucca-moth system.
Although there is the general dependence in both cases, there are instructive dif-
ferences. The reproductive interests of individual female wasps are much more
closely linked to their host than is the case with the moths, because the wasps
tend to be trapped within the inflorescence they pollinate. Moreover, the female
wasp offspring will carry pollen from the inflorescence in which they developed. In
contrast, moths can pollinate and lay eggs in several different flowers, and their
offspring are unlikely to provide the additional pollination service because they
drop to the ground and emerge as adults much later10,38,39. The difference
between the figs and yuccas in the degree to which their interests coincide with
their partners is probably reflected in the much higher proportion of the fig seeds
that support development of wasp offspring compared with the proportion of
yucca seeds that support the development of the moth offspring.
TREE vol. 14, no. 2 February 1999 51
Box 3) show predominantly vertical transmission patterns
at an ecological level, but this does not necessarily translate
into phylogenetic patterns that are concordant with their
hosts33. Presumably, this is because of sporadic cases of
horizontal transfer between distantly related species.
In an additional complexity, determining the extent to
which co-cladogenesis is occurring will frequently depend
on the taxonomic scale at which the question is asked14. For
example, the phylogenetic relationships between some lin-
eages of leaf-cutter ants show nearly perfect concordance
with the relationships of their associated fungi. However,
in some entire lineages the host phylogenetic relationships
show essentially no correspondence with those of the fungi.
In fact, there appear to be many lineages in which nonspeci-
ficity and noncongruence are the rule16,17 (Box 4). Unfortu-
nately, for most mutualisms, we do not have adequate spa-
tial and taxonomic sampling to determine the extent of
concordance between host and symbiont lineages.
Trajectories of costs and benefits
Molecular data can provide a window on the taxonomic
identities of mutualists, the structuring of their extant popu-
lations (e.g. patterns of spatial distribution and ecological
transmission), their histories of phylogenetic associations
and their evolutionary origins14,28,29,34, but provide relatively
little information about the often rapid, and sometimes
convoluted, evolutionary trajectories of costs and benefits
received4,8.
From studies that compare outcomes across several
populations of mutualistic interactions between two spe-
cies, we know that outcomes can vary among extant popu-
lations4,35,36. Several studies have documented that net costs
and benefits can vary over relatively short timescales4,36
resulting from: (1) changes in the presence or abundance
of influential third parties36,37; (2) variation in host densities
that results in shifts in patterns of transmission24; (3) vari-
ation in resource availablility3,36; or (4) variation in physical
conditions (Box 1). Furthermore, such studies raise ques-
tions concerning the degree of local adaptation in host 
and symbiont populations, such as whether hosts gener-
ally benefit most from local, presumably more highly co-
adapted symbionts.
Moreover, in evolutionary time, comparisons across
related taxa (particularly in cospeciating systems) can show
different evolutionary outcomes that represent variations
on a single theme of mutualistic interaction (e.g. leaf-
cutters, figs, yuccas, ants, plants and lycaenid butterflies).
Specifically, phylogenetic analyses reveal that parasitic 
lineages can be embedded in largely mutualistic groups
and/or vice versa19,38,39. However, theory suggests that the
species that parasitize mutualisms should not be the closest
relatives to either partner38,39. Available evidence collected
from figs (Ficus) and fig wasps (Agaonidae), and the yuccas
(Yucca) and yucca moths (Tegeticula), supports this pre-
diction19,38,39. Nonetheless, this proposition requires fur-
ther testing.
Mutualisms as model systems
Mutualisms and rates of molecular evolution
In those instances in which the host and mutualist co-
speciate, the absolute times of divergence between pairs of
cospeciating mutualists are effectively held constant. This
allows a series of potentially instructive comparisons to be
made in the accumulation of substitutions in homologous
DNA sequences. First, comparisons can be made between
the accumulation of substitutions at a given gene or set of
genes in the ?host? and in the ?symbiont? (or parasite).
Second, comparisons can be made between the rates of
accumulation of base changes between the symbionts and
their free-living relatives.
Depending on the attributes of the taxa available, these
comparisons permit the evaluation of several factors that
have been suggested to be important in influencing the rates
of molecular evolution. Cospeciating mutualists often exhibit
different generation times, different body sizes and meta-
bolic rates, different effective population sizes and different
degrees of sexual reproduction. Different taxa might also
possess very different systems of DNA repair. These con-
trasts can be productively exploited. For example, Moran and
colleagues have found that the aphid-associated Buchnera
shows much faster rates of molecular evolution than do its
REVIEWS
Box 3. Wolbachia and Buchnera: the implications of
horizontal versus vertical transmission for the 
evolution of mutualism
Theory suggests that vertical transmission selects for more benign relationships, and
that symbionts transmitted vertically should generally have benign or even positive
effects on their hosts. There is accumulating experimental and comparative support for
this proposition. A classic example is the association found between aphids and their
bacteria (Buchnera) that synthesize necessary amino acids for their hosts3,11,18,40.
However, Wolbachia appears to be a maternally inherited endosymbiont that
frequently has large negative effects on its host?s reproductive interests. At
times, the bacteria distort the host?s sex ratio, often leading to all female broods,
or produce reproductive incompatibility with other host individuals that do not
carry the same strain of Wolbachia33. Superficially, these observations contradict
the theoretical predictions.
However, to assess the relevance of these observations, the timescales over
which maternal transmission occurs and the magnitude of the negative effects of
Wolbachia must be considered. Although most cases show that at an ecological
timescale Wolbachia is transmitted vertically, there is clear phylogenetic evidence
that Wolbachia ?jumps? from lineage to lineage; that is, whether its propagation is
considered to be dominated by vertical transmission depends on temporal scale. In
addition, Wolbachia can often have complex or little, if any, negative effect on its di-
rect individual host44. Critical questions involve determining the actual routes and
frequencies of horizontal transmission, as well as the magnitude of negative effects
under real ecological situations, and then determining if there is a correspondence
between ?how bad the bugs are? and ?how much evolutionary jumping they can do?.
Box 4. Fungus-growing ants and their fungi: 
phylogenetic transitions in patterns of symbiont acquisition
The exclusively New World fungus-gardening ants in the tribe Attini (Formicidae) com-
prise over 200 described species, all obligately dependent upon the cultivation of
fungus for food16,17,46. Ants in the leaf-cutter genera Acromyrmex and Atta are eco-
logically and economically important because of the vast quantities of foliage and
flowers that they cut to culture the fungi in their often immense nests. Together with
three additional genera, leaf-cutter ants are grouped into the monophyletic higher
attines, which comprise about one-half of the species diversity of the tribe. Ants in
the remaining seven genera of lower attines are less conspicuous, frequently cryp-
tic and do not attack plants. The symbiotic associations of lower attine ants and
their fungi are diverse: some species grow their fungi entirely on dead vegetable
matter, some entirely on caterpillar frass and others on a mixed substrate that
can even include seeds.
Molecular data have been decisive in identifying the evolutionary origins and phylo-
genetic relationships of attine fungal symbionts. First, although most ant-associated
fungi are members of the family Lepiotaceae (Agaricales; Basidiomycotina), phylo-
genetic analyses based on ribosomal DNA indicate that the fungus cultivated by
several ant species in the lower attine genus Apterostigma is distantly related to
all other attine fungi, and has been secondarily acquired long after the mutualism
originated in the Amazon Basin approximately 50 million years ago16,17,46. Sec-
ond, molecular analyses indicate that several distinct lepiotoid fungal lineages
associated with lower attines are essentially identical to current free-living forms.
Together with the apparent lack of morphological modification of many lower
attine symbionts, these observations suggest the recent acquisition of novel
symbionts from free-living stock46. Thus, as can be observed on both ecological
and evolutionary scales, the presumably ancestral condition of repeatedly acquir-
ing free-living fungi has been retained in some of the lower attines but appears to
have been lost in the higher attines, which have developed longer-term associ-
ations with their generally more specialized symbionts.
52 TREE vol. 14, no. 2 February 1999
free living relatives, an observation that appears to oppose
the idea that rates of evolution in mutualists should slow
down18,40. A similar pattern has been found in lichens41.
Mutualisms and the adaptive significance of sex
Current theory regarding the adaptive value of sexual
reproduction revolves around the ideas that sexual repro-
duction serves to: (1) maintain adaptation in the face of a
constantly changing and potentially threatening biotic
world and (2) remove deleterious mutations. Potentially,
comparisons between groups of related species character-
ized either with or without sexual reproduction could be
useful to assess the relative importance of these two pro-
posed functions. For example, some groups of mutualists,
such as dinoflagellates associated with marine inver-
tebrates, fungi associated with attine ants, perhaps algae in
some lichens, clavicipitaceous (i.e. smut-like) grass endo-
phytes, and the fungal cultivars of fungus-gardening ter-
mites, are derived from free-living groups capable of both
sexual and asexual reproduction. In each case, it appears
that the balance between sexuality and asexuality has
been shifted towards the latter. Interestingly, in the case of
the endophytic fungi associated with grasses, the fungi
appear to reduce the host?s tendency to reproduce sexu-
ally42, rather than the more typical reverse pattern43.
There are several possible explanations for these pat-
terns. For example, one school of thought suggests that ?well
integrated? (e.g. intracelluar) symbionts are protected by
their hosts from a menacing organic world of constantly
evolving predators and parasites, and consequently do not
?need? sex43. An alternative, less benign, view of mutualisms
suggests that mutualistic relationships are better charac-
terized as a series of ongoing arms races. In this scenario,
sex might be the critical element that allows one member
to ?keep up?, or if suppressed in one member has allowed
the other to ?get ahead?. Further progress in this area will
depend on knowing the extent to which sex is actually ab-
sent, determining whether symbionts are represented by a
single clone or are genetically heterogeneous, and estimat-
ing the phylogenies of the partners over various spatial
and taxonomic scales. Ultimately, molecular data will play
a crucial role in distinguishing among various possible
interpretations.
Conclusions
Most organisms are involved either directly or indi-
rectly in mutualistic interactions. However, there is no gen-
eral theory of mutualism that approaches the explanatory
power that ?Hamilton?s Rule? appears to hold for the under-
standing of within-species interactions. Underlying prob-
lems revolve around explicitly defining vague terms, such
as ?alignment of interest?, and employing biologically re-
alistic currencies (i.e costs and benefits) at biologically rel-
evant scales of organization. Ideally, all of these should be
measurable and capable of being employed across radi-
cally different systems. For example, can the ?conflict of
interest? and ?costs and benefits? within and between leaf-
cutters that do or do not have vertically transmitted fungi be
estimated and then compared with those values for corals
that do or do not have vertically transmitted algae? We have
implied that factors constraining ?cheating? or ?defection?
are increasingly required because the interests of interact-
ing species are not aligned. But can it be shown that in-
creasingly stringent constraints (e.g. no options outside the
relationship and/or increased host investment in symbiont
control) operate in systems in which there are increasingly
incongruent interests?
Ultimately, we cannot begin to determine whether there
are any general principles or consistent patterns that char-
acterize mutualisms if we misunderstand individual case
studies. Ideally, for a number of cases, we need to identify
and quantify the costs and benefits to each party, and to
understand what factors influence variation in those costs
and benefits. Importantly, we need to understand conflicts
of interest and attempt to identify what factors maintain
the alignment of interests. If there is nonalignment, what
prevents the system from breaking down? To do this, it is
crucial that we identify the mutualists, and understand
their diversity, patterns of transmission and structuring 
at several spatial, temporal and evolutionary scales. This
poses the monumental task of documenting basic, de-
scriptive natural history for many distinct systems and
coupling it with the often indispensable information that
can increasingly be obtained from molecular approaches.
Acknowledgements
We thank Koos Boomsma and Jack Werren for stimulating
discussion. We thank Betsy Arnold, Jenny Apple, Egbert
Leigh, Elisabeth Kalko, Sadie Jane Ryan, Andy Dobson, 
Jon Howe, Penny Barnes, Andrew Baker, Rob Rowan,
DeWayne Shoemaker and Rod Page for help and useful
comments during the evolution of this article. STRI Post
Doctoral Fellowships supported SAR and UGM and made
this collaboration possible.
References
1 Boucher, D.H., ed. (1985) The Biology of Mutualism: Ecology and
Evolution, Oxford University Press
2 Margulis, L. and Fester, R., eds (1991) Symbiosis as a Source of
Evolutionary Innovation, MIT Press
3 Douglas, A.E. (1994) Symbiotic Interactions, Oxford Science Publications
4 Thompson, J.N. (1994) The Coevolutionary Process, University of
Chicago Press
5 Maynard Smith, J. and Szathm?ry, E. (1995) The Major Transitions in
Evolution, W.H. Freeman
6 Leigh, E.G., Jr and Rowell, T.E. (1995) The evolution of mutualism
and other forms of harmony at various levels of biological
organization, Ecologie 26, 131?158
7 Nowak, M.A., Bonhoeffer, S. and May, R.M. (1994) Spatial games and
the maintenance of cooperation, Proc. Natl. Acad. Sci. U. S. A. 91,
4877?4881
8 Doebeli, M. and Knowlton, N. (1998) The evolution of interspecific
mutualisms, Proc. Natl. Acad. Sci. U. S. A. 95, 8676?8680
9 Herre, E.A. and West, S.A. (1997) Conflict of interest in a mutualism:
documenting the elusive fig?wasp?seed tradeoff, Proc. R. Soc.
London Ser. B 264, 1501?1507
10 Pellmyr, O. and Huth, C.J. (1994) Evolutionary stability of
mutualism between yuccas and yucca moths, Nature 372, 257?260
11 Douglas, A.E. (1996) The ecology of symbiotic micro-organisms,
Adv. Ecol. Res. 26, 69?103
12 Okuma, M. and Kudo, T. (1996) Phylogenetic diversity of intestinal
bacterial communities in the termite, Reticulitermes speratus,
Appl. Environ. Microbiol. 62, 461?468
13 Askoy, S., Chen, X. and Hypsa, V. (1997) Phylogeny and potential
transmission routes of midgut-associated endosymbionts of tsetse
(Diptera: Glossinidae), Insect Mol. Biol. 6, 183?190
14 Krueger, D.M. and Cavanaugh, C.M. (1997) Phylogenetic diversity of
bacterial symbionts of Solemya hosts based on comparative sequence
analysis of 16s rRNA genes, Appl. Environ. Microbiol. 63, 91?98
15 Nason, J.D., Herre, E.A. and Hamrick, J.L. (1998) The breeding
structure of a tropical keystone plant species, Nature 391, 685?687
16 Chapela, I.H. et al. (1994) Evolutionary history of the symbiosis
between fungus-growing ants and their fungi, Science 266, 1691?1694
17 Hinkle, G. et al. (1994) Phylogeny of the attine ant fungi based on
analysis of small subunit ribosomal RNA gene sequences, Science
266, 1695?1697
18 Moran, N.A., von Dohlen, C.D. and Baumann, P. (1995) Faster
evolutionary rates in endosymbiotic bacteria than in cospeciating
insect hosts, J. Mol. Evol. 41, 727?731
19 Herre, E.A. et al. (1996) Molecular phylogenies of figs and their
pollinating wasps, J. Biogeogr. 23, 521?530
REVIEWS
TREE vol. 14, no. 2 February 1999 0169-5347/99/$ ? see front matter ? 1999 Elsevier Science. All rights reserved. PII: S0169-5347(98)01545-6 53
The study of cooperativebreeding in vertebratesaims to understand whysome animals forgo inde-
pendent reproduction and help
others to breed instead. Over the
past 30 years, the field has devel-
oped a rich set of theory1?3 and
has been wracked by some major
debates4,5. However, enough co-
operative species have been stud-
ied in detail to establish common
ground and to test theory. Indeed,
in a recent review of the field,
Emlen6 states that ?the original
paradox of cooperative breeding
largely disappeared with the wide-
spread confirmation that (1) help-
ers frequently do improve their
chances of becoming breeders?,
and (2) they frequently do obtain
large indirect genetic benefits by
helping to rear collateral kin?. With
identification of these direct and
indirect benefits to helpers, the original questions asked
by researchers would appear to be ?largely answered?.
Despite this claim, some important questions remain un-
answered. In particular, our understanding of the varying
level of helper contributions within and between species
remains poor. The approach to
cooperative breeding has often
been to compare the outcomes of
philopatry and helping with the
other options of dispersing to
float or dispersing to breed3. Evalu-
ation of the final reproductive
rewards for each strategy leads to
an ultimate understanding of why
a particular decision was made7.
Implicit in this approach is that
the outcome reflects all the costs
and benefits of dispersal versus
nondispersal, and helping versus
nonhelping, but it does not lead
to an appreciation of the nature of
each cost and benefit. Although
we have a large list of benefits to
helping8, we still lack a cohesive
framework that explains when
they apply in various taxa or eco-
logical circumstances. Less atten-
tion has been paid to the costs of
helping.
Consider the cooperatively breeding Seychelles warbler,
Acrocephalus sechellensis. In an elegant study, Komdeur9
showed that helpers much prefer to feed nestlings that are
more closely related to themselves; an important result
that emphasized the lability and adaptive nature of helping
REVIEWS
The cost of helping
Robert Heinsohn and Sarah Legge
Cooperative breeding in mammals, birds and
fish has provided evolutionary biologists
with a rich framework for studying the
causes and consequences of group-based
reproduction. Helping behaviour is
especially enigmatic because it often
entails an individual sacrificing personal
reproduction while assisting others in their
breeding attempts. The decision to help
others to reproduce is affected by
immediate and future costs analogous to
those of direct reproduction, but these
components of the equation have usually
been neglected. Recent research suggests
that the type of benefit sought could
determine the extent of help given.
Robert Heinsohn and Sarah Legge are in the Division of
Botany and Zoology, Australian National University,
Canberra, ACT 0200, Australia
(robert.heinsohn@anu.edu.au; sarah.legge@anu.edu.au).
20 Eberhard, W.G. (1980) Evolutionary consequences of intracellular
organelle competition, Q. Rev. Biol. 55, 231?249
21 Trivers, R.L. (1971) The evolution of reciprocal altruism, Q. Rev.
Biol. 46, 35?57
22 Axelrod, R. and Hamilton, W.D. (1981) The evolution of cooperation,
Science 211, 1390?1396
23 Frank, S.A. (1994) Kin selection and virulence in the evolution of
protocells and parasites, Proc. R. Soc. London Ser. B 258, 153?161
24 Bull, J.J., Molineux, I.J. and Rice, W.R. (1991) Selection on
benevolence in a host?parasite system, Evolution 45, 875?882
25 Herre, E.A. (1995) Factors affecting the evolution of virulence:
nematode parasites as a case study, Parasitology 111, 179?191
26 Perotto, S. et al. (1996) Molecular diversity of fungi from ericoid
mycorrhizal roots, Mol. Ecol. 5, 123?131
27 Sanders, I.R., Clapp, J.P. and Wiemken, A. (1996) The genetic
diversity of arbuscular mycorrhizal fungi in natural ecosystems ? 
a key to understanding the ecology and functioning of the
mycorrhizal symbiosis, New Phytol. 133, 123?134
28 Rowan, R. (1998) Diversity and ecology of zooxanthellae on coral
reefs, J. Phycol. 34, 407?417
29 Young, J.P.W. and Haukka, K.E. (1996) Diversity and phylogeny of
rhizobia, New Phytol. 133, 87?94
30 Wulff, J.L. (1997) Mutualisms among species of coral reef sponges,
Ecology 78, 146?159
31 Haygood, M.G. (1993) Light organ symbioses in fishes, Crit. Rev.
Microbiol. 19, 191?216
32 Distel, D.L., Felbeck, H. and Cavanaugh, C.M. (1994) Evidence for
phylogenetic congruence among sulfur-oxidizing
chemoautotrophic bacterial endosymbionts and their bivalve
hosts, J. Mol. Evol. 38, 533?542
33 Werren, J.H. and O?Neil, S.L. (1998) The evolution of heritable
symbionts, in Influential Passengers: Inherited Microorganisms and
Arthropod Reproduction (O?Neil, S.L., Hoffman, A.A. and Werren, J.H.,
eds), pp. 1?47, Oxford University Press
34 Gargas, A. et al. (1995) Multiple origins of lichen symbioses in fungi
suggested by SSU rDNA phylogeny, Science 268, 1492?1495
35 Wilkinson, H.H., Spoerke, J.M. and Parker, M.A. (1996) Divergence in
symbiotic compatibility in a legume?Bradyrhisobium mutualism,
Evolution 50, 1470?1477
36 Bronstein, J.L. (1994) Conditional outcomes in mutualistic
interactions, Trends Ecol. Evol. 9, 214?217
37 Gaume, L., McKey, D. and Terrin, S. (1998) Ant?plant?homopteran
mutualism: how the third partner affects the interaction between a
plant specialist ant and its myrmecophyte host, Proc. R. Soc.
London Ser. B. 265, 569?575
38 Pellmyr, O. et al. (1996) Evolution of pollination and mutualism in
the yucca moth lineage, Am. Nat. 148, 827?847
39 Pellmyr, O., Leebens-Mack, J. and Huth, C.J. (1996) Non-mutualist yucca
moths and their evolutionary consequences, Nature 380, 155?156
40 Moran, N. (1996) Accelerated evolution and Muller?s rachet 
in endosymbiotic bacteria, Proc. Natl. Acad. Sci. U. S. A. 93,
2873?2878
41 Lutzoni, F. and Pagel, M. (1997) Accelerated evolution as a
consequence of transitions to mutualism, Proc. Natl. Acad. Sci.
U. S. A. 94, 11422?11427
42 Clay, K. (1990) Fungal endophytes of grasses, Annu. Rev. Ecol. Syst.
21, 275?297
43 Law, R. and Lewis, D.H. (1983) Biotic environments and the
maintenance of sex ? some evidence from mutualistic symbioses,
Biol. J. Linn. Soc. 20, 249?276
44 Wade, M.J. and Chang, N.W. (1995) Increased male fertility in
Tribolium confusum beetles after infection with the intracellular
parasite Wolbachia, Nature 373, 72?74
45 Nefdt, R.J.C. and Compton, S.G. (1996) Regulation of seed and
pollinator production in the fig?fig wasp mutualism, J. Anim. Ecol.
65, 170?182
46 Mueller, U.G., Rehner, S.A. and Schultz, T.R. (1998) The evolution of
agriculture in ants, Science 281, 2034?2038