Evolution of Agriculture in Insects

19 Aug 2005 15:44 AR ANRV259-ES36-24.
tex XMLPublishSM (2004/02/24) P1: OJO

AR REVIEWS IN ADVANCE10.1146/annurev.ecolsys.36.102003.152614
(Some corrections may occur before final publication online and in print)
V I E W
E
S
Annu. Rev. Ecol. Evol. Syst. 2005. 36:563–95
C E
I N
doi: 10.1146/annurev.ecolsys.36.102003.152614
Copyright c 2005 by Annual Reviews. All rights reserved
N
A
D V A
THE EVOLUTION OF AGRICULTURE IN INSECTS

Ulrich G. Mueller,1,2 Nicole M. Gerardo,1,2,3
Duur K. Aanen,4 Diana L. Six,5 and Ted R. Schultz6
1
Section of Integrative Biology, University of Texas at Austin, Austin, Texas 78712;
email: umueller@mail.utexas.edu
2
Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Republic of Panama
3
Department of Ecology and Evolutionary Biology, University of Arizona, Tucson,
Arizona 85721-0088; email: ngerardo@email.arizona.edu
4
Department of Population Biology, Biological Institute, University of Copenhagen,
2100 Copenhagen, Denmark; email: dkaanen@bi.ku.dk
5
Department of Ecosystem and Conservation Sciences, University of Montana,
Missoula, Montana 59812; email: diana.six@cfc.umt.edu
6
Department of Entomology, National Museum of Natural History,
Smithsonian Institution, Washington, District of Columbia. 20013-7012;
email: schultz@lab.si.edu
Key Words Attini, Macrotermitinae, mutualism, symbiosis, Xyleborini

■ Abstract Agriculture has evolved independently in three insect orders: once in
ants, once in termites, and seven times in ambrosia beetles. Although these insect
farmers are in some ways quite different from each other, in many more ways they
are remarkably similar, suggesting convergent evolution. All propagate their cultivars
as clonal monocultures within their nests and, in most cases, clonally across many
farmer generations as well. Long-term clonal monoculture presents special problems
for disease control, but insect farmers have evolved a combination of strategies to man-
age crop diseases: They (a) sequester their gardens from the environment; (b) monitor
gardens intensively, controlling pathogens early in disease outbreaks; (c) occasion-
ally access population-level reservoirs of genetically variable cultivars, even while
propagating clonal monocultures across many farmer generations; and (d ) manage, in
addition to the primary cultivars, an array of “auxiliary” microbes providing disease
suppression and other services. Rather than growing a single cultivar solely for nutri-
tion, insect farmers appear to cultivate, and possibly “artificially select” for, integrated
crop-microbe consortia. Indeed, crop domestication in the context of coevolving and
codomesticated microbial consortia may explain the 50-million year old agricultural
success of insect farmers.
1. INTRODUCTION
The cultivation of crops for nourishment has evolved only a few times in the
animal kingdom. The most prominent and unambiguous examples include the
fungus-growing ants, the fungus-growing termites, the ambrosia beetles and, of,
1543-592X/05/1215-0563$20.00 563
19 Aug 2005 15:44 AR ANRV259-ES36-24.tex XMLPublishSM (2004/02/24) P1: OJO
564 MUELLER ET AL.
course humans. For humans, who started the transition from an ancestral hunter-
gatherer existence to farming only about 10,000 years ago (Diamond 1997, Smith
1998), sustainable, high-yield agriculture has become critical for survival in a
global economy with projected food shortages, and diverse research programs are
currently devoted to the optimization of agricultural productivity in the context of
growing environmental challenges (Green et al. 2005). Agricultural progress has
been achieved by humans through a combination of insight, creative planning, and
a fair share of contingency and luck (Diamond 1997, Schultz et al. 2005, Smith
1998). However, humans have so far not examined nonhuman agricultural systems,
such as the fungus-growing insects, for possible insights to improve agricultural
strategies.
This lack of an applied interest in insect agriculture probably derives from a
general perception that human agricultural systems (based largely on plant cul-
tivation) function in a fundamentally different manner than insect systems (all
based on fungus cultivation). However, humans have learned much of practical
value through the close examination of adaptive features of other organisms (in-
cluding insects), and comparable problems such as crop diseases affect all farmers
regardless of their phylogenetic positions or those of their crops (plant, fungus,
or otherwise). Because of the universality of crop diseases in both human and
insect agriculture, it may be fruitful to examine the short-term and long-term solu-
tions that have evolved convergently in insect agriculture for possible application
to human agriculture (Denison et al. 2003). Such a synthesis is the goal of this
review.
1.1. Behavioral and Nutritional Elements

Defining Agriculture
Insect fungiculture and human farming share the defining features of agriculture
(see Table 1): (a) habitual planting (“inoculation”) of sessile (nonmobile) cul-
tivars in particular habitats or on particular substrates, including the seeding of
new gardens with crop propagules (seeds, cuttings, or inocula) that are selected
by the farmers from mature (“ripe”) gardens and transferred to novel gardens; (b)
cultivation aimed at the improvement of growth conditions for the crop (e.g., ma-
nuring; regulation of temperature, moisture, or humidity), or protection of the crop
against herbivores/fungivores, parasites, or diseases; (c) harvesting of the cultivar
for food; and (d ) obligate (in insects) or effectively obligate (in humans) nutri-
tional dependency on the crop. Obligate dependencies of the insect farmers can be
readily demonstrated by experimental removal of their cultivated crops, resulting
in reduced reproductive output, increased mortality, or even the certain death of
the cultivar-deprived insect (Francke-Grosmann 1967, Grassé 1959, Norris 1972,
Sands 1956, Weber 1972). Our definition of agriculture does not require conscious
intent in planting and harvesting. Conscious planning, learning, and teaching have
clearly accelerated the development of complex agriculture in humans, but pre-
sumably not in insects (Schultz et al. 2005).
TABLE 1 Agricultural behaviors of farming ants, termites, beetles, and humans
19 Aug 2005 15:44
Agriculture in:
Macrotermitine Xyleborine ambrosia

AR
Agricultural behavior Attine ants termites beetles Humans
Dependency on crop for food Obligate Obligate Obligate Facultative

Engineering of optimal growth conditions for crop (e.g., Present Present Present Present
substrate preparation, moisture, or humidity regulation)
Planting of crop on improved substrate Present Present Present Present
Intensive, continuous monitoring of growth and disease Present Present Present Absent
status of all crops
ANRV259-ES36-24.tex
Sustainable harvesting of crop for food Present Present Present Present

Protection of crop from diseases and consumers Present Present Present Present
Weeding of alien organisms invading the garden Present Present Unknown Present
Use of chemical herbicides to combat pests Present Unknown Unknown Present
Use of microbial symbionts for nutrient procurement for Present Unknown Unknown Present
the crop
Use of disease-suppressant microbes for biological Present Unknown Unknown Absent1
pest control
Sociality Strictly eusocial Strictly eusocial Subsocial or communal2 Social
Task partitioning in agricultural processes Present Unknown Present Present
XMLPublishSM (2004/02/24)
Application of artificial selection for crop improvement Unknown, but ant Unknown Unknown Present
sexert symbiont
choice
P1: OJO
Learning and cultural transmission of agricultural Absent Absent Absent Present

EVOLUTION OF AGRICULTURE IN INSECTS
innovations
1
See Section 7 for some recently discovered microbes with potential disease-suppressant properties.
2
One ambrosia beetle species is eusocial, all other species appear to be subsocial or communal (see text).
565
566 MUELLER ET AL.
We restrict our review to ant, termite, and beetle fungiculturists. Cases analo-
gous to human animal husbandry, such as the tending by ants of hemipteran insects
(e.g., aphids, treehoppers; Hölldobler & Wilson 1990), are beyond the scope of this
review. We also exclude cases that fail to meet all four of the requirements of agri-
culture as defined above, including, e.g., the ant Lasius fuliginosus, which promotes
fungal growth in the walls of its nest, because the fungus is apparently not grown for
food but instead for strengthening the walls (Maschwitz & Hölldobler 1970) or for
antibiotic protection of the walls (Mueller 2002). On the same grounds, we exclude
a number of possible cases of incipient agriculture. For example, Littoria snails
may “protofarm” fungi by creating plant wounds that become infected with fungal
growth that is part of the snails’ diet, but the snails do not actively inoculate the
plant wounds or otherwise garden the fungi (Silliman et al. 2003). Many more such
protofarming species probably await discovery, particularly among invertebrates,
and all of the known insect agriculturists (fungus-growing ants, termites, and bee-
tles) probably originated from comparable protoagricultural ancestors (Mueller
et al. 2001, Schultz et al. 2005). Comparison of these protofarming insects with
“primitive” human agriculture exceeds the scope of this review.
1.2. A Coevolutionary Approach to

Understanding Agriculture
We will analyze agriculture as a type of strong coevolutionary interaction, defined
by the nutritional and behavioral criteria summarized above, in which natural se-
lection acts upon both farmers and crops as reciprocally interdependent lineages
(Futuyma & Slatkin 1983, Rindos 1984). Our coevolutionary approach to agri-
culture considers not only the interactions between a specific farming insect and
a single cultivated crop, but also its interactions with other pathogenic and mu-
tualistic microbes that have recently been discovered in insect gardens. Like the
cultivars, some of these microbes are also managed by the insect farmers for spe-
cific purposes (Figure 1, see color insert). In other words, an insect garden is
not a pure monoculture, but a sequestered and engineered ecological community
consisting of several interacting microbes, some beneficial and others detrimen-
tal to the farmers. To gain a comprehensive understanding of the principles of
insect agriculture, it will therefore be necessary to examine the nature of insect-
microbe interactions in gardens, the evolutionary origins of these interactions, and
the convergent and divergent evolutionary trajectories that culminated in the extant
agricultural systems of insects.
2. THE THREE INSECT-AGRICULTURE SYSTEMS

Behaviorally complex systems of insect agriculture are known from only three
groups of insects: ant, termites, and beetles.
EVOLUTION OF AGRICULTURE IN INSECTS 567
2.1. Ant Fungiculture

The fungus-growing ants are a monophyletic group of about 220 described and
many more undescribed species in the tribe Attini (subfamily Myrmicinae) (Price
et al. 2003, Schultz & Meier 1995). Attine ants occur only in the New World
(Argentina to the southern United States) and attain their greatest diversity in the
wet forests of equatorial South America, the region of their presumed evolutionary
origin (Mueller et al. 2001). Attine ants are obligate agriculturists; their cultivated
fungi are the sole source of food for the larvae and an important source of food for
the adults. Although adults are able to supplement their diets by feeding on plant
juices (Bass & Cherrett 1995, Murakami & Higashi 1997), the cultivated fungi
are nutritionally sufficient to support the ants even in the absence of additional
nutrients (Mueller 2002, Mueller et al. 2001). Garden fungi are transmitted verti-
cally across generations when daughter queens transport small pellets of natal-nest
mycelium within their infrabuccal pockets, pouches present in the mouthparts of
all ants (Fernández-Marı́n et al. 2004, Huber 1905, Mueller 2002). In the derived
leafcutter ants, the workers are divided into a remarkable range of differently sized
morphological castes, each specialized on a different task (Hart et al. 2002, Weber
1972) (Figure 2d, see color insert).
Different attine ant lineages cultivate their fungi on different substrates. The
ancestral gardening substrate, still used by the so-called lower attines, consists of
flower parts, arthropod frass, seeds, wood fragments, or other similar plant debris,
whereas the leafcutting genera Atta and Acromyrmex primarily use freshly cut
leaves and flowers. Despite these distinct substrate specializations, all attine sys-
tems contain at least four symbionts: (a) the fungus-growing ants; (b) their fungal
cultivars (basidiomycetes in the mushroom families Lepiotaceae and Pterulaceae;
Mueller et al. 1998, Munkacsi et al. 2004); (c) mutualistic antibiotic-producing
actinomycete bacteria (family Pseudonocardiaceae; Currie et al. 1999b); and (d )
garden parasites in the ascomycete fungal genus Escovopsis (Currie et al. 1999a,
Currie et al. 2003b). Additional bacteria and yeasts also occur in attine gardens
and may function as mutualists, e.g., by secreting digestive enzymes or antibotics
(Carreiro et al. 1997, Craven et al. 1970, Santos et al. 2004).
2.2. Termite Fungiculture

Of the more than 2600 described termite species, about 330 species in the subfamily
Macrotermitinae cultivate a specialized fungus, genus Termitomyces, for food.
Nests are generally founded by a single pair of reproductives, the future queen and
king. They seal themselves permanently in a cell of hard clay (the so-called royal
chamber) where they rear the first brood of sterile workers. In most termite species,
a new colony acquires a fungal strain from wind-dispersed sexual Termitomyces
spores shortly after nest founding and begins construction of the first gardens (De
Fine Licht et al. 2005). These spores come from fruiting bodies (mushrooms)
that arise from mature termite colonies. The fruiting of the fungus appears to be
roughly synchronized to the period when the first foraging workers emerge from a
568 MUELLER ET AL.
new nest, a few months after the nest-founding stage. Termite gardens are grown
on dead plant material that is only partially decomposed, such as leaf litter, dead
grass, dead wood, or dry leaves.
Termite gardens are built from spore-containing fecal pellets in chambers that
the termites construct either inside a mound or dispersed in the soil. Fecal pellets are
added continuously to the top of the comb and fungal mycelium rapidly permeates
the new substrate (Figure 2e). After a few weeks, the fungus starts to produce veg-
etative nodules that are consumed by the termites. These nodules are a rich source
of nitrogen, sugars, and enzymes. The nodules are also covered with indigestible
asexual spores (conidia), so that consumption serves the additional function of
inoculating the feces with spores, which pass through the gut unharmed and are
then planted in new comb with the deposition of feces (Leuthold et al. 1989).
Mature comb is also consumed (Darlington 1994), but it is nutritionally inferior
to the nodules.
2.3. Beetle Fungiculture

Ambrosia beetles make up around 3400 of the 7500 species in the weevil subfamily
Scolytinae (the “bark beetles,” including the traditionally separate Platypodinae;
Farrell et al. 2001, Harrington 2005, Wood 1982). Most ambrosia beetles con-
struct tunnel systems (galleries; Figure 2f ) in woody tissues of trees (typically
in weakened or recently dead trees or, more rarely, in vigorous hosts), although
some species are specialized to colonize pith, large seeds, fruits, and leaf petioles
(Harrington 2005, Wood 1982). The term ambrosia refers to the fungi cultivated by
the beetles on gallery walls, upon which they feed as an exclusive, or near exclusive,
food source. The beetles are obligately dependent upon the fungi, from which they
acquire essential vitamins, amino acids, and sterols (Beaver 1989, Kok et al. 1970).
The most advanced fungiculturists among the ambrosia beetles occur in the
Xyleborini, a large monophyletic tribe of about 1300 species (Farrell et al. 2001,
Jordal 2002). It is this group of ambrosia beetles that we primarily focus on in this
review. Although life histories among the Xyleborini vary considerably, most share
a number of fungicultural characteristics. There is a sexual division of labor in the
Xyleborini; only females perform gardening tasks, whereas males are short-lived
and flightless (Norris 1979). After mating, females disperse to new host substrate,
carrying the fungi in specialized pockets termed mycangia. Once within a new
host, founding females “plant” the fungi on the walls of the excavated tunnels,
lay eggs, and tend the resulting garden and brood (Norris 1979). In ways not fully
understood, they are able to control the growth of the fungal crop, as well as, to
a degree, the composition of its multiple fungal species (Beaver 1989, French &
Roeper 1972, Kingsolver & Norris 1977, Roeper et al. 1980). If the female dies, the
garden is quickly overrun by contaminating fungi and bacteria, which ultimately
results in the death of the brood (Borden 1988, Norris 1979).
The ambrosia gardens of xyleborine beetles are not pure monocultures as was
once believed, but are typically composed of an assemblage of mycelial fungi,
yeasts, and bacteria (Batra 1966, Haanstadt & Norris 1985). These assemblages
were termed multi-species complexes by Norris (1965), who suggested that it is a
complex as a whole, rather than any one individual microbe, that allows the beetles
to exploit nutrient-poor substrates such as wood. However, most subsequent work
has revealed that one “primary” fungus always dominates in beetle gardens (Baker
1963, Batra 1966, Gebhardt et al. 2004, Kinuura 1995). Furthermore, the beetles
typically carry only the primary fungus in the mycangium (although secondary
fungi are sometimes also isolated from mycangia), and the cultivation efforts of
female beetles tend to favor the primary fungus, which imparts the greatest nu-
tritional benefit (Francke-Grosmann 1967, Gebhardt et al. 2004, Morelet 1998,
Norris 1979). Some auxiliary fungi also support beetle development, but survival
on the auxiliary fungi alone is often greatly reduced (Norris 1979). These obser-
vations implicate the primary fungus as the intended crop, whereas the secondary
fungi, yeasts, and bacteria may be contaminant “weeds” or may play additional
auxiliary roles in the gardens, paralleling the hypothesized roles of the auxiliary
bacteria and yeasts in attine gardens (see above).
3. EVOLUTIONARY ORIGINS OF INSECT AGRICULTURE

Phylogenetic analyses reveal nine independent origins of insect agriculture
(Figure 2; Table 2). In ants, fungal cultivation arose only once, probably 45–65 Mya
in the Amazon rainforest (Mueller et al. 2001; Schultz & Meier 1995). In termites,
fungiculture likewise had a single origin, approximately 24–34 Mya in the African
rainforest (Aanen et al. 2002; D.K. Aanen & P. Eggleton, submitted). In ambrosia
beetles, however, agriculture arose independently seven times between 20–60 Mya,
six times in various nonxyleborine lineages, and once in the ancestor of the Xyle-
borini about 30–40 Mya (Farrell et al. 2001). Whereas the common ancestors of
the macrotermitines and of the xyleborines each domesticated a single, specific
primary cultivar clade to which their descendants have adhered throughout subse-
quent evolution (Figure 2, Table 2), attine ants maintain associations with multiple
independently domesticated cultivar lineages (which are for the most part very
closely related; Mueller et al. 1998; Munkacsi et al. 2004; Table 2). Interestingly,
there are no known cases of reversal from agricultural to nonagricultural life in any
of the nine agricultural insect lineages (Figure 2a, 2b, and 2c), suggesting that the
transition to fungiculture is a drastic and possibly irreversible change that greatly
constrains subsequent evolution.
Two main models have been suggested for the independent evolutionary tran-
sitions to agriculture in insects, the “consumption-first” versus the “transmission-
first” models (Mueller et al. 2001). In the consumption-first model (the likely
model for the termites), an insect lineage initially begins to incorporate fungi into
its more generalist diet, then becomes a specialized fungivore, and finally evolves
adaptations for cultivating fungi. In the transmission-first model (the likely model
for the beetles), the insect lineage begins its association with a fungus by serving
570
19 Aug 2005 15:44
TABLE 2 Crop ecology and evolution in ant, termite, beetle, and human agriculture
AR
Agriculture in:
Crop ecology & evolution Attine ants Macrotermitine termites Xyleborine ambrosia beetles Humans 3
Number of inferred Single origin of agriculture Single origin of agriculture Single origin of agriculture in Multiple independent
MUELLER ET AL.
evolutionary origins of in the Attini in the Macrotermitini the Xyleborini origins of agriculture
agricultural behavior No other ant practices true No other termite practices Agriculture originated
agriculture, but some ants true agriculture, but the independently in six
allow fungi to grow in their distantly related termite additional ambrosia beetle
ANRV259-ES36-24.tex
nest walls for structural or Sphaerotermes lineages; these other beetle

antibiotic purposes. sphaerothorax promotes systems are less well
the growth of bacteria in understood (see text).
food stores.
Estimated date and region 45–65 Mya in Amazonian 24–34 Mya in African 30–40 Mya in xyleborine 10,000 years ago for the
of origin of agricultural rainforests rainforests beetles; region of origin earliest known origins of
behavior unknown human agriculture;
21–60 Mya in the other six multiple regions of origin
ambrosia beetle lineages
Number of crop clades Multiple cultivar clades Single cultivar clade Multiple cultivar clades Multiple crop clades from
cultivated One cultivar clade belongs All termite cultivars belong The ambrosia cultivars diverse lineages of plants
to the Pterulaceae; at least to the genus Termitomyces. Ambrosiella and Raffaellea and fungi
three additional have a polyphyletic origin
polyphyletic cultivar clades within the genera
belong to the Lepiotaceae Ophiostoma and
P1: OJO
(tribe Leucocoprineae). Ceratocystis.

Crop transmission
Vertical cultivar inheritance Present Variable Present Present
19 Aug 2005 15:44
fromparent to offspring Vertical cultivar inheritance Vertical cultivar inheritance Vertical cultivar inheritance is
(i.e., cultivar transfer is the rule in all attine ant is typical for two the rule in all xyleborine
between generations) lineages. independently derived lineages.
AR
macrotermite lineages; all

other macrotermitines
acquire cultivars
horizontally via
wind-dispersed spores
from other colonies.
Specialization on crops
Specialization at higher Clade-clade correspondence Clade-clade correspondence No strict clade-clade Not applicable
phylogenetic levels Defined clades of ant Defined clades of termite correspondence
ANRV259-ES36-24.tex
species are specialized in species are specialized in The fungal lineages

that they only grow fungi that they only grow fungi associated with xyleborine
typical for their own typical for their own ambrosia beetles are only
ant-specific cultivar clade. termite-specific cultivar distantly related to each
clade. other, and all are also
associated with some bark
beetles.
Specialization at level of Present Present Present Not applicable
farmer species In all cases studied in detail, Single termite species Single beetle species Specialization on crops is
single species of ants grow generally associate with generally associate with historically manifested by
only a single, multiple fungal lineages, multiple, distantly related different human lineages
phylogenetically narrow but within the limits of fungal lineages; the primary (reflecting separate
group of fungi (i.e., a specific cultivar clades (see cultivar is typically a single domestication events), but
single, phylogenetically above); however, some fungal species, but no single human lineage is
defined “species” of termite species are very secondary cultivars vary and nearly as specialized on a
P1: OJO
fungus). specialized on a particular often come from different single crop as are the insect
fungus. fungal lineages than the farmers.
primary-cultivar lineage.
(continued )
571
572
19 Aug 2005 15:44
Table 2 (continued )
Agriculture in:
AR
Crop ecology & evolution Attine ants Macrotermitine termites Xyleborine ambrosia beetles Humans
Crop sharing and crop

exchange
MUELLER ET AL.
Lateral exchange of Present Unknown Unknown Present

cultivars within the same Within a species of ant, Lateral cultivar exchange is Lateral exchange of cultivars
farmer species (i.e., cultivars are probably probably rare, but may occur between farmer societies is
exchange within same occasionally exchanged when galleries containing a prominent feature of
ANRV259-ES36-24.tex
farmer generation) between colonies (e.g., brood of two or more human agriculture.
after cultivar loss when foundress females, each
cultivar-deprived colonies carrying a different cultivar
may acquire replacement (e.g., different species of
cultivars from neighboring fungi), occur in close
colonies). proximity in the same tree.
Cultivar exchange between Present Present Rare Not applicable
farmer species Horizontal cultivar Horizontal cultivar exchange Horizontal cultivar exchange
exchange occurs between (through wind-dispersed occurs at least occasionally.
sympatric ant species if spores) is typical for most
they are specialized on the termite lineages.
same cultivar lineage.

Genetic exchange between Variable Absent Absent Absent
domesticated and wild Primitive attine ants Wild populations are Wild populations are Some traditional cultivation
P1: OJO
populations (e.g., genetic probably import novel unknown for termite unknown for ambrosia regimes (e.g., potato
interbreeding between cultivar genotypes cultivars. beetle cultivars. farming in the
19 Aug 2005 15:44
domesticated and wild regularly from wild Andes) are based on

AR
populations; or import of populations; cultivars of regular genetic exchange

new domesticates from the derived higher attine between domesticated and
wild populations) ants do not appear to have wild populations, but most
free-living populations. modern cultivars are highly
derived and do not readily
outcross with wild,
ancestral populations.
Crop monoculture Present Present Present Variable
All available data indicate All available data indicate Most beetle species appear to Monoculture predominates
ANRV259-ES36-24.tex
that only a single cultivar is that only a single cultivar is be associated with a single agriculture in many but not
grown within a nest. grown within a nest. primary cultivar and one or all regions.
more secondary cultivars; it
is unclear whether
secondary cultivars are truly
cultivars, weeds, or switch
between these roles.
Crop sexuality Variable, but sexual Variable, but most cultivars Variable, but sexual Variable, but most cultivars
recombination is rare are propagated sexually recombination is probably are propagated sexually
Predominant clonal cultivar Cultivar sexuality is typical rare The great majority of human
propagation within nests for those termite lineages Asexuality is typical for both cultivars are sexually
and between nests is that acquire cultivars from primary and secondary propagated, but some
punctuated occasionally by the environment every cultivars; some secondary cultivars are largely
genetic exchanges, generation, but cultivar cultivars are sexual. clonally (e.g., potato) or
outbreeding, or other sexuality appears absent in strictly clonally propagated
P1: OJO
events generating the two agricultural (e.g., banana).

recombinants. systems with vertical

cultivar transmission.
573
574 MUELLER ET AL.
as a vector of that fungus, then begins to derive nutrition from it, and finally
becomes a fungus cultivator. In a third possibility, an insect-fungus association
evolves because the insects originally use fungi as a source of antibiotics as, for
example, in the lower termite Reticulotermes speratus that derives antibiotic pro-
tection from fungal sclerotia mixed into egg piles (Matsuura et al. 2000). Lastly,
insect-associated fungi may have undergone even more complicated evolutionary
histories, originating from the exploitation by one insect lineage (e.g., the ances-
tor of attine ants) of a preexisting insect-fungus association (the fungi ancestrally
associating with beetles) when it encounters these insect-adapted fungi in a shared
nest environment (e.g., decaying wood; Sanchez-Peña 2005). This latter hypoth-
esis, however, is not supported by the phylogenetic relationships between beetle
and ant fungi and is inconsistent with the estimated dates of origin of these insect-
fungal associations (i.e., attine agriculture probably arose well before ambrosia
beetle agriculture; see Table 2).
For attine ants, it is unclear whether agriculture arose from a state of ancestral
fungivory, antibiotic acquisition, or fungal vectoring (Mueller et al. 2001). Termite
agriculture most likely originated via the consumption-first route, because many
nonfarming termite species are attracted to and feed on fungus-infested wood,
which suggests that the nonfarming ancestors of the farming termites may have
fed on fungi as well (Batra & Batra 1979, Rouland-Lefevre 2000). The nonfarming
ancestors of the fungus-growing beetles appear to have associated with fungi even
before the origin of fungiculture, because many of the more primitive nongarden-
ing scolytines act as fungal vectors without apparent dependence on their fungal
associates (Harrington 2005, Malloch & Blackwell 1993, Six 2003, Six & Klepzig
2004). This suggests non-nutritional dependencies on fungi that predate the origins
of fungiculture in the various ambrosia beetle lineages (Six 2003). However, many
nonambrosial scolytines carry fungi in mycangia and feed as larvae on ungardened
mycelium that colonizes host plants and feed as new adults on spore layers lining
pupal chambers (Ayres et al. 2000, Barras 1973, Six & Klepzig 2004, Six & Paine
1998; also A. Adams & D.L. Six, unpublished data), suggesting a stage of nutri-
tional dependency predating the origin of fungiculture. Thus, some of the seven
agricultural origins in beetles appear to have followed the transmission-first route,
whereas others followed the consumption-first route.
Insect agriculture is restricted to the cultivation of fungi rather than plants,
which predominate in human agriculture. Although it is true that some insects
are specialized on host plants that they protect from other herbivores (e.g., Pseu-
domyrmex ants protect acacia trees in exchange for shelter and nutritional benefits;
Janzen 1966, Hölldobler & Wilson 1990), none of these insect-plant mutualisms
possesses all four of the components of agriculture listed above. One could there-
fore ask what factors have predisposed insects to evolve fungal rather than plant
agriculture. Indeed, there are several advantages of fungal agriculture over plant
agriculture, and several characteristics of plants may even preclude their easy cul-
tivation. First, unlike fungi, plants typically have stringent light and space require-
ments, excluding them from cultivation in the subterranean or otherwise enclosed
nests of insects. Such nesting habits may facilitate fungiculture by shielding fun-
gal crops from unwanted consumers (i.e., other fungivores) and wind-dispersed
diseases. Furthermore, unlike plants, which usually require regular pollination for
long-term cultivation, fungi can be maintained indefinitely in a nonsexual mycelial
state, yielding a more consistent food source. Thus, although seeds and plant ma-
terial can be readily harvested, fungi are likely more cultivatable, explaining the
predominance of fungal rather than plant agriculture among insects.
Ant, termite, and most beetle agriculturists are social. All ants and termites
are eusocial (characterized by reproductive division of labor, cooperative brood
care, and overlap of generations; Hölldobler & Wilson 1990). Only one ambrosia
beetle (Austroplatypus incompertus) is known to be eusocial (Kent & Simpson
1992); the remainder are subsocial, in which a single female cares for her brood,
or communal, in which several reproductive females cooperate in brood care and
gardening (Kirkendall et al. 1997). Sociality may have facilitated the evolution
of agriculture because of the inherent advantage to agriculture of division of la-
bor, which enables the partitioning of agricultural tasks and augments agricultural
efficiency (Hölldobler & Wilson 1990, Hart et al. 2002). In ant and termite farm-
ers, for example, agricultural tasks are partitioned in a conveyor-belt-like series
between different worker castes, each specialized on one main task: foraging; pro-
cessing and cleaning of substrate before incorporation into the garden; planting of
mycelium onto new substrate; monitoring and weeding of the garden; or disposal
of diseased or senescent garden (Bot et al. 2001a, Hart et al. 2002, Traniello &
Leuthold 2000). Task partitioning has so far not been investigated in the ambrosia
beetles because of the logistical difficulties of studying beetle behavior in their
concealed tunnels. Task partitioning likely facilitates great efficiency in defense
against nest and garden robbers (Adams et al. 2000a, b; LaPolla et al. 2002), in
monitoring gardens for diseases, and in modulating optimal environmental condi-
tions for crop growth.
4. AGRICULTURAL EVOLUTION AND ECOLOGY

A series of convergent and divergent features of agricultural evolution emerge
from a comparative analysis of ant, termite, and beetle fungiculture (summarized
in Tables 1–3).
4.1. Cultivar Transmission Between Farmer Generations

In attine ant and xyleborine beetle agriculture, fungal cultivars are transmitted
vertically by trophophoresy from parent to offspring generations (Fernández-Marı́n
et al. 2004, Francke-Grosmann 1967, Haanstad & Norris 1985, Huber 1905).
Female reproductive ants and beetles acquire inocula from their natal gardens,
carry these inocula with them in specialized pockets during dispersal flights early in
life, and use these inocula as starter cultures for their new gardens. Trophophoretic
576
TABLE 3 Disease ecology and evolution in ant, termite, beetle, and human agriculture
19 Aug 2005 15:44
Agriculture in:
Disease ecology and Xyleborine ambrosia
AR
evolution Attine ants Macrotermitine termites beetles Humans
a. Disease and pest types Crops are attacked by fungal and Specialized crop parasites have Most ambrosial gardens Crops are infected by
bacterial pests, including not been documented, but consist of a complex of fungal, bacterial, and viral
specialized (Escovopsis) and some species of Xylaria are mycelial fungi, yeasts, and diseases and are attacked
MUELLER ET AL.
unspecialized (Trichoderma) specialized as weeds of bacteria. At least some of by invertebrate and

fungal pathogens. Microbial fungus gardens. Common soil these are likely to cause vertebrate herbivores.
“weeds” (e.g., Xylaria fungi) and fungi are present in gardens as disease or act as pests. Weeds are an additional
fungivorous arthropods (e.g., well. problem in human
mites) also inhabit gardens. agriculture.
ANRV259-ES36-24.tex
b. Disease prevalence Infection of gardens by the crop Unknown Unknown Ubiquitous and diverse
parasite Escovopsis is frequent parasites and pathogens
and sometimes devastating. cause immense loss in crop
harvests.
c. Defense mechanisms by
farmers against crop
disease
- Protection from disease
Sequestration, sheltering Present Present Present Absent1
Substrate sterilization or Present Present Unknown Uncommon
cleaning
Guarding or protecting Present Present Present Uncommon

against disease vectors
Partitioning of Present An uncontaminated PresentForaging and gardening Unknown Uncommon
disease-removal tasks to worker-caste specializes on are performed by different
P1: OJO
minimize contact gardening; a contaminated worker castes in many

between contaminated worker-caste specializes on species.
workers and healthy crop disease removal.
- Eradication of disease
Continuous, intensive Present Present Present Very rare
19 Aug 2005 15:44
monitoring of all crop

Physical weeding Present Ant gardeners Present Termite gardeners Unknown Female beetles Present Physical weeding is
continuously “weed” and groom gardens to eliminate continually tend crop more integral in primitive
AR
“groom” their gardens, excising some alien microbes (e.g., gardens but whether agricultural systems, but
infected garden fragments. Xylaria stromata). directed weeding is part of less so in modern
this activity is unknown. monoculture systems.
Application of chemical Present Metapleural and Present Labial and salivary Unknown Present Chemical control of
herbicides mandibular gland secretions have gland secretions have diseases and herbivores is
general antibiotic properties; antibiotic properties; an essential element of
antibiotics derived from defensive secretions of most modern monoculture
actinomycete bacteria have soldiers have antimicrobial systems.
specific antibiotic effects against characteristics, but their
ANRV259-ES36-24.tex
Escovopsis parasites. sanitary function in the garden

is unknown.
Routine use of Present Unknown Unknown Absent2
disease-suppressant
microbes
d. Defense mechanisms of
crop against crop
diseases
Induced resistance of crop Unknown Unknown Unknown Present
Constitutive resistance of Present Cultivars produce Unknown Unknown Present
crop antibacterial and antifungal
antibiotics, which appear to
mediate resistance of some

cultivar genotypes to some
Escovopsis genotypes.
P1: OJO
1
Greenhouses provide a sequestered environment for some human crops, but greenhouse farming is costly and contributes only a small fraction (about 0.02%, Paulitz & Bélanger 2001) to
the total productivity of human agriculture; the bulk of human agriculture operates in open landscapes that expose crops to environmental stresses (fluctuations in moisture and temperature;
wind-borne pathogens; migratory herbivores; etc.)
2
Some recently developed uses of disease-suppressant microbes in agricultural pest control are discussed in the text; application of disease-suppressant microbes has yet to become an integral
part of mainstream human agriculture (but see Morrissey et al. 2004)
577
578 MUELLER ET AL.
vertical transmission also occurs in two macrotermitine groups, except that in one
of these two groups the fungal cultivar is transmitted via the king (the single species
Macrotermes bellicosus), whereas in the other group (the genus Microtermes) the
fungus is transmitted via the queen. In the few cases where fungal transmission has
been studied in the remaining macrotermitines (Johnson 1981, Johnson et al. 1981,
Korb & Aanen 2003, Sieber 1983), these termites rely on horizontal acquisition
of fungal crops from the environment in each generation.
4.2. Higher-Level Specialization (Clade-Clade Congruence)

Between Farmers and Crops
Vertical transmission of cultivars leads to the expectation of clade-clade correspon-
dences and topological congruence between the phylogenies of insect farmers and
those of their cultivars. Indeed, in all insect farming systems, major groups of
farmers (large clades or paraphyletic grades, e.g., the lower attine ants) strictly
specialize on major groups of corresponding fungal cultivars (Figure 2). The ex-
pected farmer-cultivar congruence therefore does occur at higher (i.e., broad) phy-
logenetic levels, possibly because of ancient evolutionary codependencies (e.g.,
physiological/nutritional requirements of the farmers, cultivation requirements of
the fungi, etc.) that strictly preclude switches by farmers to cultivars outside of
their specialized major cultivar groups.
Phylogenetic patterns (Figure 2) indicate, however, that within these strictly
constrained major cultivar groups, insect-farmer species occasionally switch be-
tween fungal species or strains. This combination of lower-level, within-group
switching and higher-level major-group specialization in insect farmers would be
analogous, in humans, to defined clades of specialized wheat-farmers, rice-farmers,
potato-farmers, bean-farmers, etc., each of which is able to switch between varieties
within their area of specialization (e.g., between varieties of wheat and to closely
related species such as barley), but which cannot switch across major groups (e.g.,
from wheat to beans). Among insect farmers, switches to novel major cultivar
groups have been exceedingly rare evolutionary events.
4.3. Lower-Level Specialization on Cultivars

Though low-level switching between cultivar species and strains within major
cultivar groups occurs occasionally over evolutionary time, over ecological spans
of time most insect farmer species associate with only a very narrow subgroup
of cultivars (species or strains). For example, every attine ant species surveyed
to date cultivates only a phylogenetically narrow set of cultivars (e.g., a single
species of fungus), implicating species specificity between ants and cultivars at
very recent levels of evolutionary diversification (Bot et al. 2001b, Green et al.
2002, Schultz et al. 2002). In ambrosia beetles, like ants, only one primary cul-
tivar is associated with a particular beetle species within a particular geographic
region (Gebhardt et al. 2004, Batra 1967). However, although most beetles are
associated with a species-specific, primary fungus across their entire geographic
ranges, some beetle species associate with different primary cultivars in different
geographic regions (Baker 1963, von Arx & Hennebert 1965). Among macroter-
mitine species, there exists considerable variation in cultivar specialization: Some
species are limited to a single, unique cultivar, whereas other species cultivate a
great diversity of fungal cultivars, which they sometimes share with other, usu-
ally closely related, macrotermitine species (Katoh et al. 2002; D.K. Aanen &
P. Eggleton, submitted). The factors underlying variation in termite specializa-
tion are unknown, but different cultivars may serve different primary functions,
providing specific, termite-adapted enzymes in some cases (leading to termite-
cultivar specialization), while providing generalized food in other cases (permit-
ting exchange between termite species; D.K. Aanen, V.I. Ros, H.H. de Fine Licht,
C. Roulant-LeFévre, J. Mitchel, et al., in review).
4.4. Cultivar Sharing and Exchange Between Farmer Species

Even though each attine ant species is specialized on a single cultivar species, a
given cultivar species may be cultivated by several sympatric species of ants, and
these sympatric ant species may not necessarily be closely related to one another
(e.g., they may represent different ant genera) (Bot et al. 2001b, Green et al. 2002,
Villesen et al. 2004). Cultivar transfer between ant species may occur via direct
or indirect avenues. Direct avenues may include raiding of neighboring colonies
(Adams et al. 2000a, Rissing et al. 1989) or, in polygynous species, cofounding
of colonies by multiple queens that exchange cultivars or recombine them in the
cofounded garden. Indirect avenues may include cultivar escapes from gardens,
followed by a free-living (feral) existence and subsequent reincorporation into a
symbiosis when a different attine colony imports the free-living strain into its nest
(Mueller et al. 1998).
For ambrosia beetles, the available phylogenetic evidence points to cultivar
sharing between different sympatric beetle species (Farrell et al. 2001, Gebhardt
et al. 2004) but few investigations have addressed this question. Distantly related
ambrosia beetle species are sometimes associated with the same cultivar (Gebhardt
et al. 2004), implicating fungal exchange, either direct or indirect, as explained
above for the ants. Cultivar exchange between and within beetle species may occur
when different female beetles colonize the same tree and the fungal associates
cross-contaminate adjacent galleries.
In contrast to attine ants and ambrosia beetles that all transmit their cultivars
vertically between generations, most macrotermitine species acquire their fungi
horizontally each generation. This implies that new termite-cultivar combinations
arise each generation, which should facilitate cultivar exchange between species,
as well as between lineages of the same species. Cultivar surveys of sympatric
macrotermitine communities indeed indicate that cultivars are generally shared
between closely related species via interspecific cultivar exchanges (Aanen et al.
2002; but see the exceptions mentioned above in Section 4.3). Intra-specific cultivar
exchanges have so far not been investigated in macrotermitines.
580 MUELLER ET AL.
4.5. Propagation of Sexual Versus Asexual Cultivars and

Links to Free-Living Cultivar Populations
All vertically transmitted insect cultivars, including the cultivars of attine ants, am-
brosia beetles, and termites in the genus Microtermes and the species Macrotermes
bellicosus, seem to be asexually propagated by their insect farmers across multi-
ple farmer generations. In contrast, the horizontally transmitted termite cultivars
(propagated by all other macrotermitine genera) undergo regular meiosis and sex-
ual recombination (see above).
Evidence for cultivar asexuality in attine ants comes from DNA fingerprinting
studies that indicate that all gardens of a single leafcutter colony contain a single
cultivar clone (monoculture) (Kweskin 2003, Poulsen & Boomsma 2005; J. Scott
& U.G. Mueller, unpublished data); that identical cultivar clones occur in differ-
ent colonies of the same geographically widespread attine ant species (Bot et al.
2001b, Green et al. 2002, Mueller et al. 1996); and that different sympatric ant
species occasionally share genetically identical cultivar clones (see above; Bot et al.
2001b, Green et al. 2002, Mueller et al. 1998). Contrary to previous suggestions
(Chapela et al. 1994), however, attine cultivar clones are not ancient. Although at-
tine cultivars are clonally propagated across many ant generations (Mueller 2002),
this clonality is punctuated by occasional recombination events, involving either
sexual (meiosis, mating) or parasexual (e.g., mitotic recombination, exchange
of haploid nuclei) processes. Evidence for occasional recombination includes:
(a) fruiting structures (mushrooms), which are known for cultivars of nearly all
genera of attine ants, contradicting the expectation of loss of fruiting ability un-
der strict clonality spanning millions of years (Mueller 2002); (b) rates of allele
sequence divergence in attine cultivars that are similar to those of closely related,
sexually reproducing fungi (Mueller et al. 1998; S.A. Rehner, unpublished data);
and (c) cultivars of the lower attine ants that have close genetic links to free-living
fungal populations (Green et al. 2002, Mueller et al. 1998), suggesting that these
fungi are capable of moving in and out of the symbiosis, that cultivar lineages
may regularly interbreed with wild lineages, or both. Taken together, the genetic
and natural-history information suggest predominantly asexual cultivar propaga-
tion within ant nests and across many ant generations, punctuated by occasional
genetic recombination events.
As in attine ants, a single cultivar monoculture is grown in a single termite
colony (Aanen et al. 2002, Katoh et al. 2002). Within termite nests, the Termito-
myces cultivar is propagated asexually by inoculating fresh garden substrate with
asexual spores (Leuthold et al. 1989), and probably also by transplanting mycelium
from older to younger gardens. Although Termitomyces species have no known
free-living populations existing entirely independent of the termite farmers, they
have nonetheless retained the ancestral (presymbiotic) condition of regular sexual
reproduction, and most Termitomyces cultivars are spread from one termite nest to
another horizontally via sexual spores produced by fruiting bodies (mushrooms)
growing on the external surfaces of mature nests. The Termitomyces cultivar of the
termite Macrotermes natalensis, for example, has an outcrossing mating system

(De Fine Licht et al. 2005). Asexual cultivar propagation spanning several termite
generations only occurs in those species with vertical uniparental propagation.
Phylogenetic patterns implicate occasional horizontal cultivar exchange between
nests of the same and different termite species (Aanen et al. 2002), but whether
such horizontal exchange is associated with cultivar sexual reproduction remains
unknown.
In xyleborine beetles, the primary fungi are strictly asexual (Jones & Blackwell
1998, Rollins et al. 2001), whereas the less specific, auxiliary fungi are often
sexual (Francke-Grosmann 1967). A preponderance of asexual reproduction in
fungal cultivars also occurs in the primary fungi of all other ambrosia beetles,
whereas, again, the more incidental fungi are often sexual (Six 2003, Six & Paine
1999), suggesting that this may have been the ancestral condition at the origin of
the xyleborine beetle-fungus symbiosis.
4.6. Coevolutionary Modifications

Farmer-cultivar specialization enhances the potential for coadaptation, in which
evolutionary modification in one of the partners causes a reciprocal coevolution-
ary modification in the other partner (Futuyma & Slatkin 1983). It is relatively
easy to identify evolutionary modifications in the farmer species, such as special-
ized morphological structures for the trophophoretic transport of cultivars by fe-
males during the dispersal flight (e.g., mycangia in the beetles, infrabuccal pocket
in the ants), modifications of mandibles and guts of beetle and ant larvae for
fungus-feeding (Browne 1961, Schultz & Meier 1995), or the suite of behavioral,
glandular, or physiological modifications that form the basis of insect farming.
Examples of evolutionary modifications in the cultivars have been more difficult
to identify, however, because the cultivated fungi are inherently more difficult to
study.
The clearest examples of cultivar modifications are the hyphal-tip swellings
(gongylidia) produced by the cultivars of the higher attines and the analogous
nodules produced by macrotermitine cultivars. Both gongylidia and nodules are
nutrient-rich structures designed for easy harvesting by the farmers, ingesting, and
feeding to the larvae or nymphs. Nutrient-rich structures are not known for beetle
gardens, although the ambrosia morphology of the beetle cultivars suggests evolu-
tionary modification designed specifically for efficient consumption and digestion
by the beetle larvae. Ambrosial growth consists of tightly packed conidiophores
with copious spores and is only formed in the presence of the beetles (French
& Roeper 1972). Ambrosia formation has not been reported from nonsymbiotic
fungus species. Interestingly, two of the major genera of fungi associated with
ambrosia beetles (Ambrosiella, Raffaelea) are each polyphyletic, and the multiple
lineages within each genus have converged on the same ambrosial morphology
(Blackwell & Jones 1997, Jones & Blackwell 1998), suggesting evolutionary con-
vergence due to selection. Other likely coevolutionary modifications that have
582 MUELLER ET AL.
yet to be investigated include predominantly asexual reproduction in the insect

cultivars while under cultivation and the cultivars’ capacity to survive storage in
the dispersal pockets of the beetles and ants, or the passage through the alimentary
canal of the termites.
4.7. Symbiont Choice and “Artificial Selection” of Cultivars

From an evolutionary perspective, insect agriculture represents a case of cooper-
ative interaction between farmer and cultivar lineages, each exploiting the other
for its own reproductive purposes (Herre et al. 1999, Mueller 2002). Such coop-
erative interactions are frequently unstable and can erode over evolutionary time,
for example, when mutant overexploiters arise (so-called cheater cultivars) and
invade a mutualism. A series of additional farmer-cultivar conflicts are predicted
that could destablilize the mutualism (Mueller 2002, Aanen & Boomsma 2005,
Schultz et al. 2005), but at least two evolutionary mechanisms preserve the coop-
erative nature of the farmer-cultivar association: First, partner feedback, inherent
in vertical cultivar transmission, is an automatic feedback mechanism in which an
uncooperative partner reduces the other partner’s fitness to the extent that it reduces
its own fitness as well; and, second, partner (symbiont) choice in which farmers
favor associations with productive cultivars and discriminate against inferior cul-
tivars in specific choice situations (e.g., a choice between cultivar strains that may
coexist in a garden or in proximate gardens, exercised either by workers during the
planting of new gardens or by reproductives when choosing a cultivar strain for
dispersal). In cases where the evolutionary rates differ between two cooperating
partners, partner choice is a particularly important mechanism (Sachs et al. 2004).
The slower-evolving partner (e.g., the insect farmer) is expected to exert the choice
between variants of the faster evolving partner (e.g., the fungal cultivar), and thus
the slower-evolving farmer imposes selection favoring beneficial symbiont vari-
ants (e.g., productive cultivars) and prevents the spread of nonbeneficial cultivar
mutants (e.g., degenerate or suboptimal cultivars; Mueller 2002, Sachs et al. 2004).
Symbiont choice has yet to be investigated for termite and beetle farmers, but ant
farmers are able to discern surprisingly fine genotypic differences between cultivars
(Mueller et al. 2004; also N.K. Advani & U.G. Mueller, submitted), suggesting that
cultivar diversity in ant gardens, arising, for example, through mutation in a garden
or through the import of novel strains, may evolve under an analog of “artificial
selection.”
5. ROLE OF DISEASE IN INSECT AGRICULTURE

“Weedy” fungi frequently invade the gardens of ants, termites, and beetles, and may
coexist at low or manageable levels along with the crop. If the gardening insects are
removed or if they abandon their nests, the garden is quickly overrun by these weeds
(Batra & Batra 1979, Norris 1979). One such group of weeds, wood-degrading
fungi in the endophytic genus Xylaria, is found in most gardens of fungus-growing

ants and termites, probably because it is introduced with garden substrate (Fisher
et al. 1995; V.I. Ros, A.J. Debets, T. Læssøe, D.K. Aanen, submitted; N.M. Gerardo
& U.G. Mueller, personal observation). Though weeds like Xylaria do not directly
attack the cultivar, they compete with it for nutrients and thus decrease crop yield
(V.I. Ros, A.J. Debets, T. Læssøe, D.K. Aanen, submitted). Escovopsis species,
ascomycete fungi found in colonies of fungus-growing ants, are specialized para-
sites that subsist directly on the cultivars and reduce the nutrients available to the
ants (Currie 2001a, Currie et al. 2003b). Weed fungi and bacteria are also known
in termite and beetle agriculture, but have yet to be studied in detail (Six 2003; V.I.
Ros, A.J. Debets, T. Læssøe, D.K. Aanen, submitted; D.K. Aanen, unpublished
data).
Escovopsis infections reduce garden productivity, which in turn reduces ant
colony growth and the likelihood of colony survival (Currie 2001a, 2001b, Currie
et al. 1999a). Escovopsis is geographically widespread and taxonomically diverse.
The parasite has been isolated from colonies of every attine genus throughout
their geographic ranges, and particular Escovopsis lineages specialize on para-
sitizing particular cultivar lineages. This high degree of host specificity suggests
a long history of host-parasite coevolution in Escovopsis, in which the cultivars,
the ants, and their mutualistic bacteria have likely coadapted to defend against
Escovopsis attack and in which each Escovopsis species has become narrowly
specialized to overcome the defenses of some hosts but not others (Gerardo et al.
2004).
6. DISEASE AND MICROBIAL MANAGEMENT

STRATEGIES
In response to the persistent selection pressure imposed by weeds and pathogens,
fungus-farming insects have evolved an arsenal of strategies for preventing and
suppressing infection (Table 3).
6.1. Sequestration of Gardens

All insect agriculturists sequester and separate their gardens from the surrounding
environment, e.g., by growing their gardens in underground chambers or gal-
leries in wood, or by covering them with a protective mycelial veil in the case of
some Apterostigma ants (Villesen et al. 2004). Although sequestered nests are the
ancestral condition in fungus-growing ants, termites, and beetles, and although
sequestered nests may serve other agricultural functions such as the regulation of
temperature and humidity, sequestration no doubt also buffers the garden against
fungivores, wind-borne pathogens, and arthropod vectors of diseases (e.g., mites,
collembolans).
584 MUELLER ET AL.
6.2. Maintenance of Cultivar Genetic Variability

Although attine ants clonally propagate their cultivars across generations over short
evolutionary time spans, no attine cultivar has been found to be an ancient clone.
Instead, the evidence indicates that the lower attines occasionally acquire new culti-
vars from wild (free-living), sexually reproducing fungal populations and that both
lower and higher attines occasionally acquire new cultivars from the nests of other
attines. The cultivars of higher attines, which are not known to have free-living
populations (Mueller 2002), nonetheless retain the ability to fruit (Mueller 2002)
and demonstrate patterns of DNA-sequence diversity that suggest occasional ge-
netic recombination through self-mating (S.A. Rehner, personal communication)
or through true intercrossing between different cultivar strains (M. Bacci, personal
communication). Thus, whereas the crop employed by any attine ant colony at
any given time is a clonally propagated monoculture, the genetic variability and
resilience necessary for long-term disease management resides in the fungal pop-
ulaton external to the nest. As already noted, sexual reproduction is the norm in
the cultivars of those termites that reacquire their cultivars horizontally each gen-
eration, and sexual reproduction may also occur in the fungi of those termites that
transmit their cultivars vertically across generations. Whether the primary culti-
vars of the ambrosia beetles occasionally reproduce sexually remains unknown.
At least for the termites and ants, then, and possibly for the beetles as well, access
to a population-level reservoir of cultivar genetic variability is a consistent feature
of insect agriculture that may provide alternative crops for dealing with disease.
6.3. Intensive Monitoring of Gardens, Weeding,

and Herbicide Application
All insect agriculturists constantly inspect their gardens, and no part of the garden
is left untended for periods of time sufficient to allow the establishment and spread
of diseases and fungivores (Batra & Batra 1979, Currie & Stuart 2001). Insect
agriculturists, particularly the ants and termites, are able to invest in such intensive
monitoring because their societies possess a nonreproducing worker caste, a large
portion of which is dedicated to garden care. In the beetles, the relatively small size
of the garden allows for intensive monitoring by a single female or by a small family
of females. Intensive monitoring ensures that diseases are discovered and eradi-
cated in the early stages of infection before they are able to spread and cause signifi-
cant crop loss. Early detection is an effective defense against novel disease mutants
that might evolve greater virulence if left untreated, because in the early stages of
infection these strains can more readily be controlled with standard treatments.
Garden treatment in attine ants includes the use of secretions from their meta-
pleural and mandibular glands to clean substrate as it is brought into the nest,
probably removing some or most weeds and pathogens from the surface of the
substrate before it is added to the fungus garden (Maschwitz et al. 1970, Ortius-
Lechner et al. 2000). Although antibiotic-producing glands have not been studied
in fungus-growing termites, some secretions of nonfungus-growing termites have
antibiotic properties (Rosengaus et al. 1998, 2004). Antimicrobial glands in am-

brosia beetles remain unknown and unstudied. In addition to glandular secretions,
attine ants have another antimicrobial defense. Some or all of their integuments are
covered with actinomycete bacteria. These bacteria are known to inhibit Escovop-
sis growth (Currie et al. 1999b), and experimental reduction of actinomycetes in
colonies increases Escovopsis infection (Currie et al. 2003a). Garden bacteria in
the genus Burkholderia (Santos et al. 2004) also provide antibiotics that provide
protection against the garden parasite Escovopsis and against entomopathogenic
diseases of the ants themselves. Termite gardens contain actinomycetes and other
bacteria (Batra & Batra 1979), and beetle gardens contain a great diversity of bac-
terial secondary symbionts; however, the exact roles of these bacterial associates
remain unknown.
6.4. Microbial Buffering

Beyond the known antibiotic-producing, disease-suppressing bacteria in attine
colonies (Currie 2001a, Currie et al. 1999b, Santos et al. 2004), other secondary
bacteria and fungi occur in insect gardens (Carreiro et al. 1997, Craven et al. 1970;
C. Wang & U.G. Mueller, unpublished data), but their roles remain largely un-
known. Although some of these secondary microbes may be neutral or detrimental
to garden health and productivity, others may provide disease-modulating effects
through competitive exclusion, antibiotic suppression of disease-causing microbes,
resistance induction, or other mechanisms of microbial interaction. Competitive
exclusion, disease suppression, and resistance induction have been demonstrated in
both experimental and natural microbial systems (Hood 2003, Paulitz & Bélanger
2001, Wille et al. 2001). Some secondary microbes may even facultatively switch
between beneficial and detrimental roles, depending on garden growth conditions,
seasonal factors, or interactions with the insects or other microbes. For example,
although the detrimental effects of Escovopsis are obvious in natural garden out-
breaks and in interactions with cultivars in vitro (Currie 2001a, b; Currie et al.
1999a; Gerardo et al. 2004), it remains an untested possibility that Escovopsis
may provide beneficial effects when present at low levels in the garden matrix.
Secondary microbes in termite gardens remain uninvestigated, but the possible
significance of a secondary microbial flora in beetle gardens has been recognized
for some time (Norris 1965). Norris (1965) suggested that it is the microbial com-
plex as a whole (filamentous fungi, yeasts, and bacteria), rather than the dominant
ambrosia fungus per se, that allows the beetles to exploit nutrient-poor substrates
such as wood. Norris did not speculate on any additional roles for the secondary
microbes, such as suppressing diseases, but such auxiliary roles deserve further
study.
6.5. Management of Crop-Associated Microbial Consortia

If, as recent evidence indicates, secondary microbes serve ancillary functions
in gardens by buffering against disease organisms or by producing antibiotics,
586 MUELLER ET AL.
enzymes, and metabolites, it is possible that insect farmers have evolved the ca-
pacity to manage these microbial consortia. Such microbial management strategies
by insect farmers could include the following three methods.
6.5.1. STERILIZATION OF SUBSTRATE BEFORE INCORPORATION INTO THE GARDEN

(ANTS, TERMITES) OR USE OF ESSENTIALLY STERILE SUBSTRATE FROM THE OUTSET
(BEETLES) In termites, the passage of substrate through the gut before incorpo-
ration into the garden probably eliminates many unwanted microbes and may
increase the abundance of desired microbes (Figure 1). In attine ants, the con-
siderable effort spent cleaning substrate surfaces appears to partially sterilize the
substrate (Weber 1972). In the case of the ambrosia beetles, gardening occurs in
a closed system because the beetles do not need to leave the nest to forage and
because galleries are excavated in what is essentially a sterile medium, the sap-
wood or heartwood of living or recently killed trees, which are generally free of
endophytic fungi and other microbes. This closed system greatly reduces the po-
tential for accidental introduction of unwanted microbes and likely facilitates the
management of desired microbes in beetle gardens.
6.5.2. SPATIALLY STRUCTURED GARDEN MATRIX Structuring of gardens allows in-

sect farmers to assess properties of particular, localized crop-microbe consortia.
Any unwanted mutant genotypes, arising locally under particular microbe-microbe
competitons, thus can be identified indirectly through the detection of their detri-
mental effects on the properties of the subgarden, and that subgarden piece can
then be excised. Conversely, novel microbial mutants with beneficial effects can
be identified indirectly by their beneficial effects on the subgarden, and preferen-
tially subcultured and propagated across the rest of the garden (Figure 1). Such
“symbiont-community choice” is possible only because of the fixed garden matrix,
enabling farmers to assess properties of local consortia.
6.5.3. CONTINUOUS CO-OCCURRENCE OF GARDENS OF ALL AGES IN CLOSE PROXI-

MITY, RANGING FROM “UNPLANTED” TO MATURE GARDENS Coexistence of gar-
dens at different developmental stages is inherent in the vertical structuring of attine
and termite gardens into younger subgardens at the top and older subgardens at the
bottom. A range of differently-aged gardens allows farmers to efficiently practice
one-way, selective transfer of only beneficial crop-microbe consortia from mature
to younger gardens. Age-structuring also delays the spread of mutant microbes
from older garden material to younger, more sterile gardens and thus prevents
deterioration of the symbiont-community.
6.6. Multipartner Coevolution and Coevolving

Antibiotic Defenses
One hypothetical advantage of secondary mutualistic microbes is that, unlike the
insect farmers, microbes can potentially evolve at the same rate as the coevolving
garden pests, enabling mutualistic insect-microbe systems to respond rapidly to the

emergence of novel disease genotypes (Currie 2001a, Mueller & Gerardo 2002).
Although such rapid microbial antibiotic defenses would obviously confer clear
advantages, their evolutionary maintenance remains unclear. One possibility is
that any single farmer society may have access to a diverse array of microbes from
which it can select particular, desired types as needed. This scenario raises the
question of how the insect farmers could maintain such a diverse array in their
colonies in the face of both competition between microbes and the regular bottle-
necking of the entire microbial “library” that presumably occurs at the founding
of every new insect colony. Alternatively, the associated secondary microbes may
be inherently fast-mutating, so that novel beneficial genotypes can rapidly arise
to muster an appropriate defensive response. This scenario raises the question of
how the most beneficial genotypes are recognized by the insect farmers and cho-
sen for selective “amplification” against particular pathogens. The lack of clear
evolutionary mechanisms for maintaining functional associations with coevolv-
ing, mutualistic microbes is not trivial, and future research needs to assess not
only the diversity of microbial genotypes within single farmer colonies, but also
to identify the mechanisms underlying adaptive symbiont-choice selection of ben-
eficial, novel microbial genotypes. Future research also needs to address whether
the coevolution of several, mutualistically-aligned partners (i.e., a “multidefense
alliance” of ants, cultivar, and auxiliary microbes), each mustering its own defense,
provides for a more evolutionarily stable disease-management strategy compared
to a strategy in which the insect farmers act alone in a coevolutionary arms race
against particular pathogens.
7. PRINCIPLES OF INSECT AGRICULTURE:

LESSONS FOR HUMAN AGRICULTURE?
Perhaps the most striking feature of insect agriculture is the long-term cultivation
of clonal monocultures. Monoculture increases agricultural efficiency through an
economy of scale (Wolfe 1985), and clonality preserves the desirable properties of
the crop by eliminating sexual recombination, but these advantages come at two
costs: (a) increased vulnerability to the rapid spread of disease mutants (Barrett
1981, Shipton 1977, Mitchell et al. 2002, Mundt 2002, Peacock et al. 2001, Piper
et al. 1996, Wolfe 1985), and (b) decreased resistance to fast-evolving diseases due
to decreased genetic variability in the crop (Barrett 1981, Gustafson et al. 2003,
Hamilton et al. 1990, Jaenike 1978, Zhu et al. 2000). These economic tradeoffs
(i.e., monoculture/clonality efficiency versus disease vulnerability) apply to both
human and insect farmers.
The insect farmers’ solution to the monoculture-disease problem appears to be
not a single, “magic bullet” strategy (e.g., sole reliance on pesticides), but rather a
combination of several strategies consisting of (a) crop sequestration, (b) intensive
monitoring of crops for diseases, (c) access to a population-level reservoir of crop
588 MUELLER ET AL.
genetic variability and (d) management of disease-suppressant microbes associated

with the crop (Table 3). Of these strategies, large-scale crop sequestration is the
least feasible in human agriculture because human crops need exposure to sunlight
and because greenhouse cultivation is costly (Paulitz & Bélanger 2001). Intensive
(e.g., daily) monitoring of every single crop plant for diseases may be feasible
for some crops (e.g., in greenhouse environments); however, hourly monitoring
of the kind implemented in insect agriculture seems cost-prohibitive for human
agriculture at large.
A more novel approach is to design human agricultural systems that more
efficiently take advantage of the microbial consortia that are known to play ben-
eficial roles in crop nutrient uptake and disease resistance (Morrissey et al. 2004,
Paulitz & Bélanger 2001, Wardle et al. 2004). Microbes of the rhizosphere (e.g.,
nitrogen-fixing bacteria and mycorrhizal fungi) have long been managed as crit-
ical associates of certain crops and trees (Finlay 2004, Johansson et al. 2004).
More recently, disease-suppressant bacteria have been discovered that live on
the root exudates of crops and produce antibiotics that protect the crop against
pathogens (Haas & Keel 2003, Mazzola 2004, Morrissey et al. 2004, Weller et al.
2002, Whipps 2001). Disease-suppressant effects on crop plants have also been
documented for phyllosphere microbes (Lindow & Brandl 2003) and endophytic
microbes (Narisawa et al. 2002, Sturz et al. 2000). Agricultural research on rhi-
zosphere, phyllosphere, and endophyte microbes of human crops is a very new
field, however, and many beneficial microbes remain to be discovered and put to
use (e.g., inoculation of crops with phyllosphere microbes to deter herbivores or
to suppress airborne diseases).
Two problems commonly encountered in human agricultural experiments with
beneficial microbial consortia are, first, that the composition of microbial species
is difficult to manage and stabilize (Garbeva et al. 2004, Mazzola 2004), and,
second, that beneficial microbes can rapidly evolve into detrimental ones (Alves
et al. 2003, Morrissey et al. 2002). The farming insects’ solution to these prob-
lems appears to consist of (a) selection on spatially limited microbial consortia
(i.e., high-resolution, spatial separation of evolutionary processes, preventing the
uncontrolled spread of microbes from inferior consortia); (b) propagation of crops
with fast generational turnovers, thus minimizing the time for the evolution of any
deleterious traits in the microbes; and (c) partial or complete sterilization of the
substrate prior to planting, thus minimizing the influx of microbial contaminants
into a largely closed agricultural system.
Perhaps it is from strategies such as these that humans have the most to learn
from insect farmers, certainly if disease-suppressant microbes are ever to be man-
aged in human agriculture (Morrissey et al. 2004). In developing these strategies,
agriculturists would need to keep in mind that, during the domestication process,
current human crops were not necessarily selected for capacities to interact with
auxiliary microbes, i.e., the alleles in the wild ancestors optimally mediating such
interactions may have been lost during the domestication process. Thus, a full
evaluation of the potential uses of auxiliary microbes in human agriculture may
require the study of the microbial consortia associated with the wild populations
from which human-domesticated crops were originally derived. Such domestica-
tion within the context of coevolving and codomesticated microbial consortia may
well be the key element explaining the 50-million year old agricultural success of
the insect farmers.
ACKNOWLEDGMENTS
This review synthesizes work supported by the National Science Foundation
(awards DEB-97,07209 and IRCEB DEB-01,10073 to U.G.M. & T.R.S.; CA-
REER DEB-99,83879 to U.G.M.; DEB-03,08757 to N.M.G.; INT-04,34171 to
DLS), the Smithsonian Tropical Research Institute (to U.G.M., N.M.G. & D.K.A.),
the Smithsonian Institution (to T.R.S.), and the Danish Natural Science Research
Council (to D.K.A.). Some of the ideas presented here evolved out of discussions
with P. Abbot, L. Ancel-Meyers, M. Bacci, K. Boomsma, J. Bull, Y. Carriere,
C. Currie, A. Herre, A. Himler, N. Mehdiabadi, A. Mikheyev, T. Murakami, F.
Pagnocca, S. Rehner, J. Sachs, R. Samuels, J. Scott, S. Solomon, L. Thomashow,
and B. Wcislo. Special thanks to P. Abbot, R. Adams, K. Boomsma, S. Brady,
S. Bruschi, E. Caldera, H. De Fine Licht, M. Dijkstra, A. Green, A. Himler, D.
Kronauer, J. LaPolla, C. Marshall, A. Mikheyev, J. Miller, A. Mosegaard, D. Nash,
J. Pedersen, M. Poulsen, C. Rabeling, J. Scott, B. Slippers, S. Solomon, J. Sosa-
Calvo, and one anonymous reviewer for constructive comments on the manuscript;
and to M. Bacci, A. Himler, C. Currie, and S. Rehner for unpublished information.
We are most grateful to Greg Dimijian, Susanne Kühnholz, and Karen Machielsen
for photos in Figure 2; and especially to Barrett Klein for producing Figure 1.
The Annual Review of Ecology, Evolution, and Systematics is online at

http://ecolsys.annualreviews.org
LITERATURE CITED
Aanen DK, Boomsma JJ. 2005. Evolutionary niak JM. 2000a. Garden sharing and garden
dynamics of the mutualistic symbiosis be- stealing in fungus-growing ants. Naturwis-
tween fungus-growing termites and Termit- senschaften 87:491–93
omyces fungi. See Vega & Blackwell 2005, Adams RMM, Mueller UG, Schultz TR, Nor-
pp. 191–210 den B. 2000b. Agro-predation: usurpation
Aanen DK, Eggleton P, Rouland-Lefèvre C, of attine fungus gardens by Megalomyrmex
Guldberg-Frøslev T, Rosendahl S, Boomsma ants. Naturwissenschaften 87:549–54
JJ. 2002. The evolution of fungus-growing Alves BJR, Boddey RM, Urquiaga S. 2003. The
termites and their mutualistic fungal sym- success of BNF in soybean in Brazil. Plant
bionts. Proc. Natl. Acad. Sci. USA 99:14887– Soil. 252:1–9
92 Ayres MP, Wilkens RT, Ruel JJ, Lombardero
Abe T, Bignell DE, Higashi M, eds. 2000. Ter- MJ, Vallery E. 2000. Nitrogen budgets of
mites: Evolution, Sociality, Symbiosis, Ecol- phloem-feeding bark beetles with and with-
ogy. Dordrecht: Kluwer Acad. 488 pp. out symbiotic fungi (Coleoptera: Scolyti-
Adams RMM, Mueller UG, Green AM, Naroz- dae). Ecology 81:2198–10
590 MUELLER ET AL.
Baker JM. 1963. Ambrosia beetles and their Browne F. 1961. The biology of Malayan
fungi, with particular reference to Platypus Scolytidae and Platypodidae. Malay. For.
cylindricus Fab. Symp. Soc. Gen. Microbiol. Rec. 22:1–255
13:232–65 Carreiro SC, Pagnocca FC, Bueno OC, Bacci
Barras SJ. 1973. Reduction of progeny and de- M, Hebling MJA, de Silva OA. 1997. Yeasts
velopment in the southern pine beetles fol- associated with the nests of the leaf-cutting
lowing removal of symbiotic fungi. Can. En- ant Atta sexdens rubropilosa Forel, 1908. An-
tomol. 105:1295–99 tonie van Leeuwenhoek 71:243–48
Barrett JA. 1981. The evolutionary conseque- Chapela IH, Rehner SA, Schultz TR, Mueller
nces of monocultures. In Genetic Consequ- UG. 1994. Evolutionary history of the sym-
ences of Man-Made Change, ed. JA Bishop, biosis between fungus-growing ants and their
LM Cook, pp. 209–48. London: Academic. fungi. Science 266:1691–94
409 pp. Craven SE, Dix MD, Michaels GE. 1970. At-
Bass M, Cherrett JM. 1995. Fungal hyphae as tine fungus gardens contain yeasts. Science
a source of nutrients for the leaf-cutting ant 169:184–86
Atta sexdens. Physiol. Entomol. 20:1–6 Currie CR. 2001a. A community of ants,
Batra LR. 1966. Ambrosia fungi: extent of fungi, and bacteria: a multilateral approach
specificity to ambrosia beetles. Science to studying symbiosis. Annu. Rev. Microbiol.
153:193–95 55:357–80
Batra LR. 1967. Ambrosia fungi: a taxonomic Currie CR. 2001b. Prevalence and impact of
revision and nutritional studies of some a virulent parasite on a tripartite mutualism.
species. Mycologia 59:976–1017 Oecologia 128:99–106
Batra LR. 1979. Insect-Fungus Symbiosis. Currie CR, Bot ANM, Boomsma JJ. 2003a. Ex-
Monclair: Allanheld, Osmun & Co. 288 pp. perimental evidence of a tripartite mutual-
Batra LR, Batra SWT. 1979. Termite-fungus ism: bacteria protect ant fungus gardens from
mutalism. See Batra 1979, pp. 117–63 specialized parasites. Oikos 101:91–102
Beaver RA. 1989. Insect-fungus relationships Currie CR, Mueller UG, Malloch D. 1999a. The
in the bark and ambrosia beetles. In Insect- agricultural pathology of ant fungus gardens.
fungus Interactions, 14th Symp. R. Entomol. Proc. Natl. Acad. Sci. USA 96:7998–8002
Soc. London, ed. N Wilding, NM Collins, PM Currie CR, Scott JA, Summerbell RC, Mal-
Hammond, JF Webber, pp. 121–43. London: loch D. 1999b. Fungus-growing ants use
Academic antibiotic-producing bacteria to control gar-
Blackwell M, Jones K. 1997. Taxonomic di- den parasites. Nature 398:701–4
versity and interactions of insect-associated Currie CR, Stuart AE. 2001. Weeding and
ascomycetes. Biodivers. Conserv. 6:689– grooming of pathogens in agriculture by ants.
99 Proc. R. Soc. London Ser. B 268:1033–39
Borden JH. 1988. The striped ambrosia beetles. Currie CR, Wong B, Stuart AE, Schultz TR,
In Dynamics of Forest Insect Populations, Rehner SA, et al. 2003b. Ancient tripartite
ed. AA Berryman, pp. 579–96. New York: coevolution in the attine ant-microbe sym-
Plenum. 624 pp. biosis. Science 299:386–88
Bot ANM, Currie CR, Hart AG, Boomsma Darlington JECP. 1994. Nutrition and evolu-
JJ. 2001a. Waste management in leaf-cutting tion in fungus-growing termites. In Nourish-
ants. Ethol. Ecol. Behav. 13:225–37 ment and Evolution in Insect Societies, ed.
Bot ANM, Rehner SA, Boomsma JJ. 2001b. JH Hunt, CA Nalepa, pp. 105–30. Boulder,
Partial incompatibility between ants and CO: Westview. 449 pp.
symbiotic fungi in two sympatric species de Fine Licht HH, Andersen A, Aanen DK.
of Acromyrmex leaf-cutting ants. Evolution 2005. Termitomyces sp. associated with
55:1980–91 the termite Macrotermes natalensis has a
heterothallic mating system and multinucle- ization of fungal parasites on cultivars in the
ate cells. Mycol. Res. In press attine ant symbiosis. Proc. R. Soc. London
Denison RF, Kiers ET, West SA. 2003. Dar- Ser. B 271:1791–98
winian agriculture: When can humans find Grassé PP. 1959. Une nouveau type de sym-
solutions beyond the reach of natural selec- biose: La meule alimentaire des termites
tion? Q. Rev. Biol. 78:145–68 champignonnistes. Nature 3293:385–89
Diamond J. 1997. Guns, Germs, and Steel: the Green AM, Adams RM, Mueller UG. 2002. Ex-
Fates of Human Societies. New York: Nor- tensive exchange of fungal cultivars between
ton. 480 pp. two sympatric species of fungus-growing
Farrell BD, Sesqueira AS, O’Meara BC, Nor- ants. Mol. Ecol. 11:191–95
mark BB, Chung JH, et al. 2001. The evo- Green RE, Cornell SJ, Scharlemann JPW,
lution of agriculture in beetles (Curculion- Balmford A. 2005. Farming and the fate of
idae: Scolytinae and Platypodinae). Evolu- wild nature. Science 307:550–55
tion 55:2011–27 Gustafson DM, Boe A, Jin Y. 2003. Genetic
Fernández-Marı́n H, Zimmerman JK, Wcislo variation for Puccinia emaculata infection in
WT. 2004. Ecological traits and evolutionary switchgrass. Crop Sci. 43:755–59
sequences of nest establishment in fungus- Haanstad JO, Norris DM. 1985. Microbial sym-
growing ants (Hymenoptera, Formicidae, bionts of the ambrosia beetle Xyloterinus
Attini). Biol. J. Linn. Soc. 81:39–48 politus. Microb. Ecol. 11:267–76
Finlay RD. 2004. Mycorrhizal fungi and their Haas D, Keel C. 2003. Regulation of antibiotic
multifunctional roles. Mycologist 18:91–96 production in root-colonizing Pseudomonas
Fisher PJ, Stradling DJ, Sutton BC, Petrini LE. spp. and relevance for biological control of
1995. Microfungi in the fungus gardens of plant disease. Annu. Rev. Phytopathol. 41:
the leaf-cutting ant Atta cephalotes: a pre- 117–53
liminary study. Mycol. Res. 100:541–46 Hamilton WD, Axelrod R, Tanese R. 1990. Sex-
Francke-Grosmann H. 1967. Ectosymbiosis in ual reproduction as an adaptation to resist
wood-inhabiting insects. In Symbiosis, ed. parasites (a review). Proc. Natl. Acad. Sci.
SM Henry, 2:141–205. New York: Aca- USA 87:3566–73
demic. 443 pp. Harrington TC. 2005. Ecology and evolution
French JRJ, Roeper RA. 1972. Interactions of mycophagous bark beetles and their fun-
of the ambrosia beetle Xyleborus dispar gal partners. See Vega & Blackwell 2005,
(Coleoptera: Scolytidae) with its symbiotic pp. 257–91
fungus, Ambrosiella hartigii (Fungi Imper- Hart AG, Anderson C, Ratnieks FL. 2002. Task
fecti). Can. Entomol. 104:1635–41 partinioning in leafcutting ants. Acta Ethol.
Futuyma DJ, Slatkin M. 1983. Coevolution. 5:1–11
Sunderland, MA: Sinauer. 555 pp. Herre EA, Knowlton N, Mueller UG, Rehner
Garbeva P, van Veen JA, van Elsas JD. 2004. SA. 1999. The evolution of mutualisms: ex-
Microbial diversity in soil: selection of mi- ploring the paths between conflict and coop-
crobial populations by plants and soil type eration. Trends Ecol. Evol. 14:49–53
and implications for disease suppressiveness. Hölldobler B, Wilson EO. 1990. The Ants. Cam-
Annu. Rev. Phytopathol. 42:243–70 bridge: Harvard Univ. Press. 732 pp.
Gebhardt H, Bergerow D, Oberwinkler F. 2004. Hood ME. 2003. Dynamics of multiple in-
Identification of the ambrosia fungus of fection and within-host competition by the
Xyleborus monographus and X. dryographus anther-smut pathogen. Am. Nat. 162:122–
(Curculionidae, Scolytinae). Mycol. Prog. 33
3:95–102 Huber J. 1905. Über die Koloniegründung bei
Gerardo NM, Mueller UG, Price SL, Currie CR. Atta sexdens. Biol. Cent. 25:606–19, 625–35
2004. Exploitation of a mutualism: special- Jaenike J. 1978. A hypothesis to account for the
592 MUELLER ET AL.
maintenance of sex within populations. Evol. ior in Insects and Arachnids, ed. JC Choe, BJ
Theory 3:191–94 Crespi, pp. 181–215. Cambridge, UK: Cam-
Janzen DH. 1966. Coevolution of mutualism bridge Univ. Press. 561 pp.
between ants and acacias in Central Amer- Kok LT, Norris DM, Chu HM. 1970. Sterol
ica. Evolution 20:249–75 metabolism as a basis for a mutualistic sym-
Johansson JF, Paul LR, Finlay RD. 2004. Mi- biosis. Nature 225:661–62
crobial interactions in the mycorrhizosphere Korb J, Aanen DK. 2003. The evolution of uni-
and their significance for sustainable agricul- parental transmission of fungal symbionts in
ture. FEMS Microbiol. Ecol. 48:1–13 fungus-growing termites (Macrotermitinae).
Johnson RA. 1981. Colony development and es- Behav. Ecol. Sociobiol. 53:65–71
tablishment of the fungus comb in Microter- Kweskin M. 2003. Molecular and behavioral
mes sp. nr. usambaricus (Isoptera, Macroter- ecology of fungus-growing ants and their
mitinae) from Nigeria. Insect. Soc. 28:3–12 fungi. MA thesis. Univ. Tex., Austin. 79 pp.
Johnson RA, Thomas RJ, Wood TG, Swift MJ. LaPolla JS, Mueller UG, Seid M, Cover
1981. The inoculation of the fungus comb in SP. 2002. Predation by the army ant
newly founded colonies of the Macrotermiti- Neivamyrmex rugulosus on the fungus-
nae (Isoptera). J. Nat. Hist. 15:751–56 growing ant Trachymyrmex arizonensis. In-
Jones KG, Blackwell M. 1998. Phylogenetic sect. Soc. 49:251–56
analysis of ambrosial species in the genus Leuthold RH, Badertscher S, Imboden H. 1989.
Raffaelea based on 18S rDNA sequences. The inoculation of newly formed fungus
Mycol. Res. 102:661–65 comb with Termitomyces in Macrotermes
Jordal BH. 2002. Elongation factor 1 resolves colonies (Isoptera, Macrotermitinae). Insect.
the monophyly of the haploid ambrosia bee- Soc. 36:328–38
tles Xyleborini (Coleoptera: Curculionidae). Lindow SE, Brandl MT. 2003. Microbiology of
Insect Mol. Biol. 11:453–65 the phyllosphere. Appl. Environ. Microbiol.
Katoh H, Miura T, Maekawi K, Shinzato N, 69:1875–83
Matsumoto T. 2002. Genetic variation of Malloch D, Blackwell M. 1993. Dispersal bi-
symbiotic fungi cultivated by the macroter- ology of the ophiostomatoid fungi. In Cer-
mitine termite Odontotermes formosanus atocystis and Ophiostoma: Taxonomy, Ecol-
(Isoptera: Termitidae) in the Ryukyu Archi- ogy and Pathogenicity, ed. MJ Wingfield, KA
pelago. Mol. Ecol. 11:1565–72 Seifert, J Webber, pp. 195–206. St. Paul: Am.
Kent DS, Simpson JS. 1992. Eusociality Phytopathol. Soc. 304 pp.
in the beetle Austroplatypus incomper- Maschwitz U, Hölldobler B. 1970. Der Kar-
tus (Coleoptera: Curculionidae). Naturwis- tonbau bei Lasius fuliginosus Latr. (Hy-
senschaften 79:86–87 menoptera: Formicidae). Z. Vgl. Physiol.
Kingsolver JG, Norris DM. 1977. External mor- 66:176–89
phology of Xyleborus ferrugineus (Fabr.) Maschwitz U, Koob K, Schildknecht H. 1970.
(Coleoptera: Scolytidae) I. Head and protho- Ein Beitrag zur Funktion der Metatho-
rax of adult male and females. J. Morphol. rakaldrüse der Ameisen. J. Insect Physiol.
154:147–56 16:387–404
Kinuura H. 1995. Symbiotic fungi associated Matsuura K, Tanaka C, Nishida T. 2000. Sym-
with ambrosia beetles. Jpn. Agric. Res. Q. biosis of a termite and a sclerotium-forming
29:57–63 fungus: Sclerotia mimic termite eggs. Ecol.
Kirkendall LR, Kent DS, Raffa KF. 1997. Inter- Res. 15:405–14
actions among males, females, and offspring Mazzola M. 2004. Assessment and manage-
in bark and ambrosia beetles: the significance ment of soil microbial community structure
of living in tunnels for the evolution of so- for disease suppression. Annu. Rev. Phy-
cial behavior. In Evolution of Social Behav- topathol. 42:35–59
Mitchell CE, Tilman D, Groth JV. 2002. Effects rooms by fungus-growing ants. Proc. R. Soc.
of grassland plant species diversity, abun- London Ser. B 271:1777–82
dance, and composition on foliar fungal dis- Murakami T, Higashi S. 1997. Social or-
ease. Ecology 83:1713–26 ganization in two primitive attine ants,
Morelet M. 1998. Une espece nouvelle de Cyphomyrmex rimosus and Myrmicocrypta
Raffaelea, isolee de Platypus cylindrus, ednaella, with reference to their fungus sub-
coleoptere xylomycetophage des chenes. strates and food sources. J. Ethol. 15:17–
Extr. Ann. Soc. Sci. Nat. Archeol. Toulon Var 25
50:185–93 Narisawa K, Kawamata H, Currah RS, Hashiba
Morrissey JP, Dow JM, Mark GL, O’Gara F. T. 2002. Suppression of Verticillium wilt in
2004. Are microbes at the root of a solution to eggplant by some fungal root endophytes.
world food production? EMBO Rep. 10:922– Eur. J. Plant Pathol. 108:103–9
26 Norris DM. 1965. The complex of fungi essen-
Morrissey JP, Walsh UF, O’Donnell A, tial to growth and development of Xyleborus
Moënne-Loccoz Y, O’Gara F. 2002. Ex- sharpi in wood. Mater. Org. Beih. 1:523–29
ploitation of genetically modified inoculants Norris DM. 1972. Dependence of fertility and
for industrial ecology applications. Antonie progeny development of Xyleborus ferrug-
van Leeuwenhoek 81:599–606 ineus upon chemicals from its symbiotes.
Mueller UG. 2002. Ant versus fungus versus In Insect and Mite Nutrition, ed. JC Ro-
mutualism: Ant-cultivar conflict and the de- driguez, pp. 299–310. North-Holland: Am-
construction of the attine ant-fungus symbio- sterdam. 702 pp.
sis. Am. Nat. 160(Suppl.):S67–98 Norris DM. 1979. The mutualistic fungi of
Mueller UG, Gerardo N. 2002. Fungus-farming Xyleborini beetles. See Batra 1979, pp. 53–
insects: Multiple origins and diverse evolu- 65
tionary histories. Proc. Natl. Acad. Sci. USA Ortius-Lechner D, Maile R, Morgan ED,
99:15247–49 Boomsma JJ. 2000. Metapleural gland secre-
Mueller UG, Lipari SE, Milgroom MG. 1996. tions of the leaf-cutter ant Acromyrmex oc-
Amplified Fragment Length Polymorphism tospinosus: New compounds and their func-
(AFLP) fingerprinting of fungi cultured by tional significance. J. Chem. Ecol. 26:1667–
the fungus-growing ant Cyphomyrmex min- 83
utus. Mol. Ecol. 5:119–22 Paulitz TC, Bélanger RR. 2001. Biological con-
Mueller UG, Poulin J, Adams RMM. 2004. trol in greenhouse systems. Annu. Rev. Phy-
Symbiont choice in a fungus-growing ant topathol. 39:103–33
(Attini, Formicidae). Behav. Ecol. 15:357– Peacock L, Hunter T, Turner H, Brain P. 2001.
64 Does host genotype diversity affect the distri-
Mueller UG, Rehner SA, Schultz TR. 1998. bution of insect and disease damage in willow
The evolution of agriculture in ants. Science cropping systems? J. Appl. Ecol. 38:1070–
281:2034–38 81
Mueller UG, Schultz TR, Currie CR, Adams Piper JK, Handley MK, Kulakow PA. 1996. In-
RMM, Malloch D. 2001. The origin of the at- cidence and severity of viral disease symp-
tine ant-fungus mutualism. Q. Rev. Biol. 76: toms on eastern gamagrass in monoculture
169–97 and polycultures. Agric. Ecosyst. Environ.
Mundt CC. 2002. Use of multiline cultivars and 59:139–47
cultivar mixtures for disease management. Poulsen M, Boomsma JJ. 2005. Mutualis-
Annu. Rev. Phytopathol. 40:381–410 tic fungi control crop-diversity in fungus-
Munkacsi AB, Pan JJ, Villesen P, Mueller UG, growing ants. Science 307:741–44
Blackwell M, et al. 2004 Convergent coevo- Price SL, Murakami T, Mueller UG, Schultz
lution in the domestication of coral mush- TR, Currie CR. 2003. Recent findings in
594 MUELLER ET AL.
fungus-growing ants: Evolution, ecology, sexdens rubropilosa. FEMS Microbiol. Lett.

and behavior of a complex microbial sym- 239:319–23
biosis. In Genes, Behavior, and Evolution Schultz TR, Meier R. 1995. A phylogenetic
in Social Insects, ed. M Kikuchi, S Higashi, analysis of the fungus-growing ants (Formi-
pp. 255–80. Sapporo: Hokkaido Univ. Press. cidae: Attini) based on morphological char-
314 pp. acters of the larvae. Syst. Entomol. 20:337–
Rindos D. 1984. The Origins of Agriculture. 70
Orlando: Academic. 325 pp. Schultz TR, Mueller UG, Currie CR, Rehner
Rissing SW, Pollock GB, Higgins MR, Ha- SA. 2005. Reciprocal illumination: A com-
gen RH, Smith DR. 1989. Foraging spe- parison of agriculture in humans and ants.
cialization without relatedness or dominance See Vega & Blackwell 2005, pp. 149–90
among co-founding ant queens. Nature 338: Schultz TR, Solomon SA, Mueller UG, Ville-
420–22 sen P, Boomsma JJ, et al. 2002. Cryptic
Roeper RA, Treeful LM, O’Brien KM, Foote speciation in the fungus-growing ants Cyph-
RA, Bunce MA. 1980. Life history of the am- omyrmex longiscapus Weber and Cyphomyr-
brosia beetle Xyleborus affinis (Coleoptera: mex muelleri Schultz and Solomon, new
Scolytidae) from in vitro culture. Great Lakes species (Formicidae, Attini). Insect. Soc. 49:
Entomol. 13:141–44 331–43
Rollins F, Jones KG, Krokene P, Solheim H, Shipton PJ. 1977. Monoculture and soil borne
Blackwell M. 2001. Phylogeny of asexual pathogens. Annu. Rev. Phytopathol. 15:387–
fungi associated with bark and ambrosia bee- 407
tles. Mycologia 93:991–96 Sieber R. 1983. Establishment of fungus comb
Rosengaus RB, Guldin MR, Traniello JFA. in laboratory colonies of Macrotermes mich-
1998. Inhibitory effect of termite fecal pel- aelseni and Odontotermes montanus (Isopt-
lets on fungal spore germination. J. Chem. era, Macrotermitinae). Insect. Soc. 30:204–9
Ecol. 24:1697–706 Silliman BR, Newell SY. 2003. Fungus farm-
Rosengaus RB, Traniello JFA, Lefebvre ML, ing in snails. Proc. Natl. Acad. Sci. USA 100:
Maxmen AB. 2004. Fungistatic activity of 15643–48
the sternal gland secretion of the dampwood Six DL. 2003. Bark beetle-fungus symbiosis. In
termite Zootermopsis angusticollis. Insect. Insect Symbiosis, ed. T Miller, K Kourtzis,
Soc. 51:259–64 pp. 99–116. Boca Raton, FL: CRC. 368 pp.
Rouland-Lefevre C. 2000. Symbiosis with Six DL, Klepzig KD. 2004. Dendroctonus bark
fungi. See Abe et al. 2000, pp. 289–306 beetles as model systems for studies on sym-
Sachs J, Mueller UG, Wilcox TP, Bull JJ. 2004. biosis. Symbiosis 37:207–32
The evolution of cooperation. Q. Rev. Biol. Six DL, Paine TD. 1998. Effects of mycangial
79:135–60 fungi and host tree species on progeny sur-
Sanchez-Peña, SR. 2005. New view on origin of vival and emergence of Dendroctonus pon-
attine ant-fungus mutualism: exploitation of derosae (Coleoptera: Scolytidae). Environ.
a preexisting insect-fungus symbiosis (Hy- Entomol. 27:1393–1401
menoptera: Formicidae). Ann. Entomol. Soc. Six DL, Paine TD. 1999. Allozyme diversity
Am. 98:151–64 and gene flow in Ophiostoma clavigerum
Sands WA. 1956. Some factors affecting the (Ophiostomatales: Ophiostomataceae), the
survival of Odontotermes badius. Insect. Soc. mycangial fungus of the Jeffrey pine beetle,
3:531–36 Dendroctonus jeffreyi (Coleoptera: Scolyti-
Santos AV, Dillon RJ, Dillon VM, Reynolds dae). Can. J. For. Res. 29:324–31
SE, Samuels RI. 2004. Occurrence of the Smith BD. 1998. The Emergence of Agricul-
antibiotic producing bacterium Burkholderia ture. New York: Sci. Am. Libr. 230 pp.
sp. in colonies of the leaf-cutting ant Atta Sturz AV, Christie BR, Nowak J. 2000.
Bacterial endophytes: Potential role in de- Weller DM, Raaijmakers JM, Gardener BB,
veloping sustainable systems of crop produc- Thomashow LS. 2002. Microbial population
tion. Crit. Rev. Plant Sci. 19:1–30 responses for specific soil suppressiveness to
Traniello JFA, Leuthold RH. 2000. The behav- plant pathogens. Annu. Rev. Phytopathol. 40:
ioral ecology of foraging in termites. See Abe 309–48
et al. 2000, pp. 141–68 Whipps JM. 2001. Microbial interactions and
Vega F, Blackwell M, eds. 2005. Insect-Fungal biocontrol in the rhizosphere. J. Exp. Bot. 52:
Associations: Ecology and Evolution. Ox- 487–511
ford: Oxford Univ. Press. 333 pp. Wille P, Boller T, Kaltz O. 2001. Mixed inocu-
Villesen P, Mueller UG, Schultz TR, Adams lation alters infection success of strains of the
RMM, Bouck MC. 2004. Evolution of ant- endophyte Epichole bromicola on its grass
cultivar specialization and cultivar switching host Bromus erectus. Proc. R. Soc. London
in Apterostigma fungus-growing ants. Evo- Ser. B 269:397–402
lution 58:2252–65 Wolfe MS. 1985. The current status and
von Arx JA, Hennebert GL. 1965. Deux prospects of multiline cultivars and variety
champignons ambrosia. Mycopathol. Mycol. mixtures for disease resistance. Annu. Rev.
Appl. 25:309–15 Phytopathol. 23:251–73
Wardle DA, Bardgett RD, Klironomos JN, Wood SL. 1982. The bark and ambrosia beetles
Setälä H, van der Putten WH, Wall DH. 2004. of North and Central America (Coleoptera:
Ecological linkages between aboveground Scolytidae) a taxonomic monograph. Great
and belowground biota. Science 304:1629– Basin Nat. Mem. 6:1–1359
33 Zhu Z, Chen H, Fan J, Wang Y, Li Y, Chen
Weber NA. 1972. Gardening Ants: The Attines. J, et al. 2000. Genetic diversity and disease
Philadelphia: Am. Philos. Soc. 146 pp. control in rice. Nature 406:718–22
Mueller.qxd 8/19/05 1:04 PM Page 1
EVOLUTION OF AGRICULTURE IN INSECTS C-1
See legend on next page

C-2 MUELLER ET AL.
Figure 1 Comparison of agriculture in attine ants, xyleborine beetles, macrotermitine ter-

mites, and humans. The time-series (left to right) highlight the roles of beneficial auxil-
iary microbes (blue shading) that suppress diseases (black dots) or aid in buffering against
contaminant microbes (orange shading). (a) Ant agriculture. Ants attempt to clean conta-
minant microbes from garden substrate (not shown) and remove garden diseases (black
dots) through active weeding (top ant). The ants (bottom ant) then plant a crop-microbe
consortium (crop plus beneficial auxiliary blue microbes) onto the prepared substrate,
spreading beneficial microbes through the garden matrix. (b) Beetle agriculture. Primary
fungus (crop) lining the tunnel grows intermixed with secondary microbes (blue shading)
and occasional contaminant microbes (orange dots). No disease microbes (black dots) are
indicated because they very rarely occur in young gardens near a tunnel head. The exact
roles of the secondary microbes in beetle fungiculture are still unknown. (c) Termite agri-
culture. Hypothetical passage of a mixture of crop spores, auxiliary microbes, and sub-
strate (ingested plant material) through the gut of a termite, followed by defecation of the
substrate-crop-microbe consortium in fecal pellets that the termite adds to new garden.
Other (external) modes of crop-microbe copropagation may exist in termite farmers, par-
alleling the planting of crop-microbe consortia in attine ants. Selective passage of
microbes through the alimentary canals of attine ants and ambrosia beetles is unknown,
but has never been investigated. No disease microbes (black dots) are indicated because
no specialized pathogens have yet been identified in the fungus-growing termite system.
(d) Human agriculture (wheat). A seed, fortuitously planted in soil enriched in antibiotic-
secreting rhizosphere bacteria (blue-shaded soil), grows into a vigorous, disease-resilient
crop plant (Weller et al. 2002). However, crops are often planted in microbially subopti-
mal soil (orange-shaded soil), leading to higher disease loads (black dots) on such plants.
Traditional human planting schemes passage crops through a seed stage without copropa-
gating disease-suppressant rhizosphere microbes. Illustrations by Barrett Klein.
EVOLUTION OF AGRICULTURE IN INSECTS C-3
Figure 2 Evolutionary histories of insect agriculture. (a–c) Comparison of the patterns of

evolutionary diversification in the insect farmers (left cladograms) and their cultivated
fungi (right cladograms). In the left cladograms, farmer lineages are black and nonfarmer
relatives are gray, whereas in the right cladograms, cultivated fungal lineages (cultivars)
are black and noncultivated feral fungal lineages are gray. Independent origins of agricul-
tural behavior are indicated for each farmer clade in the left cladograms, and independent-
ly domesticated fungal lineages appear as separate cultivar lineages in the right clado-
grams. (d) Garden of the fungus-growing ant Atta texana (photo by Greg Dimijian). The
workers are cleaning and shredding leaf cuttings before expanding new gardens through
the addition of leaf material. (e) Garden of the fungus-growing termite Macrotermes belli-
cosus (photo by Karen Machielsen). The fungus is grown on fecal pellets that are stacked
into lamellar walls of the fungus garden (comb). ( f ) Gallery of the ambrosia beetle
Trypodendron lineatum (photo by Susanne Kühnholz) with ambrosia fungus (black) lining
the main gallery and beetle brood developing in niches adjacent to the gallery. Galleries are
constantly patrolled by adult beetles (not shown). Figure adapted from Mueller & Gerardo
2002.

Evolution of Agriculture in Insects

Uploaded by

Document Informationclick to expand document informationAgriculture has evolved independently in three insect orders. All propagate their cultivars as clonal monocultures within their nests. Insect farmers have evolved a combination of strategies to manage crop diseases.

Document Informationclick to expand document information

Copyright:

Available Formats

Evolution of Agriculture in Insects

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Evolution of Agriculture in Insects

Uploaded by

Copyright:

Available Formats

19 Aug 2005 15:44 AR ANRV259-ES36-24.

tex XMLPublishSM (2004/02/24) P1: OJO

THE EVOLUTION OF AGRICULTURE IN INSECTS

Key Words Attini, Macrotermitinae, mutualism, symbiosis, Xyleborini

564 MUELLER ET AL.

1.1. Behavioral and Nutritional Elements

Macrotermitine Xyleborine ambrosia

Agricultural behavior Attine ants termites beetles Humans

Dependency on crop for food Obligate Obligate Obligate Facultative

Sustainable harvesting of crop for food Present Present Present Present

Learning and cultural transmission of agricultural Absent Absent Absent Present

566 MUELLER ET AL.

1.2. A Coevolutionary Approach to

2. THE THREE INSECT-AGRICULTURE SYSTEMS

EVOLUTION OF AGRICULTURE IN INSECTS 567

2.1. Ant Fungiculture

2.2. Termite Fungiculture

568 MUELLER ET AL.

2.3. Beetle Fungiculture

EVOLUTION OF AGRICULTURE IN INSECTS 569

3. EVOLUTIONARY ORIGINS OF INSECT AGRICULTURE

nest walls for structural or Sphaerotermes lineages; these other beetle

(tribe Leucocoprineae). Ceratocystis.

macrotermite lineages; all

species are specialized in species are specialized in The fungal lineages

EVOLUTION OF AGRICULTURE IN INSECTS

Crop sharing and crop

Lateral exchange of Present Unknown Unknown Present

same cultivar lineage.

domesticated and wild regularly from wild Andes) are based on

populations; or import of populations; cultivars of regular genetic exchange

events generating the two agricultural (e.g., banana).

recombinants. systems with vertical

574 MUELLER ET AL.

EVOLUTION OF AGRICULTURE IN INSECTS 575

4. AGRICULTURAL EVOLUTION AND ECOLOGY

4.1. Cultivar Transmission Between Farmer Generations

evolution Attine ants Macrotermitine termites beetles Humans

unspecialized (Trichoderma) specialized as weeds of bacteria. At least some of by invertebrate and

Guarding or protecting Present Present Present Uncommon

minimize contact gardening; a contaminated worker castes in many

monitoring of all crop

Escovopsis parasites. sanitary function in the garden

mediate resistance of some

578 MUELLER ET AL.

4.2. Higher-Level Specialization (Clade-Clade Congruence)

4.3. Lower-Level Specialization on Cultivars

EVOLUTION OF AGRICULTURE IN INSECTS 579

4.4. Cultivar Sharing and Exchange Between Farmer Species

580 MUELLER ET AL.

4.5. Propagation of Sexual Versus Asexual Cultivars and

EVOLUTION OF AGRICULTURE IN INSECTS 581

termite Macrotermes natalensis, for example, has an outcrossing mating system

4.6. Coevolutionary Modifications

582 MUELLER ET AL.

yet to be investigated include predominantly asexual reproduction in the insect

4.7. Symbiont Choice and “Artificial Selection” of Cultivars