University of Nebraska - Lincoln University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln
Faculty Publications in the Biological Sciences Papers in the Biological Sciences
2001
Contrasting Effects of Plant Richness and Composition on Insect Contrasting Effects of Plant Richness and Composition on Insect
Communities: A Field Experiment Communities: A Field Experiment
Nick M. Haddad
University of Minnesota
David Tilman
University of Minnesota
John Haarstad
University of Minnesota
Mark Ritchie
Utah State University
Johannes M.H. Knops
University of Nebraska-Lincoln
, jknops@unl.edu
Follow this and additional works at: https://digitalcommons.unl.edu/bioscifacpub
Part of the Life Sciences Commons
Haddad, Nick M.; Tilman, David; Haarstad, John; Ritchie, Mark; and Knops, Johannes M.H., "Contrasting
Effects of Plant Richness and Composition on Insect Communities: A Field Experiment" (2001).
Faculty
Publications in the Biological Sciences
. 130.
https://digitalcommons.unl.edu/bioscifacpub/130
This Article is brought to you for free and open access by the Papers in the Biological Sciences at
DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Faculty Publications in the
Biological Sciences by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln.
vol. 158, no. 1 the american naturalist july 2001
Contrasting Effects of Plant Richness and Composition on
Insect Communities: A Field Experiment
Nick M. Haddad,
1,
* David Tilman,
1
John Haarstad,
1
Mark Ritchie,
2
and Johannes M. H. Knops
1,†
1. Department of Ecology, Evolution, and Behavior, University of
Minnesota, St. Paul, Minnesota 55108;
2. Department of Fisheries and Wildlife, Utah State University,
Logan, Utah 84322-5210
Submitted June 22, 2000; Accepted February 23, 2001
abstract: We experimentally separated the effects of two com-
ponents of plant diversity—plant species richness and plant func-
tional group richness—on insect communities. Plant species rich-
ness and plant functional group richness had contrasting effects
on insect abundances, a result we attributed to three factors. First,
lower insect abundances at higher plant functional group richness
were explained by a sampling effect, which was caused by the
increasing likelihood that one low-quality group, C
4
grasses, would
be present and reduce average insect abundances by 25%. Second,
plant biomass, which was positively related to plant functional
group richness, had a strong, positive effect on insect abundances.
Third, a positive effect of plant species richness on insect abun-
dances may have been caused by greater availability of alternate
plant resources or greater vegetational structure. In addition, a
greater diversity of insect species, whose individual abundances
were often unaffected by changes in plant species richness, may
have generated higher total community abundances. After con-
trolling for the strong, positive influence of insect abundance on
insect diversity through rarefaction, insect species richness in-
creased as plant species richness and plant functional group rich-
ness increased. Although these variables did not explain a high
proportion of variation individually, plant species richness and
plant functional group richness had similar effects on insect di-
versity and opposing effects on insect abundances, and both factors
may explain how the loss of plant diversity influences higher
trophic levels.
Keywords: abundance, composition, diversity, functional groups,
insects, species richness.
* Present address: Department of Zoology, Box 7617, North Carolina State
University, Raleigh, North Carolina 27695-7617; e-mail: nick_haddad@ncsu.edu.
†
Present address: Department of Biological Sciences, University of Nebraska,
Lincoln, Nebraska 68588.
Am. Nat. 2001. Vol. 158, pp. 17–35. 䉷 2001 by The University of Chicago.
0003-0147/2001/15801-0002$03.00. All rights reserved.
The loss of plant diversity has been reported to cause higher
insect abundances, particularly abundances of specialist in-
sect pests (Elton 1958; Pimentel 1961; Root 1973; Kareiva
1983; Risch et al. 1983; Strong et al. 1984; Andow 1991),
and to lower insect species richness (Murdoch et al. 1972;
Southwood et al. 1979; Strong et al. 1984; Siemann et al.
1998; Knops et al. 1999). Herbivore abundances are thought
to be higher in plant monocultures, where specialist her-
bivores are more likely to find and to remain on their hosts
and/or generalist predators are less abundant (Root 1973).
Empirical studies have shown that specialist herbivores have
higher reproductive rates and higher immigration into but
lower emigration from monocultures than polycultures
(Bach 1980a, 1980b, 1984; Risch 1981; Kareiva 1985; Elms-
trom et al. 1988). Insect species richness, especially the rich-
ness of specialist herbivores, is thought to increase with
increasing plant species richness because a greater diversity
of plants provides a greater diversity of resources for insects
(Murdoch et al. 1972; Southwood et al. 1979; Strong et al.
1984; Siemann et al. 1998; Knops et al. 1999). Higher di-
versity of herbivorous insects may then support a higher
diversity of insect predators and parasitoids (Hunter and
Price 1992; Knops et al. 1999).
Despite this previous work, it remains unclear whether
these changes in insect communities are driven more by
changes in the number of plant species or changes in plant
community composition that are usually associated with
changes in plant diversity. Interest in this question has
emerged from recent studies of effects of plant biodiversity,
which have shown that plant community composition can
strongly affect ecosystem processes (e.g., Tilman et al.
1997b; Hooper and Vitousek 1998; Symstad et al. 1998).
Plant composition is often determined by separating plants
into functional groups, that is, groups of species that differ
in physiology, phenology, and morphology—and thus also
in their effects on ecosystem processes (Chapin et al. 1996).
One important conclusion of recent studies is that the
effects of diversity on ecosystem properties may result from
both changes in the number of species and changes in
plant functional group composition (Hooper and Vitousek
18 The American Naturalist
1997, 1998; Tilman et al. 1997b; Symstad et al. 1998; Hec-
tor et al. 1999).
Although plant functional groups in these studies were
not chosen specifically for their effects on insects, insects
might be expected to respond to changes in plant func-
tional group composition for two reasons. First, plants
within the same functional group, which are often deter-
mined with respect to plant resource use, have some sim-
ilarities in tissue quality. For example, legumes are typically
separated from other forbs because they fix nitrogen. Their
high tissue nitrogen would then provide a higher-quality
resource for herbivores (e.g., Mattson 1980; Strong et al.
1984). Likewise, grasses are often divided by their pho-
tosynthetic pathway (which also corresponds to differences
in growing season). Warm-season grasses that photosyn-
thesize via the C
4
pathway generally have low tissue ni-
trogen, higher toughness, and structural characteristics
that protect starches and nutrients from herbivores (Cas-
well et al. 1973). Thus, they are poorer-quality food re-
sources for herbivores (Boutton et al. 1978; Kroh 1978;
Pinder and Kroh 1987). The second reason insect com-
munities might respond to plant composition is that func-
tional groups separate plants into species that are more
similar taxonomically. Insects are often specialists within
a plant genus or family, and functional groups encompass
the food plants of these herbivores.
We tested two primary hypotheses regarding the effects
of plant species richness, plant functional group richness,
and plant functional group composition on insect com-
munities in a grassland experiment. First, we tested the
hypothesis that higher plant diversity causes lower total
insect abundances and that the negative effects of plant
diversity are best explained by higher plant functional
group richness rather than by higher plant species richness.
This hypothesis arises from work in agroecosystems, where
monocultures have higher insect abundances (measured
either for individual species or for the entire insect com-
munity) than polycultures composed of very different
plant functional groups, such as corn (grass), beans (leg-
ume), and squash (forb; Bach 1980a; Risch 1981; Andow
1990), or agricultural crops and weeds (Pimentel 1961;
Tahvanainen and Root 1972; Root 1973). If, as we hy-
pothesized, higher plant functional group richness causes
lower total insect abundances and explains the negative
relationship between plant diversity and insect abun-
dances, then changes in plant species richness within func-
tional groups would have no effect on insect abundances.
An alternative hypothesis is that plant species richness has
an opposite, positive effect on total insect abundances. This
hypothesis has some support from evidence of associa-
tional susceptibility, whereby a less preferred host attracts
more herbivores in polyculture because it is planted with
a more preferred host (Bach 1980b; Brown and Ewel 1987;
Wahl and Hay 1995) and from evidence of polycultures
increasing abundances of generalist herbivores that can
exploit alternative hosts (Risch 1981).
Our second primary hypothesis was that higher plant
species richness, rather than higher plant functional group
richness, best explains higher insect species richness. In
other experiments, insect species richness has been shown
to be positively related to plant species richness (Knops et
al. 1999) and to plant functional group richness (Siemann
et al. 1998; Symstad et al. 2000). We did not replicate the
results of previous studies. Here, we separated the effects
of plant species richness, plant functional group richness,
and plant functional group composition on insect diversity.
Our hypothesis that plant species richness has the strongest
positive effect on insect species richness was based on the
observation that the diversity of insects is often correlated
with the diversity of resources (e.g., Murdoch et al. 1972;
Strong et al. 1984; Siemann et al. 1998; Knops et al. 1999),
regardless of plant functional types. In addition to studies
of the entire insect community, we also analyzed the re-
sponses of the abundance and diversity of each insect
trophic group and herbivore feeding guild.
The two community attributes that are the focus of this
article, insect species richness and insect abundance, may
each help to inform the response of the other factor. Insect
species richness is often positively correlated with total insect
abundance because sampling more individuals leads to
higher counts of species. In our analyses of total insect
species richness, we used rarefaction to estimate insect spe-
cies richness before testing for effects of plant species rich-
ness, plant functional group richness, or plant composition.
Knowledge of the number of species may in turn help to
interpret responses of insect abundances in the entire com-
munity. Most previous studies of insect responses to plant
diversity have focused on one insect species (Tahvanainen
and Root 1972; Bach 1980a; Risch 1981; Andow 1991; but
see Root 1973). We examined the responses of several of
the most abundant individual herbivore species to plant
species richness and plant functional group richness, as well
as responses of the entire insect community. The total insect
community integrates the individual responses of many in-
sect species, and the mean response of insects per species
provides a standard to compare effects of plant diversity on
total community abundance to the numerical responses of
individual species.
Material and Methods
Plant Species Functional GroupRichness # Plant
Richness Experiment
To separate the effects of plant diversity on insects, we stud-
ied insect communities in a well-replicated, randomized
Plant Diversity and Insect Communities 19
Table 1: Plant species and their functional group
designations in the experiment
Plant species Plant functional group
Achillea millefolium Forb
Agropyron repens C
3
grass
Agropyron smithii C
3
grass
Amorpha canescens Legume
Andropogon gerardii C
4
grass
Asclepias tuberosa Forb
Astragalus canadensis Legume
Baptisia leucantha Legume
Bouteloua curtipendula C
4
grass
Bouteloua gracilis C
4
grass
Bromus inermis C
3
grass
Buchloe dactyloides C
4
grass
Calamagrostis canadensis C
3
grass
Coreopsis palmata Forb
Elymus canadensis C
3
grass
Koeleria cristata C
3
grass
Leersia oryzoides C
3
grass
Lespedeza capitata Legume
Liatris aspera Forb
Lupinus perennis Legume
Monarda fistulosa Forb
Panicum virgatum C
4
grass
Petalostemum candidum Legume
Petalostemum purpureum Legume
Petalostemum villosum Legume
Poa pratensis C
3
grass
Quercus ellipsoidalis Woody
Quercus macrocarpa Woody
Rudbeckia hirta Forb
Schizachyrium scoparium C
4
grass
Solidago nemoralis Forb
Solidago rigida Forb
Sorghastrum nutans C
4
grass
Sporobolus cryptandrus C
4
grass
Stipa spartea C
3
grass
Vicia villosa Legume
Zizea aurea Forb
experiment where both plant species richness and plant
functional group richness were manipulated (described in
Tilman et al. 1997b; Siemann et al. 1998; Knops et al. 1999).
The experiment was developed to test for effects of plant
biodiversity on a host of community and ecosystem prop-
erties, including responses of insects. The experiment was
conducted at Cedar Creek Natural History Area in east-
central Minnesota and consisted of 342 experimental plots,
each 169 m
2
, that formed an -plot grid. The ex-18 # 19
periment was created in 1994 when plots were planted with
either zero, one, two, four, eight, 16, or 32 perennial, sa-
vannah grassland species representing zero to five functional
groups (table 1; Tilman et al. 1997b). Insects were sampled
in 285 of the 342 plots that were designed to separate the
effects of plant species richness and plant functional group
richness. Plots were created in three ways. First, we created
163 plots by randomly drawing one, two, four, eight, or 16
species from a pool of 18 species that represented five func-
tional groups. Second, to balance the design with a similar
number of plots containing each combination of plant spe-
cies richness and plant functional group richness, we created
76 additional plots by first randomly drawing one to three
functional groups (from the pool of five functional groups)
and then randomly drawing two, four, or eight species in
those functional groups that were contained in a larger pool
of 34 species. The expanded pool of species was needed to
create, for example, plots with eight species from one func-
tional group, which was not possible in random draws from
the pool of 16 species. Third, we created 46 additional plots
with the highest level of plant species richness and plant
functional group richness by planting 32 species chosen
from the pool of 34 species. We attempted to maintain
treatment levels by periodic removal of weeds and appli-
cation of herbicides, and four species with poor germination
success were replaced in 1995. However, because all species
did not germinate in all plots where they were planted and
because weeding did not eradicate all unwanted species,
imposed levels of plant species richness were approximate.
All plots were burned in May 1997 to prevent litter accu-
mulation, which could affect plant species composition. We
recognize that burning could influence insect communities.
Siemann et al. (1997) found individualistic responses of
many species to fire at Cedar Creek; however, they found
little effect of fire frequency on total insect species richness
or abundance. In addition, in a review of the history of fire
at Cedar Creek, Tilman et al. (2000) found evidence for
annual to biennial burn frequency.
Plant species were classified into five functional groups
based on their physiological, phenological, and morpho-
logical characteristics, which included their resource
requirements, seasonality of growth, and life history. Func-
tional groups were chosen based on plant attributes within
ecosystems and not with respect to their impacts on in-
sects. However, the classification organized plants in ways
that have relevance to insects, particularly in their taxo-
nomic relatedness and in their relative tissue quality. Cool-
season grasses that photosynthesize via the C
3
pathway
have higher tissue nitrogen than do warm-season grasses
that photosynthesize via the C
4
pathway. Forbs are her-
baceous dicots. Legumes are forbs that fix nitrogen, the
limiting nutrient at Cedar Creek (Tilman 1987). Woody
plants produce a perennial stem.
Plant community and nutrient responses to the manip-
ulations were described in Tilman et al. (1997b). In each
plot, we estimated actual plant species richness and per-
centage cover by each plant species in four -m sub-0.5 # 1
plots and then took the average values from the four sub-
20 The American Naturalist
plots to generate one estimate per plot. We measured peak
aboveground living plant biomass (a strong correlate of
aboveground plant productivity because there was little
overwintering aboveground production) by clipping four
-m strips per plot that were then combined, dried0.1 # 3.0
to a constant dry weight at 55⬚C, and weighed. To measure
plant tissue nitrogen and carbon, plant samples were ground
and analyzed on a Carlo Erba NA 1500 elemental analyzer
(Carlo Erba Instruments, Milan).
Insect Samples
Insects and terrestrial arthropods were collected three times
during the season of greatest plant production, on June 20,
July 28, and August 22, 1997. We swept each of the 285
plots with a 38-cm-diameter muslin net, which we swung
25 times while walking in a line 3 m from the plot edge.
We swept plots rather than sampling by another method
(like suction sampling) because we decided to cover a larger
area that would better represent plot level characteristics of
the plant and insect community and because previous work
at Cedar Creek has demonstrated that insect community
responses to changes in diversity were similar when insects
were collected by sweeping or by a D-vac (Siemann 1998;
N. M. Haddad, unpublished data). One bias of sweep sam-
pling is against some leaf minors and galling insects, which
are often specialized and would thus respond to changes in
plant species richness and resource concentration. Speci-
mens were identified to species or morphospecies within
known genera or families and counted.
Insect abundance was quantified as the total number
of individuals. A second analysis that is not presented
here quantified insect abundance as insect biovolume, an
approximation of biomass, which was calculated as the
average product of the maximum length, width, and
thickness of each species (Siemann et al. 1996; Haddad
et al. 2000). The number of individual insects and insect
biovolume were highly correlated (Pearson correlation
; ), and results were quali-coefficient p 0.608 P p .001
tatively similar to the results of analyses of total number
of individuals.
To determine the effects of plant species richness, plant
functional group richness, and plant functional group
composition on insect trophic structure, insect species
were classified into one of five trophic categories based on
field observations and literature review. Herbivores, par-
asitoids, predators, detritivores, and omnivores were clas-
sified by whether they fed, respectively, on live plant tissue,
within other animals, on insects that they killed, on dead
plant or animal tissues or by-products, or on combinations
of food sources. Herbivores were further divided into one
of four feeding guilds: chewing, sucking, boring, or seed/
pollen feeding. Insects occupying different trophic levels
in different stages of their life cycles were classified based
on their larval stage unless we could identify a species’
adult food resources within the experiment. A small num-
ber of individuals with aquatic larval stages were difficult
to classify using the above criteria and were excluded from
trophic analyses.
Analysis
We used backward elimination multiple regression to sep-
arate the effects of plant species richness, plant functional
group richness, and plant functional group composition
on insect abundance and diversity. The effects of plant
species richness and plant functional group richness are
necessarily correlated (i.e., a plot with one species must
have one functional group). However, well-replicated and
randomized experiments where the levels of plant species
richness and plant functional group richness are specifi-
cally manipulated can be used to distinguish the effects of
each variable through multiple regression. In most of our
analyses, several measures indicated that these variables
were not markedly collinear and that their effects could
be legitimately separated.
The experimental design included many replicates at
each combination of one to eight species and one to five
functional groups. At higher levels of diversity, however,
the design was incomplete. Because plant species com-
position was constrained by fixed pools of 18 or 34 species,
plots with 16 or 32 species always had four or five func-
tional groups (see description of experiment; Tilman et al.
1997b; Siemann et al. 1998). When we conducted separate
analyses that included only plots containing up to eight
species, our results did not differ qualitatively from those
using all plots. Because of this, we decided to retain plots
at all levels of plant species richness.
Although there was a strong relationship between planned
levels of plant species richness and actual plant species rich-
ness (Pearson correlation ; ),coefficient p 0.66 P p .001
the two values differed because plant cover plots were small,
weedy species were not completely eradicated, and some
species failed to establish, especially in high-diversity plots.
Because actual plant species richness explained more vari-
ation in every analysis of insect species richness and abun-
dance (see also Knops et al. 1999), we used it as our in-
dependent variable in analyses. We analyzed the effects of
plant composition by including a dummy variable indicat-
ing the presence or absence of each plant functional group.
To account for the effect of the number of individual insects
within a plot on insect species richness, we rarefied our total
community data to estimate species richness based on the
plot with lowest total abundance using EcoSim software
(Gotelli and Entsminger 2000).
We included three other variables that may affect insect
