486
John F. Valentine and J. Emmett Duffy
It is also possible, of course, that the failure to
demonstrate predator limitation of mesograzers reflects a true absence of top-down control. To evaluate this possibility, and the relative importance of
resource and predator control of mesograzer populations, G.J. Edgar (see below) initiated a series of
studies designed to test between them. Experimental manipulations of mesograzer abundance, both
enhancement and reduction, in field microcosms
(Edgar 1990; Edgar and Aoki 1993) resulted in rapid
compensatory responses of the mesograzer assemblage and convergence of estimated secondary production values, consistent with diffuse competition
and limitation by a common resource, presumably
periphyton production (the “production ceiling” hypothesis, Edgar 1993). These experiments were complemented with field surveys using artificial plants of
standardized construction in a wide range of shallow
marine habitats and geographic regions to characterize epifaunal assemblages, and relate their characteristics to environmental parameters (Edgar 1993;
Edgar and Klumpp 2003). Specifically, abundance
and size of epifaunal animals were used to estimate
epifaunal biomass and, based on published data on
specific production rates, biomass was used to estimate a standardized index of total epifaunal secondary production, P 20 , representing the aggregate
daily secondary production of epifauna normalized
to 20
◦ C (Edgar 1993). Surprisingly, P 20 showed no
significant variation among standardized habitats deployed at 17 shallow, clear-water sites worldwide,
whereas the index was strongly reduced at deeper,
turbid, and nutrient-depleted sites (Edgar 1993), and
varied with architecture of the artificial substratum
(Edgar and Klumpp 2003). These patterns are consistent with limitation of epifaunal productivity by
resources available for primary production (a “production ceiling”, Edgar 1993), whereas the lack of
variation among sites is unexpected if spatial variation in predation strongly influences epifaunal populations. Thus, the suggestion was that total secondary
production of epifauna was regulated by resource
availability rather than predation.
The resource ceiling hypothesis and its supporting data are intriguing and potentially quite important. They raise a philosophical question: which aspects of ecological structure and function are we
concerned about and why? Although the resource
ceiling hypothesis, an explicitly bottom-up hypothesis, appears superficially to be incompatible with
hypotheses of predator regulation of epifauna, it is
not necessarily so, and focuses attention on the somewhat difficult question of what we mean by “regulation” or “control”. The resource ceiling hypothesis suggests that a specific ecosystem-level process,
the rate of production of an entire trophic level or
functional group, is controlled (limited) by resource
availability. Yet the source of this secondary production varies considerably among systems and, in
particular, with predator density. In one of the same
studies that supported the resource ceiling hypothesis, Edgar and Aoki (1993) showed that fish predation tended to eliminate larger-bodied animals,
shifting the epifaunal assemblage toward species resistant to predation, and toward the smallest bodysize classes in species vulnerable to predation (e.g.,
Edgar and Aoki 1993). Thus, both epifaunal demography and species composition changed substantially with predator density despite similar secondary production levels. Such shifts are likely to
have profound consequences both for grazer effects
on plants, and for the transmission of seagrass epiphytic production to higher trophic levels, including
harvested fishes and shellfish. First, recent studies
have shown that even superficially similar mesograzer species differ considerably in their grazing
rates, food preferences, and per capita impacts on
the plant community (Fig. 5; Duffy and Hay 2000;
Duffy and Harvilicz 2001; Duffy et al. 2001, 2003).
As major macrophyte-grazing taxa tend to be largerbodied species, which will often be severely reduced
by predation, predation on these grazers may cascade
down to affect biomass and plant community composition. Second, the similarity of total secondary
production estimates between treatments with and
without predators presumably does not translate to
equivalent secondary production flowing to higher
trophic levels, since production by assemblages exposed to fishes is almost entirely due to minute size
classes not effectively captured by fishes (Edgar
and Aoki 1993). Again, as emphasized by Paine
(1980), energy-flow diagrams (including secondary
production levels) and community interaction webs
for the same community may have quite different
appearances.
For mesograzers specifically, this lack of concordance between energetic and interaction estimates
of ecological importance has been demonstrated in
several experiments. For example, one microcosm
experiment compared the grazing impacts and population growth rates of ampithoid amphipods (a mixture of Ampithoe longimana and Cymadusa compta)
John F. Valentine and J. Emmett Duffy
It is also possible, of course, that the failure to
demonstrate predator limitation of mesograzers reflects a true absence of top-down control. To evaluate this possibility, and the relative importance of
resource and predator control of mesograzer populations, G.J. Edgar (see below) initiated a series of
studies designed to test between them. Experimental manipulations of mesograzer abundance, both
enhancement and reduction, in field microcosms
(Edgar 1990; Edgar and Aoki 1993) resulted in rapid
compensatory responses of the mesograzer assemblage and convergence of estimated secondary production values, consistent with diffuse competition
and limitation by a common resource, presumably
periphyton production (the “production ceiling” hypothesis, Edgar 1993). These experiments were complemented with field surveys using artificial plants of
standardized construction in a wide range of shallow
marine habitats and geographic regions to characterize epifaunal assemblages, and relate their characteristics to environmental parameters (Edgar 1993;
Edgar and Klumpp 2003). Specifically, abundance
and size of epifaunal animals were used to estimate
epifaunal biomass and, based on published data on
specific production rates, biomass was used to estimate a standardized index of total epifaunal secondary production, P 20 , representing the aggregate
daily secondary production of epifauna normalized
to 20
◦ C (Edgar 1993). Surprisingly, P 20 showed no
significant variation among standardized habitats deployed at 17 shallow, clear-water sites worldwide,
whereas the index was strongly reduced at deeper,
turbid, and nutrient-depleted sites (Edgar 1993), and
varied with architecture of the artificial substratum
(Edgar and Klumpp 2003). These patterns are consistent with limitation of epifaunal productivity by
resources available for primary production (a “production ceiling”, Edgar 1993), whereas the lack of
variation among sites is unexpected if spatial variation in predation strongly influences epifaunal populations. Thus, the suggestion was that total secondary
production of epifauna was regulated by resource
availability rather than predation.
The resource ceiling hypothesis and its supporting data are intriguing and potentially quite important. They raise a philosophical question: which aspects of ecological structure and function are we
concerned about and why? Although the resource
ceiling hypothesis, an explicitly bottom-up hypothesis, appears superficially to be incompatible with
hypotheses of predator regulation of epifauna, it is
not necessarily so, and focuses attention on the somewhat difficult question of what we mean by “regulation” or “control”. The resource ceiling hypothesis suggests that a specific ecosystem-level process,
the rate of production of an entire trophic level or
functional group, is controlled (limited) by resource
availability. Yet the source of this secondary production varies considerably among systems and, in
particular, with predator density. In one of the same
studies that supported the resource ceiling hypothesis, Edgar and Aoki (1993) showed that fish predation tended to eliminate larger-bodied animals,
shifting the epifaunal assemblage toward species resistant to predation, and toward the smallest bodysize classes in species vulnerable to predation (e.g.,
Edgar and Aoki 1993). Thus, both epifaunal demography and species composition changed substantially with predator density despite similar secondary production levels. Such shifts are likely to
have profound consequences both for grazer effects
on plants, and for the transmission of seagrass epiphytic production to higher trophic levels, including
harvested fishes and shellfish. First, recent studies
have shown that even superficially similar mesograzer species differ considerably in their grazing
rates, food preferences, and per capita impacts on
the plant community (Fig. 5; Duffy and Hay 2000;
Duffy and Harvilicz 2001; Duffy et al. 2001, 2003).
As major macrophyte-grazing taxa tend to be largerbodied species, which will often be severely reduced
by predation, predation on these grazers may cascade
down to affect biomass and plant community composition. Second, the similarity of total secondary
production estimates between treatments with and
without predators presumably does not translate to
equivalent secondary production flowing to higher
trophic levels, since production by assemblages exposed to fishes is almost entirely due to minute size
classes not effectively captured by fishes (Edgar
and Aoki 1993). Again, as emphasized by Paine
(1980), energy-flow diagrams (including secondary
production levels) and community interaction webs
for the same community may have quite different
appearances.
For mesograzers specifically, this lack of concordance between energetic and interaction estimates
of ecological importance has been demonstrated in
several experiments. For example, one microcosm
experiment compared the grazing impacts and population growth rates of ampithoid amphipods (a mixture of Ampithoe longimana and Cymadusa compta)