236 Walter P. Carsonetal.
For example, HilleRisLamberset al. (2002) con-
cluded that density-dependent mortality is just
as prevalent in temperate as in tropical forests,
although they acknowledged that the strength
of Janzen–Connell effects could be stronger in
the tropics. In addition, the logic that implicates
the Janzen–Connell hypothesis in the latitudinal
gradient in species richness should also apply
to habitats with substantively different levels of
diversity within the same general latitude (Dirzo
andBoegeChapter5,thisvolume).Tropicalforests
with prolonged flooded or dry seasons generally
have depressed diversity relative to aseasonal trop-
ical forests (Ferreira 2000, Fajardoet al.2005).
This seasonality could also decrease the abun-
dance and impact of natural enemies, therefore
decreasin gthe stren gth of Janzen–Connell effects
(Dirzo and Boege Chapter 5). Nevertheless, very
few studies have investigated the Janzen–Connell
hypothesis in abiotically stressful tropical habi-
tats (but see Johnet al. 2002), and none have
attempted to compare results from abiotically
stressful and abiotically more benign forests. Over-
all, studies are required that assess the occurrence
and strength of Janzen–Connell effects and their
consequences both amon gand within latitudes
or wherever there are steep stress gradients and
sharply contrastin gpatterns of species diversity
(e.g., length of dry season).
The Janzen–Connell hypothesis is
difficult to test and falsify and may be
most prevalent for uncommon or
rare species
As Janzen (1970) pointed out (p. 521) heavy pre-
dation may keep some species rare and widely
spaced and these species may also be the best com-
petitors. If so, the species most likely to form dense
aggregations and reduce diversity in local stands
may be the ones least likely to be studied by ecolo-
gists either because they are so uncommon or it is
reasoned that such rare species are unlikely to be
regulated by density dependence (Wright 2002).
This means that if only a fraction of woody species
in diverse tropical forest are actually kept in check
by their predators, the Janzen–Connell model still
holds if these species are the superior competi-
tors in the community or can establish and grow
rapidly throughout the habitat, or both. Therefore
the failure to find Janzen–Connell effects for what
could be hundreds of tree species does not reject
the hypothesis (though it does reduce the impor-
tance of the hypothesis for explainin gcoexistence
of all species in the community). For example,
Conditet al. (1992) studied patterns of recruit-
ment around reproductive adults of 80 species.
They found repelled recruitment syndromes for
just15speciesandconcludedthatJanzen–Connell
effects occurred over short distances and for
few species (note that recent work suggests they
underestimated Janzen–Connell effects; see Leigh
1999). If, however, only a few of these tree
species were excellent competitors and pests actu-
ally caused these patterns, then these pests would
be keystone species and Janzen–Connell would
be operating. Hyattet al.(2003) did not recog-
nize this possibility when they concluded that
there was “no general support for the distance
dependent prediction of the hypothesis and...
further testin gto explore this hypothesis [Janzen–
Connell] as a diversity-maintainin gmechanism
is unnecessary.” Yet Hyattet al. (2003) did find
that there were “individual cases of conformity
to the hypothesis,” which is all that is needed
for the hypothesis to work if the specific cases
represent tree species that are excellent competi-
tors, highly shade tolerant, or habitat generalists,
or some combination of these traits (e.g., Silman
et al. 2003). Testin gwhether repelled recruit-
ment syndromes are stron gfor uncommon tree
species, particularly shade-tolerant species with
large seeds that are putatively attractive to seed
predators or enemies, may give insight into this
problem of testin gthe Janzen–Connell hypoth-
esis. Additionally, exclusion experiments (using
exclosures or insecticides) nested beneath these
trees may lead to dense and depauperate stands
of juvenile conspecifics in less than a decade.
Asdiscussedabove,theJanzen–Connellmodelis
at its core a community-level model where the key
prediction is that predation ultimately leads indi-
rectly to the maintenance of high woody species
diversityintropicalforests.Toidentifypredationas
the indirect cause of woody species diversity will
require long-term experiments that preclude other