Tropical Arboreal Ants 3392004b, Davidsonet al.2004, Davidson 2005).
Ideally also, future tests of the “manipulated
competition hypothesis” should compare EFN and
trophobiont data with dietary assays of ants from
the same communities (contraFigure 20.1, see
legend). Further, some ant taxa respond to mix-
tures of amino acids and sugars in individualistic
and non-additive ways (e.g., Lanzaet al.1993,
Blüthgen and Fiedler 2004b), and rigorous tests
of the hypothesis must eventually take such com-
plications into account. Finally, in at least some
EFN plants, changes in nectar quality can be
induced by herbivory (Stephenson 1982, Smith
et al.1990), and the same may be true of honey-
dew (Fischer and Shingleton 2001, Fischeret al.
2002). In the single case where sugars and amino
acids were monitored simultaneously, amino acid
content but not sugar content increased following
simulated herbivory (Smithet al.1990). However,
more studies are needed to determine whether
constitutive EFNs are proportionally richer in
sugars and poorer in N than are nectars produced
after herbivory.
ANT ASSOCIATIONS WITH
ENDOSYMBIONTS
In general, predatory ants might be anticipated
to be more effective than are herbivorous or
omnivorous ants in deterring herbivorous insects.
Although stable isotope technology has proven
useful in ranking arboreal ant taxa from lesser
to greater dependence on predation and scav-
enging (Blüthgenet al.2003, Davidsonet al.
2003),δ^15 N values are likely imperfect predic-
tors of trophic levels. Thus, although the very
lowδ^15 Nvaluesof somearborealexudate-feeders,
especially formicines and cephalotines, suggest
that these taxa feed as herbivores or highly her-
bivorous omnivores, an alternative or additional
explanation is possible. Isotopic fractionation dur-
ing putative N-recycling by symbiotic bacteria
may produce differential retention of light N, pref-
erentially released during biochemical reactions
and then transaminated to convert non-essential
to essential amino acids. Because most ants are
omnivores, the potential exists for gradually mag-
nifying colony N reserves by recycling N acquired
through consumption of fungi (authors’ unpub-
lished data), and/or hunted or scavenged prey,
including hemolymph (Zientzet al.2005). Given
the opportunity, selection to counteract stoichio-
metric imbalances should have favored such rela-
tionships in N-deprived taxa, as in the N-limited
sap-feeding trophobionts often tended by ants
(e.g., Douglaset al.2001). Recently published
evidence shows N-recycling and upgrading by
microsymbionts of ants (Feldhaaret al.2007).
Moreover, coupled with autocoprophagy and anal
trophalaxis, the considerable urate stores in the
Dolichoderusfat body suggest possible urate recy-
cling in this genus (Cook and Davidson 2006).
If any or all of these groups do recycle or fix N,
and those processes measurably lowerδ^15 N, then
exudate-feeding, omnivorous ants in these taxa
would be less “herbivorous” than isotopes indi-
cate, and therefore potentially more beneficial and
less harmful to the plants on which they forage.
An important goal of continued research in
this area is to understand what fractions of ants’
N budgets come from plant secretions (including
those filtered through trophobionts), as opposed
to carnivory followed by N-recycling. This goal
will likely prove difficult to achieve in anything
other than “closed systems” of ants and myrme-
cophytes. Moreover, identifying ant taxa with the
greatestpotentialforprovidinganti-herbivorepro-
tection to plants would require distinguishing N
acquiredviapredationversusscavenging.General
effectsof antsontheforestmaybeeasiesttodefine
by direct tests, that is, ant removal experiments,
and would be most useful if focused on particular
taxonomic and functional groups (see below).FORAGING FUNCTIONAL GROUPS
Many studies of ant–ant, ant–trophobiont, and
ant–plant interactions are formulated in terms
of “generic” ants, differing perhaps in body size
and/or colony size, but with few or no other
defining features. Nevertheless, achieving a pre-
dictive understanding of interspecific interactions
in ant communities, as well as of the roles these
insects play in tropical ecosystems, will likely
depend more on advancements in our knowledge
of their functional biology than on attempts to