Reprinted from Nature, Vol. 341, No.6238, pp. 142-144, 14th
September, 1989
@ Macmi//an Magazines Lid., 1989
Ecosystem.level patterns ofprimary productivity and herbivoryin
terrestrial habitats
s. J. McNaughton, M. Oesterheld, D. A. Frank
& K. J. Williams
Biological Research Laboratories, Syracuse University,
Syracuse.New York 13244-1220. USA
FIG. 1 Relationship between net above.ground primary
productivity (NAP,abscissa) and herbivore biomass. consumption. and
net secondary produc.tivity (NSP). Units are kJ m-2 yr-1, except
for biomass which is kJ m-2, Keyto type of ecosystem: 1. desert; 2.
tundra; 3, temperate grassland; 4.temperate successional old field;
5. unmanaged tropical grassland; 6,temperate forest; 7. tropical
forest; 8. salt marsh; and 9. agriculturaltropical grassland. The
data set consisted of 51 points for 8. 69 for C and35 for NSP.
tnrough primary producers is indicative of the three
majorenergetic properties of the trophic level directly above.
Inspection of the goodness of fit between consumption andnet
primary productivity, revealed that forests fell below theline and
grasslands above. Dominance by heterotherm her-bivores in forests
and homeotherms in grasslands, and thegreater prevalence of
secondary chemicals in trees than ingrasses11, were initially
accepted as explanation of this pattern.Much of the net
productivity of forests, however, is confined towood, a poor
quality food44 largely unavailable to most her-bivores. Therefore,
we re-examined the relationship, confiningnet productivity to
foliage (NFP). Then, for the larger numberof studies in which
foliage production and consumption wereconcurrently measured in
forests, the relationship was (Fig. 2)
log C = 2.04 (Iog NFP) -4.80
(,2 = 0.594, p< 0.00001, d.f. = 73). Restricting primary
produc-tivity data to foliage, therefore, decreased the unexplained
vari-ance substantially by bringing forested ecosystems into line
withthe patterns across other ecosystems.
We infer that secondary chemicals and differences in the
basicphysiologies of consumers are much less important
influencesupon consumption at the ecosystem level than they are at
theindividual plant and population levels. Other than
contributingto an essentially unavailable food class, namely
wood44, secon-dar.y chemicals aooear to be more imDortant a"
re"ulalnr" ill
ECOSySTEMS are structurally organized as food webs withinwhich
energy is transmitted between trophic levels and dissipatedinto the
environment. Energy flow between two trophic levels isgiven by the
amount of production at the lower level and by theproportion of
production that is consumed, assimilated and res-pired at the
higher level. Considerable evidence indicates thatfood-web
structure varies predictably in different habitatsl-5, butmuch less
is known about quantitative relationships among foodweb fluxes.
Many of the energetic properties of herbivores inAfrican game parks
are associated with rainfall and, by inference,with net primary
productivity6.'. Respiratory costs per unit produc-tion at the
consumer trophic level are higher for homeothermsthan for
heterotherms8. Plant secondary chemicals affect herbivoredietary
choices9.lo and the allocation of plant resources tothose chemicals
varies with resource availabilityll. How thesephenomena are
translated into ecosystem fluxes is unknown. Wepresent evidence
that herbivore biomass, consumption and produc-tivity are closely
correlated with plant productivity, suggestingthat the latter is a
principal integrator and indicator of functionalprocesses in food
webs.
A data base was compiled, using sets of data reported in
theliterature, which combined measurements of herbivore
biomass,consumption and/or net secondary productivity with
measuresof net primary productivity6.7.12-39. Unpublished estimates
ofconsumption and net primary productivity in unmanagedAfrican
grasslands, from a study described earlier40, were alsoincluded.
Cases in which herbivore trophic-level properties wereestimated
from primary productivity were excluded. Values ofherbivore
biomass, consumption and net secondary productivityrarely include
all populations of herbivores in an ecosystem.Consequently, data
were included only 1fthe preponderance ofthe herbivore level was
accounted for in the original study. Onlyprocesses occurring above
ground were included, as comprehen-sive below-ground measurements
were too few in numbers forquantitative analysis. Fluxes expressed
in the literature as unitsof mass were converted to energy using
standard conversionfactors41-43. Values of net primary production,
net secondaryproduction and consumption are presented in kJ m-2
yr-l, andvalues of herbivore biomass are in kJ m-2. Net primary
produc-tivity ranged from -125 to -29,000kJm-2yr-l. Wide rangesof
ecosystems and primary productivities were thus encom-passed by the
assembled data.
There was a positive association between primary productivityand
all three energetic properties of herbivore communities,with
different types of ecosystems segregating along commonlines (Fig. I
). Herbivore biomass (8) was related (coefficient ofdetermination
,2 = 0.579, p < 0.00001, degrees of freedom(d.f.) = 49) to net
primary productivity (NAP) by
log 8 = !.52 (log NAP) -4.79
where log indicates common logarithms. Consumption ( C) andNAP
were related (,2 = 0.367, p < 0.00001, d.f. = 67) by
log C = 1.38 (Iog NAP) -2.32
Finally, net secondary productivity (NSP) was related (r =0.364,
p < 0.0001, d.f. = 34) to NAP by
log NSP= 1.10 (Iog NAP) -3.27
These data indicate that net energy flux into the food-web
base
