Tropical Forest Community Ecology

Tropical Arboreal Ants 337

gain no detectable N from plant or trophobiont
exudates (Davidsonet al.2003). At a range of
taxonomic levels (species/species groups, genera,
and tribes), N-deprivation declined marginally
with increasing trophic level (i.e., higherδ^15 N
ratio) and significantly with increasing body
size. Overall, use of N-free or less “N-dense”
chemical weaponry by N-OH taxa appeared to
reduce N-deprivation. (However, among rela-
tives with similarly N-free weaponry, the pat-
tern with trophic level was reversed, that
is, the most N-limited taxa were often more
carnivorous.) Groups with proteinaceous ven-
oms tended to be more carnivorous and
less N-deprived than did those with alka-
loidal venoms or N-free chemical weaponry,
and most small-bodied ants fell into the latter
category.
In general then, exudate-foraging ants tend
to have highly imbalanced and N-poor diets,
and despite deploying excess CHOs to obtain
protein, many such taxa remain N-deprived.
This is particularly true of several small-bodied
taxa with relatively carnivorous diets for their
subfamilies. Perhaps not coincidentally, such
taxa include several genera rich in specialized
plant-ants (e.g., Davidson and McKey 1993):
AztecaandCrematogaster(Davidson 2005), and
alsoMyrmelachista(with just a single plant-ant
surveyed to date; Davidson unpublished data).
Together, these observations suggest that myrme-
cophytes could have evolved filters to selectively
attract the most carnivorous of N-limited ant taxa
that had previously responded to N-poor diets
by reducing N investment in chemical weaponry.
The next section describes how myrmecophilous
plants, interacting more casually and opportunis-
tically with ants, might vary CHO:N ratios of ant
rewards to similar effect.

THE ROLE OF STOICHIOMETR YIN

OPPORTUNISTIC ANT–PLANT

INTERACTIONS

In contrast to dimensionless ERs, threshold sugar

and amino acid values for feeding cannot be

compared independently of one another among

taxa because they are influenced by such factors

100 w,c

M

M CO W

SUCmin – ants

EFN – plants

AAmin – ants

EFN – plants

80

40

20

0

100

% of taxa

% of taxa

80

60

40

20

0

135

% Total AAs

7

10 30 50

% Total sugars

(a)

(b)

70 100

Figure 20.1 Comparisons of SUCmin and AAmin for

54 Amazonian and 33 Bornean ant taxa (Davidson

2005 and unpublished) versus (a) total sugars and

(b) total amino acids (AAs) in EFNs of 16 Australian

tropical rainforest plant taxa (Blüthgen and Fiedler

2004a; all EFNs used mainly by ants). Arrows designate

(a) individual values (not included in columns) for

wound secretions (W) or trophobiont honeydews

from the same study: C, Cicadelidae; CO, Coccidae;

M, Membracidae; and (b) all values for these same

wound and trophobiont secretions (asterisk). All

concentrations are wt/vol. Median SUCmin values were

identical (4.00) in Brunei and Peru, and lumped values

were significantly lower than % total sugars in EFN

(12.90,X 12 =26.67,P<0.0001 in a Wilcoxon test).

Median AAmin values did not differ between

Amazonian and Bornean ants (1.00 and 0.21,

respectively,P0.05), and lumped values

significantly exceeded the median for EFN (0.10,

X 12 =18.28,P<0.0001 in a Wilcoxon test).

as proximity to nest sites (Davidson 1978) and

natural enemies (Nonacs and Dill 1991, Blüthgen

and Fiedler 2004b). Nonetheless, we can ask how

the distribution of minimum acceptable concen-

trations across all assayed ant taxa compares with

total sugar and amino acid concentrations in EFN

and honeydew for tropical plants and tropho-

bionts. Figure 20.1 presents such comparisons