Testing and Falsifying the Janzen–Connell Hypothesis 217to 0 years (Wright and Dubor 2001 where the
distribution of seeds that showed evidence of ver-
tebrateandinvertebratepredationwasevaluated).
Most studies did not repeatedly sample individu-
als through time but only initially and at the end.
Ninety-two percent of the studies were conducted
at a single site and 40% of studies were conducted
on or very near Barro Colorado Island (BCI). The
bulk of the studies occurred in lowland and moist
tropical forests (72%), with only 13% occurrin gin
other tropical forests (e.g., floodplain, dry, etc.) and
15% in temperate regions (Table 13.3).The major-
ity of studies (58%) either explicitly or implicitly
considered a number of factors contributin gto
plant performance in addition to density or dis-
tance (e.g., light level, drought, heterospecific
abundance).
Only seven studies investigated the impact
of Janzen–Connell effects on species diversity
(Connellet al. 1984, Conditet al. 1992, Willset al.
1997, Webb and Peart 1999, Wills and Condit
1999, Harmset al. 2000, HilleRisLamberset al.
2002, see also Willset al. 2006). Of these stud-
ies, three used data only from BCI and two from
BCI and Pasoh (Table 13.1). Thus, the bulk of
our generalizations regarding the Janzen–Connell
hypothesis come from a single forest plot (BCI).
In Table 13.2, we summarize Janzen–Connell
results, on a per species basis, for 173 species in
49 of the 53 studies from Table 13.1. The four
remainin gstudies did not list their focal species
(Connellet al. 1984, Willset al. 1997, Harms
et al. 2000, Peters 2003). Three of these studies
occurred on BCI or at Pasoh (Willset al. 1997,
Harmset al. 2000, Peters 2003) and thus there
would be some overlap with the species from these
studies and those listed in Table 13.2. Nonethe-
less, we acknowledge that Table 13.2 under-
represents the species for which this hypothesis
has been tested, especially since Peters (2003)
found that approximately 80% of 732 species at
BCI and Pasoh showed patterns consistent with
the Janzen–Connell hypothesis.
The majority of species studied were canopy
and understory trees (81%) whereas few lianas,
palms, and shrubs were considered (Table 13.4).
Density dependence was evaluated for 125 species
(Table 13.2). A species was considered to exhibit
density dependence in Table 13.2 if density
dependence occurred in any part of its life history.
Negative density dependence occurred for 40%
of the species whereas 57% showed no density
dependence or exhibited positive density depen-
dence and 3% had results that varied among
studies. Distance dependence was evaluated for
129 species and 36% had patterns consistent
with Janzen–Connell. Sixty percent had either
no density dependence or survivorship decreased
with distance from adult conspecifics and 4%
had results that varied amon gstudies.
Most species studied (79.8%) were from lowland
tropical forests, some of which had pronounced
dry seasons. Only 7% of species were from dry
tropical forests and less than 2% were from
mature tropical floodplain sites or swampy tropi-
cal habitats. Temperate forests accounted for 10%
of the species studied while there was only one
studyintemperatearidforestsandoneinmontane
tropical forests. A surprisin g63% of the species
studied were studied on or near BCI.
Only 18% of the studies reported seed dry
weights (X=1.8 g±3.9 SD). The abundance
of adults was not reported for 27 species and
most species were simply classified as common
(100 species), moderately abundant (one species),
or uncommon (two species).The mean abundance
of species where reported was 45 adults per ha
(±70.6 SD).
This brief review of studies leads us to the
followin gconclusions. There have now been
ample studies of the distance- and density-
dependent predictions of the Janzen–Connell
hypothesis and many species show Janzen–
Connell effects (Tables 13.1 and 13.2). That said,
however, there have been few studies outside
of the lowland tropics, and Dirzo and Boege
(Chapter 5, this volume) predict that pest pressure
will be reduced when resource availability is more
seasonal and episodic (e.g., dry forest). Addition-
ally, there have been too few studies of life-forms
other than trees (Tables 13.3 and 13.4) and too
few studies of species at locations other than on or
near BCI.
Far greater attention needs to be centered on
the causes of Janzen–Connell effects and the
degree to which they occur in later life-history
stages (post small-sapling stages). Uncommon or
rare species have been largely ignored yet may