266 A. Elizabeth Arnold
detected. However, the systemic defense hypoth-
esis is generally not supported by field observa-
tions: new endophyte infections accumulate in
tissues followin ginitial colonization, with stron g
evidence suggesting that early colonists do not
deter later infections (until leaves are saturated
with endophytes; see Arnold and Herre 2003).
Instead, these data raise the possibility that endo-
phytes evade or otherwise do not activate plant
defenses.
Do endophytes interact directly with
pathogens?
Endophytes have the capacity to interact directly
with pathogens within the leaves they share.
When Arnold et al. (2003) found that the
anti-pathogen effects of endophytes were appar-
ently restricted to the leaves that bore those
endophytes, they concluded that direct or indi-
rect interactions between endophytes and the
Phytophthorapathogen were responsible for lim-
itin gthe patho gen’s spread. Under this scenario,
the metaphor of “leaf as landscape” is apt: either
through direct or indirect competition, or perhaps
mycoparasitism, a robust endophyte community
may limit the ability of invadin gpatho gens to
growrapidlyorextensivelywithinleaves.Whether
endophyte communities are more resistant to
invasion when more diverse or more fully packed
with individuals remains to be assessed, and lends
itself to straightforward experiments.
Do endophytes serve as
entomopathogens?
Several major pathogens of insects, includ-
ingBeauveria bassiana(Lewis and Bin g1991),
Aspergillussp. (Caoet al.2002), andPaecilomyces
sp. (Arnold 2002) have been isolated as endo-
phytes from temperate and tropical plants.
In agroecosystems, endophyte infections by
Beauveria bassianahave been successful in lim-
itin gdama ge to maize by the European corn
borer, a major pest (Lewis and Bin g1991).
Entomopathogenic infections of insects generally
occur via cuticular penetration, rather than by
consumption of infected plant tissues (Rawlins
1984).Thus, plants harborin gentomopatho genic
endophytes benefit from the production of many
fungal propagules on senescent tissues. The
frequency of entomopathogens among tropical
endophytes has not been assessed, but is worth
explorin gwith bioassays (to identify novel ento-
mopathogens)andintermsof screening unnamed
cultures and foliage samples with specific primers
to recover known species of entomopathogenic
fungi. More generally, the potential for plants
to harbor entomopathogenics as symbionts is
worth explorin gin tropical forests (see Elliottet al.
2000).ENDOPHYTES: MUTUALISTS OF
INSECT HERBIVORES?
Inasmuch as endophytes may act to protect
plants against insects, it is also possible that
they serve as attractants of folivores, and/or
may improve forage quality. Herbivorous insects
are directly implicated in pathogen movement
and/or infection success. At Los Tuxtlas, Mexico,
García-Guzman and Dirzo (2001) showed that
folivory and visible symptoms of pathogen dam-
age were positively associated: 43% of surveyed
leaves were damaged by both herbivores and
pathogens, whereas 16% were damaged only by
herbivores, and less than 2% were damaged only
by pathogens. The authors concluded that cuticu-
lar woundin gby insects is important for infection
by pathogenic fungi. The same appears to be true
for endophytes: Arnold (unpublished data) found
that parts ofGustavia superbaleaves that were
damaged by hesperiid larvae had significantly
higher rates of endophyte infection, higher endo-
phyte species richness, and a different community
of endophytes relative to undamaged areas of the
same leaves (Figure 15.4). Similarly, Faeth and
Wilson (1996) surface-sterilized and then artifi-
cially herbivorized livin gleaves ofQuercus emoryi
in Arizona, pairin geach herbivore-dama ged leaf
with an undamaged leaf of similar age and posi-
tion.Attheendof thegrowingseason,endophytes
were more common in damaged leaves, and dam-
aged areas of leaves, than in undamaged tissue.
Leaf dama ge by chewin gor scrapin ginsects,
includin gLepidoptera, Orthoptera, Coleoptera,