Tropical Forest Community Ecology

Testing and Falsifying the Janzen–Connell Hypothesis 211

INTRODUCTION

The Janzen–Connell hypothesis states that
specialist pests and pathogens keep key plant
species rare enough or reduce their competitive
ability enough so as to make space available
for many other species (Janzen 1970, Connell
1971). This idea has a lon ghistory in ecol-
ogy. Ridley (1930, p. xvi), nearly 80 years ago,
remarked: “Where too many plants of one species
are grown together, they are very apt to be
attacked by some pest, insect or fungus....It
is largely due to this also, in Nature, that one-
plant associations are prevented and nullified by
better means of dispersal for the seeds.” Later,
Gillett (1962) presented a non-equilibrium form
of this hypothesis. Finally, Janzen (1970) and
Connell (1971) established this hypothesis as one
of the commonplaces of tropical biology. They
both presented evidence that the seedlings and
saplings of trees exhibit repelled recruitment pat-
terns around adults thereby potentially creating
space for numerous plant species. MacArthur
(1972, 191 ff.) accepted this general idea as true.
Nonetheless,ameta-analysisbyHyattetal.(2003)
found little support for the distance-dependent
prediction of the hypothesis and concluded that
“further testin gto explore this hypothesis as
a diversity-maintainin gmechanism is unneces-
sary.” Although Leighetal. (2004) provided much
indirect evidence in favor of the role of pest pres-
sure in maintainin gtropical tree diversity, the
evidence they provided could be said to allure,
ratherthanextort.Intheend,theywerenotableto
establish the truth of this hypothesis beyond rea-
sonable doubt. Below we review studies that tested
different aspects of the Janzen–Connell hypothe-
sis and address a number of issues and challenges
associated with this hypothesis.

A REVIEW OF STUDIES

TESTING FOR JANZEN–CONNELL

EFFECTS

We searched Web of Science to locate articles

published between 1970 and September 2006

that cited Janzen (1970) and explicitly addressed

the Janzen–Connell hypothesis. We identified

53 appropriate studies (Tables 13.1 and 13.2).

We excluded a relatively small number of studies

(less than five) that assessed only static distribu-

tion patterns of one life-history stage as a function

of distance from the nearest adult conspecific

or a function of conspecific juvenile density. We

felt that these studies were less informative than

those that sampled focal plants through time

to address how distance- and density-dependent

effects influence performance (survivorship and

growth rate), or compared two or more life-history

stages to assess changes in distribution patterns

due to density- or distance-dependent factors.

The majority of the studies (58%) focused

on a single species, 21% studied between two

and nine species, and 21% studied 10 or more

species (Table 13.1). Most studies were purely

observational (51%) whereas 34% were purely

experimental, and 15% used both experimental

and observational approaches. About 75% of the

studies were restricted to seeds and seedlings and

17% focused on saplings. We found only one study

that focused on adults (>10 cm dbh; Stoll and

Newberry 2005) and only two studies considered

all life-history stages (Connellet al. 1984, Silva

Matosetal.1999).Nearlyhalf of thestudies(47%)

failed to mention the seed size of the species under

study, a trait originally thought to be important by

both Janzen and Connell. Fifty of the 53 studies

found evidence consistent with either density or

distance dependency but of these, half provided

no evidence for the mechanism underlyin gthe

pattern. Where a putative mechanism was tenta-

tively identified, there was a near even split among

vertebrates (eight studies), invertebrates (10 stud-

ies), and pathogens (seven studies). Several studies

speculated that intraspecific competition could

underlie Janzen–Connell patterns (Connellet al.

1984, Silva Matosetal. 1999, Stoll and Newberry

2005). Host specificity, another trait thought to

be important by both Janzen and Connell, was

reported in only one third of the studies, proba-

bly because it was unknown. For the nine studies

where it was evaluated, five reported high host

specificity, three reported low specificity, and one

reported the occurrence of both specialists and

generalists.

Mean study duration was 3.5 years (±4 SD)

and ranged from 18 years (Connellet al. 1984)