problem was that the inefficiency of energy transfer
should predict that longer food chains should occur
in more productive environments, but there was
little empirical support for this pattern (Pimm and
Lawton 1977; Pimm 1982; Post 2002; however, see
Oksanenet al.1981). If anything, longer food chains
and more complex food webs seemed to occur in
less productive situations, rather than in the most
productive environments. This apparent inconsis-
tency must be tempered by the realization that
there were very few detailed descriptions of any
food chains or webs at the time, and most of them
were from aquatic systems (Cohen 1978). The rela-
tively small number of highly resolved food webs
that have since accumulated (Winemiller and
Pianka 1990; Martinez 1991; Dunneet al.2002) sug-
gest that the empirical data used to motivate these
theoretical studies were far from ideal.
Experiments designed to evaluate the roles of en-
ergy limitation or population dynamics in limiting
food chain length take two forms: (1) manipulations
of energy inputs with concordant measurements of
resulting food chain lengths and (2) manipulations
of food chain lengths with concordant measure-
ments of population dynamics. In the first approach,
if more energy allowed longer food chains to devel-
op and persist, then energetics presumably played
some role in setting food chain length. The basic
theory underlying this idea has been developed by
Fretwell (1977) and Oksanenet al.(1981). Assuming
that efficiency of energy transfer between trophic
levels is about 10%, energy availability would have
to be manipulated by over an order of magnitude
(e.g. at least 10-fold) to see the addition or loss of a
top trophic level. It is essential to be able to clearly
place the organisms involved on particular trophic
levels – this means that relatively linear food chains
without substantial omnivory are required. Very
few experiments satisfy these requirements, but
those that do suggest an important role for energy
in determining food chain length.
Jenkinset al.(1992) tested the effects of produc-
tivity on the relatively simple food webs that
develop in water-filled tree-holes in tropical Aus-
tralia. They examined food chain development
after experimentally reducing productivity over
2ordersofmagnitude,areductionthatshould
result in the loss of at least one trophic level from
the top of the food chain. Decreasing productivity
reduced the number of coexisting species, the
number of trophic links and maximum food
chain length. There seemed to be clear evidence
that energy played a role in limiting food chain
length in this system.
Kaunzinger and Morin (1998) used a simple mi-
crobial system to test for effects of productivity on
food chain length. The longest three-level food
chains consisted of a basal level (the bacterium
Serratia marcescens) consumed by a ciliated protist
(Colpidium striatum), which was in turn consumed
by a top predator (the ciliateDidinium nasutum).
The system lacks omnivores, so the trophic position
of each species was known without error. Produc-
tivity was manipulated by varying the nutrient
concentration of growth medium consumed by the
bacteria. Three-level food chains, those containing
the top predatorDidinium, persisted only at higher
levels of productivity (Fig. 1.7). At lower productiv-
ity levels the third trophic level failed to persist,
clearly supporting the role of energy transfer in
limiting the length of simple linear food chains.
Patterns of change in the abundance of species on
each trophic level are also consistent with simple
prey-dependent models of predator–prey inter-
actions (e.g. Leibold 1996), but are not consistent
with ratio-dependent models (e.g. Abrams and
Ginzburg 2000).
There is scant evidence for comparable patterns
in natural systems, perhaps because of the technical
difficulties in unambiguously assigning species to
trophic levels or measuring productivity. Postet al.
(2000) failed to find a relationship between food
chain length and productivity in a survey of natural
lakes, but did find that larger lakes tended to sup-
port longer food chains. Their finding is superficial-
ly similar to the idea that larger predators located
higher in the food chain should require larger home
ranges (Slobodkin 1960), even disproportionally
larger than one allometrically would expect from
their size (Haskellet al.2002). Postet al. relied on the
indirect measurement of the trophic position of top
predators using stable isotope ratios, rather than
using knowledge of the structure of the entire
food chain. Postet al.(2000) and Post (2002) have
suggested that the dependence of food chain length
on energy inputs shown by Jenkinset al.(1992) and20 SHAPE AND STRUCTURE