, 1608 (2013);339 Science et al.Lucas A. Garibaldi
Wild Pollinators Enhance Fruit Set of Crops Regardless of Honey Bee
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles
): April 1, 2013 www.sciencemag.org (this information is current as of
The following resources related to this article are available online at
version of this article at: including high-resolution figures, can be found in the onlineUpdated information and services,
can be found at: Supporting Online Material
found at: can berelated to this article A list of selected additional articles on the Science Web sites
, 14 of which can be accessed free:cites 67 articlesThis article
1 articles hosted by HighWire Press; see:cited by This article has been
Ecology subject collections:This article appears in the following
registered trademark of AAAS. is aScience2013 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on April 1, 2013www.sciencemag.orgDownloaded from
Wild Pollinators Enhance Fruit
Set of Crops Regardless of
Honey Bee Abundance
Lucas A. Garibaldi,
Marcelo A. Aizen,
Saul A. Cunningham,
Luísa G. Carvalheiro,
Lawrence D. Harder,
Natacha P. Chacoff,
Jan H. Dudenhöffer,
Breno M. Freitas,
Steven K. Javorek,
Christina M. Kennedy,
Kristin M. Krewenka,
Margaret M. Mayfield,
Brian A. Nault,
Simon G. Potts,
Taylor H. Ricketts,
Colleen L. Seymour,
Carlos H. Vergara,
Blandina F. Viana,
Thomas C. Wanger,
Alexandra M. Klein
The diversity and abundance of wild insect pollinators have declined in many agricultural landscapes.
Whether such declines reduce crop yields, or are mitigated by managed pollinators such as honey
bees, is unclear. We found universally positive associations of fruit set with flower visitation by wild
insects in 41 crop systems worldwide. In contrast, fruit set increased significantly with flower visitation
by honey bees in only 14% of the systems surveyed. Overall, wild insects pollinated crops more
effectively; an increase in wild insect visitation enhanced fruit set by twice as much as an equivalent
increase in honey bee visitation. Visitation by wild insects and honey bees promoted fruit set
independently, so pollination by managed honey bees supplemented, rather than substituted for,
pollination by wild insects. Our results suggest that new practices for integrated management of
both honey bees and diverse wild insect assemblages will enhance global crop yields.
Human persistence depends on many nat-
ural processes, termed ecosystem ser-
vices, which are usually not accounted
for in market valuations. The global degrada-
tion of such services can undermine the ability
of agriculture to meet the demands of the grow-
ing, increasingly affluent, human population (1,2).
Pollination of crop flowers by wild insects is
one such vulnerable ecosystem service (3), as the
abundance and diversity of these insects are de-
clining in many agricultural landscapes (4,5).
Globally, yields of insect-pollinated crops are
often managed for greater pollination through
the addition of honey bees (Apis mellifera L.)
as an agricultural input (Fig. 1) (6–8). Therefore,
the potential impact of wild pollinator decline on
crop yields is largely unknown. Nor is it known
whether increasing application of honey bees (9)
compensates for losses of wild pollinators, or
even promotes these losses.
Fruit set, the proportion of a plant’sflowers
that develop into mature fruits or seeds, is a key
component of crop yield (fig. S1). Wild insects
may increase fruit set by contributing to polli-
nator abundance, species number (richness),
equity in relative species abundance (evenness),
or some combination of these factors. Increased
pollinator abundance, and therefore the rate of
visitation to crop flowers, should augment fruit
set at a decelerating rate until additional in-
dividuals do not further increase fruit set (e.g.,
pollen saturation) or even decrease fruit set (e.g.,
pollen excess) (10–12). Richness of pollinator
species should increase the mean, and reduce
the variance, of fruit set (13) because of comple-
mentary pollination among species (14,15), fa-
cilitation (16,17), or “sampling effects”(18),
among other mechanisms (19,20). Pollinator
evenness may enhance fruit set via comple-
mentarity, or diminish it if a dominant species
(e.g., honey bee) is the most effective pollinator
(21). To date, the few studies on the importance
of pollinator richness for crop pollination have
revealed mixed results (22), the effects of even-
ness on pollination services remain largely un-
known, and the impact of wild insect loss on
fruit set has not been evaluated globally for
We tested four predictions arising from the
assumptions that wild insects effectively polli-
nate a broad range of crops, and that their role
can be replaced by increasing the abundance of
honey bees in agricultural fields: (i) For most
crops, both wild insect and honey bee visitation
enhance pollen deposition on stigmas of flow-
ers; (ii) consequently, for most crops, wild insect
and honey bee visitation both improve fruit set;
(iii) visitation by wild insects promotes fruit set
only when honey bees visit infrequently (i.e.,
there is a negative interaction effect between
wild insect visitation and honey bee visitation);
and (iv) pollinator assemblages with more spe-
cies benefit fruit set only when honey bees visit
infrequently (i.e., there is a negative interaction
effect between richness and honey bee visitation).
To test these predictions, we collected data at
600 fields on all continents, except Antarctica,
for 41 crop systems (Fig. 1). Crops included a
Sede Andina, Universidad Nacional de Río Negro (UNRN)
and Consejo Nacional de Investigaciones Científicas y Técnicas
(CONICET), Mitre 630, CP 8400, San Carlos de Bariloche, Río
Department of Animal Ecology and Trop-
ical Biology, Biocentre, University of Würzburg, Am Hubland,
D-97074 Würzburg, Germany.
Department of Ecology, Evo-
lution and Natural Resources, Rutgers University, New Brunswick,
NJ 08901, USA.
Laboratorio Ecotono, Centro Regional Univer-
sitario Bariloche (CRUB), Universidad Nacional del Comahue and
Instituto de Investigaciones en Biodiversidad y Medioambiente
(INIBIOMA), CP 8400, San Carlos de Bariloche, Río Negro, Ar-
Department of Ecology, Swedish University of Agricul-
tural Sciences, SE-750 07 Uppsala, Sweden.
Sciences, Box 1700, Canberra, ACT, Australia.
Sciences Policy and Management, 130 Mulford Hall, University
of California, Berkeley, CA 94720, USA.
School of Biology,
University of Leeds, Leeds LS2 9JT, UK.
Center, P.O. Box 9517, 2300RA Leiden, Netherlands.
ment of Biological Sciences, University of Calgary, Calgary,
Alberta T2N 1N4, Canada.
Department of Entomology,
Robert H. Smith Faculty of Agriculture, Food and Environ-
ment, Hebrew University of Jerusalem, Rehovot, Israel.
partment of Entomology, Evolution and Natural Resources,
Rutgers University, New Brunswick, NJ 08901, USA.
of Ecology, Ecosystem Functions, Leuphana University, 21335
Department of Environmental Systems
Science, ETH Zürich, 8092 Zürich, Switzerland.
Ecología Regional, Facultad de Ciencias Naturales e IML,
Universidad Nacional de Tucumán, CC 34, CP 4017, Yerba
Buena, Tucumán, Argentina.
Ecological Farming Systems,
Agroscope Reckenholz-Tänikon Research Station ART, Zürich,
Universidade Federal do Ceará, Departamento
de Zootecnia–CCA, Campus Universitário do Pici, Bloco 808.
60.356-000 Fortaleza–CE, Brazil.
Universidade Federal da
Bahia, Departamento de Zoologia, Instituto de Biologia, Rua
Barão de Geremoabo, s/n. 40.170-110 Salvador–BA, Brazil.
New Zealand Institute for Plant and Food Research, Private
Bag 4704, Christchurch, New Zealand.
Department of En-
tomology, Michigan State University, East Lansing, MI 48824,
Agriculture and Agri-Food Canada, Atlantic Food and
Horticulture Research Centre, Kentville, Nova Scotia, Canada.
Development by Design Program, Nature Conservancy, Fort
Collins, CO 80524, USA.
Agroecology, Department of Crop
Sciences, Georg-August-University, Grisebachstr. 6, 37077
School of Biological Sciences and Ecol-
ogy Centre, University of Queensland,Brisbane,QLD,Australia.
Department of Biology, National Center for Research in
Natural Sciences, CRSN-Lwiro, D.S. Bukavu, Sud-Kivu, Demo-
cratic Republic of Congo.
Department of Entomology, Cornell
University, New York State Agricultural Experiment Station, 630
West North Street, Geneva, NY 14456, USA.
School of Agri-
culture, Policy and Development, University of Reading, Read-
ing RG6 6AR, UK.
Department of Physical Geography and
Quaternary Geology, Stockholm University, SE 106 91 Stockholm,
Gund Institute for Ecological Economics, University
of Vermont, Burlington, VT 05401, USA.
Department of Biol-
ogy, Lund University, SE-22362Lund,Sweden.
diversity Research Division, South African National Institute of
Biodiversity, Private Bag X7, Claremont, 7735, South Africa.
Institute of Ecology and Evolution, Community Ecology, Uni-
versity of Bern, 3012 Bern, Switzerland.
University of Koblenz-Landau, Fortstrasse7, D-76829 Landau,
Institute of Environmental Sciences, Jagiellonian
University, ul. Gronostajowa 7, Kraków 30-387, Poland.
partment of Forest Entomology, Forestry and Forest Products
Research Institute, 1 Matsunosato, Tsukuba, Ibaraki 305-8687,
Departamento de Ciencias Químico-Biológicas, Uni-
versidad de las Américas Puebla, Cholula, Puebla, Mexico.
Department of Entomology, University of California, Davis, CA
*Corresponding author. E-mail: firstname.lastname@example.org
29 MARCH 2013 VOL 339 SCIENCE www.sciencemag.org1608
on April 1, 2013www.sciencemag.orgDownloaded from
wide array of animal-pollinated, annual and
perennial fruit, seed, nut, and stimulant crops;
predominantly wind-pollinated crops were not
considered (fig. S2 and table S1). The sampled
fields were subject to a diversity of agricultural
practices, including extensive monocultures and
small or diversified systems (fig. S2 and table
S1), fields stocked with low to high density of
honey bees (Fig. 1 and table S2), and fields with
low to high abundance and diversity of wild
insects (fig. S3 and table S2). For each field, we
measured flower visitation per unit of time (here-
after “visitation”) for each insect species, from
which we estimated species richness and even-
ness (23). We quantified pollen deposition for
14 systems as the number of pollen grains per
stigma, and fruit set (fig. S1) for 32 systems as
the percentage of flowers setting mature fruits
or seeds. Spatial or temporal variation of pollen
deposition and fruit set were measured as the
coefficient of variation (CV) over sample points
or days within each field (10). The multilevel
data provided by fields within systems were
analyzed with general linear mixed-effects mod-
els that included crop system as a random effect,
and wild insect visitation, honey bee visitation,
evenness, richness, and all their interactions as
fixed effects. Best-fitting models were selected
on the basis of the Akaike information criterion
In agreement with the first prediction, crops
in fields with more flower visits received more
pollen on stigmas, with an overall 74% stronger
influence of visitation by honey bees than by
wild insects (Fig. 2A and table S3). Honey bee
visitation significantly increased pollen deposi-
tion (i.e., confidence intervals for individual re-
gression coefficients, b
, did not include zero)
in 7 of 10 crop systems, and wild insects in 10
of 13 systems (fig. S4). Correspondingly, in-
creased wild insect and honey bee visitation
reduced variation in pollen deposition among
samples (fig. S5).
Contrary to the second prediction, fruit set
increased significantly with wild insect visita-
tion in all crop systems, but with honey bee
visitation in only 14% of the systems (Fig. 2B).
In addition, fruit set increased twice as strongly
with visitation by wild insects as with visitation
by honey bees (Fig. 2A). These partial regres-
sion coefficients did not differ simply because
of unequal abundance, nor because of dispar-
ate variation in visitation between wild insects
and honey bees. In crop systems visited by both
honey bees and wild insects, honey bees ac-
counted for half of the visits to crop flowers
[mean = 51%; 95% confidence interval (CI) = 40
to 62%], and among-field CVs for visitation by
honey bees (mean = 73%; 95% CI = 57 to 88%)
and by wild insects (mean = 79%; 95% CI = 62
to 96%) were equivalent. Furthermore, wild in-
sect visitation had stronger effects than honey
bee visitation, regardless of whether honey bees
were managed or feral (fig. S6) and, compar-
ing across systems, even where only wild insects
or honey bees occurred (Fig. 2B). Wild insect
visitation alone predicted fruit set better than did
honey bee visitation alone (D
= 16; table S4,
model F versus model M). Correspondingly,
the CV of fruit set decreased with wild insect
visitation but varied independently of honey bee
visitation (fig. S5).
Pollinator visitation affected fruit set less
strongly than did pollen deposition on stigmas
(compare regression coefficients in Fig. 2A). This
contrast likely arose from pollen excess, filtering
of pollen tubes by post pollination processes,
and/or seed abortion (11,24), and so reflects pol-
lination quality, in part. Intriguingly, the differ-
ence in coefficients between pollen deposition
and fruit set for honey bees greatly exceeded
that for wild insects (Fig. 2A); this finding in-
dicates that wild insects provide better-quality
pollination, such as greater cross-pollination
(14,16,17,19). These results occurred regardless
of which crop systems were selected (fig. S7),
sample size (fig. S8), the relative frequency of
honey bees in the pollinator assemblage (domi-
nance) among systems, the pollinator depen-
dence of crops, or whether the crop species were
herbaceous or woody, or native or exotic (fig.
S9). Poor-quality pollination could arise if for-
aging behavior on focal resources typical of honey
bees (16,17) causes pollen transfer between
flowers of the same plant individual or the same
cultivar within a field, thereby limiting cross-
pollination and increasing the incidence of self-
pollen interference and inbreeding depression (24).
The smaller difference in coefficients between
pollen deposition and fruit set for wild insects,
and the stronger effect of wild insect visitation
on fruit set, suggest that management to promote
diverse wild insects has great potential to im-
prove the global yield of animal-pollinated crops.
The third prediction was also not supported.
Fruit set consistently increased with visitation
by wild insects, even where honey bees visited
frequently (i.e., no statistical interaction; Fig. 2,
A and C). In particular, the best-fitting model
(lowest AIC) for fruit set included additive ef-
fects of visitation by both wild insects and hon-
ey bees (table S4, model P), which suggests that
managed honey bees supplement the polli-
nation service of wild insects but cannot re-
place it. Overall, visitations by wild insects and
honey bees were not correlated among fields
(fig. S10), providing no evidence either for
Fig. 1. Relative visitation by honey bees and wild insects to flowers of 41 crop systems on six continents. Honey bees occur as domesticated colonies in
transportable hives worldwide, as a native species in Europe (rarely) and Africa, or as feral populations in all other continents except Antarctica.
www.sciencemag.org SCIENCE VOL 339 29 MARCH 2013 1609
on April 1, 2013www.sciencemag.orgDownloaded from
competition for the resources obtained from crop
flowers (pollen, nectar) or for density compensa-
tion (13) between wild insects and honey bees
at the field scale. Even if honey bees displace
wild insects (or vice versa) at the flower scale
(16,17), this is unlikely to scale up to the field,
as indicated by our data, if mass-flowering crops
provide floral resources in excess of what can
be exploited by local pollinator populations.
Therefore, insect pollinators appear not to be
limited by crop floral resources, but crop yield
was commonly pollen-limited, as crops set more
fruit in fields with more visitation by pollinators
Contrary to the fourth prediction, fruit set
increased with flower-visitor richness indepen-
dently of honey bee visitation (fig. S11). Corre-
spondingly, the CVs of fruit set decreased with
richness; in contrast, evenness did not affect the
mean or CV of fruit set (figs. S12 and S13). Vis-
itation by wild insects increased strongly with
richness (Fig. 3) and improved model fit (lower
AIC), even when richness was included in the
model (table S4, model B versus model G).
However, richness did not enhance model fit
when added to a model with wild insect visi-
tation (table S4, model F versus model G), which
suggests that the effects of richness on fruit set
reflect increased wild insect visitation (i.e., co-
linear effects; fig. S13). Like wild insect visita-
tion (fig. S10), richness did not correlate with
honey bee visitation (table S5). Previous studies
have shown that agricultural intensification re-
duces both species richness of pollinator assem-
blages and wild insect visitation (4,5,13,19).
Our results for multiple crop systems further
demonstrate that fields with fewer pollinator
species experience less visitation by wild insects
and reduced fruit set, independent of species
evenness or honey bee visitation. Globally, wild
insect visitation is an indicator of both species
richness and pollination services, and its measure-
ment can be standardized easily and inexpen-
sively among observers in field samples (25).
Large, active colonies of honey bees provide
abundant pollinators that can be moved as needed,
hence their appeal for pollination management
in most animal-pollinated crops (6–8,26). By
Fig. 3. Globally, rate of visitation to crop flowers by wild insects increases with
flower-visitor richness. (A) The line is the overall regression, and each point is a
field in a crop system. (B)Slopes(b
T95% CI) represent the effect of richness
on wild insect visitation for individual crop systems. Data from individual crop
systems were standardized by zscores prior to analysis (after log-transformation
for visitation), permitting direct comparison of regression coefficients.
Fig. 2. Wild insect visitation to crop flowers enhances reproduction in all crops examined
(regression coefficient b
> 0), whereas honey bee visitation has weaker effects overall. (A)
Overall partial regression coefficients (b
T95% CI) for the direct and interacting effects of
visitation by wild insects and honey bees on pollen deposition or fruit set (models R and Q
in tables S3 and S4, respectively). (B)Slopes(b
T95% CI) represent the effects of visitation
by wild insects or honey bees on fruit set for individual crop systems. Cases at the right are
systems in which only wild insects or only honey bees were present. Data from individual
crop systems were standardized by zscores prior to analysis, permitting comparison of
regression coefficients in all panels. Letters after crop names indicate different regions
(table S1); for example, Mango_A and Mango_B are located in South Africa and Brazil,
respectively. (C) Given the absence of interaction between the effects of visitation by wild
insects and honey bees, maximum fruit set is achieved with high visitation by both wild
insects and honey bees (upper right area of graph). The plane in orange is the overall
regression (model P in table S4; the inclination of the surface in the yand xdirections
reflects the b
for visitation of wild insects and honey bees, respectively), and each point
is a field in a crop system (fruit set increases from cyan to dark blue).
29 MARCH 2013 VOL 339 SCIENCE www.sciencemag.org
on April 1, 2013www.sciencemag.orgDownloaded from
comparison, methods for maintaining diverse
wild insects for crop pollination are less devel-
oped, and research on such pollination services
is more recent (3,16,17,20,26,27) (table S1).
Although honey bees are generally viewed as a
substitute for wild pollinators (3,6–8), our re-
sults show that they neither maximize pollination
nor fully replace the contributions of diverse wild
insect assemblages to fruit set for a broad range
of crops and agricultural practices on all conti-
nents with farmland. These conclusions hold even
for crops stocked routinely with high densities
of honey bees for pollination, such as almond,
blueberry, and watermelon (Fig. 2 and table S2).
Dependence on a single species for crop pollina-
tion also carries the risks associated with predator,
parasite, and pathogen development (4,20,28).
Our results support integrated management
policies (29) that include pollination by wild in-
sects as ecosystem service providers, along with
managed species—such as honey bees, bumble
bees (Bombus spp.), leafcutter bees (Megachile
spp.), mason bees (Osmia spp.), and stingless
bees (Meliponini)—as agricultural inputs, where
they are not invasive species. Such policies should
include conservation or restoration of natural or
seminatural areas within croplands, promotion
of land-use heterogeneity (patchiness), addition
of diverse floral and nesting resources, and con-
sideration of pollinator safety as it relates to pes-
ticide application (3,16,17,20,27). Some of
these recommendations entail financial and op-
portunity costs, but the benefits of implementing
them include mitigation against soil erosion as
well as improvements in pest control, nutrient
cycling, and water-use efficiency (30). Without
such changes, the ongoing loss of wild insects
(4,5) is destined to compromise agricultural
References and Notes
1. L. Roberts, Science 333, 540 (2011).
2. D. Tilman et al., Science 292, 281 (2001).
3. A.-M. Klein et al., Proc. R. Soc. London Ser. B 274, 303
4. S. G. Potts et al., Trends Ecol. Evol. 25, 345 (2010).
5. L. A. Garibaldi et al., Ecol. Lett. 14, 1062 (2011).
6. R. A. Morse, Trends Ecol. Evol. 6, 337 (1991).
7. R. M. Goodwin, H. M. Cox, M. A. Taylor, L. J. Evans,
H. M. McBrydie, New Zeal. J. Crop Hort. 39, 7 (2011).
8. R. R. Rucker, W. N. Thurman, M. Burgett, Am. J.
Agric. Econ. 94, 956 (2012).
9. M. A. Aizen, L. D. Harder, Curr. Biol. 19, 915 (2009).
10. L. A. Garibaldi, M. A. Aizen, A.-M. Klein, S. A. Cunningham,
L. D. Harder, Proc. Natl. Acad. Sci. U.S.A. 108, 5909 (2011).
11. N. P. Chacoff, M. A. Aizen, V. Aschero, Proc. R. Soc.
London Ser. B 275, 907 (2008).
12. W. F. Morris, D. P. Vázquez, N. P. Chacoff, Ecology 91,
13. R. Winfree, C. Kremen, Proc. R. Soc. London Ser. B 276,
14. P. Hoehn, T. Tscharntke, J. M. Tylianakis, I. Steffan-Dewenter,
Proc. R. Soc. London Ser. B 275, 2283 (2008).
15. N. Blüthgen, A.-M. Klein, Basic Appl. Ecol. 12, 282 (2011).
16. L. G. Carvalheiro et al., Ecol. Lett. 14, 251 (2011).
17. S. S. Greenleaf, C. Kremen, Proc. Natl. Acad. Sci. U.S.A.
103, 13890 (2006).
18. B. J. Cardinale et al., Nature 443, 989 (2006).
19. A.-M. Klein, C. Müller, P. Hoehn, C. Kremen, in
Biodiversity, Ecosystem Functioning, and Human
Wellbeing: An Ecological and Economic Perspective,
D. E. Bunker, A. Hector, M. Loreau, C. Perrings,
S. Naeem, Eds. (Oxford Univ. Press, Oxford, 2009),
20. T. Tscharntke, A.-M. Klein, A. Kruess, I. Steffan-Dewenter,
C. Thies, Ecol. Lett. 8, 857 (2005).
21. H. Hillebrand, D. M. Bennett, M. W. Cadotte, Ecology 89,
22. B. J. Cardinale et al., Nature 486, 59 (2012).
23. See supplementary materials on Science Online.
24. M. A. Aizen, L. D. Harder, Ecology 88, 271 (2007).
25. C. Kremen, K. S. Ullman, R. W. Thorp, Conserv. Biol. 25,
26. K. S. Delaplane, D. F. Mayer, Crop Pollination by Bees
(CABI, New York, 2000).
27. C. Kremen et al., Ecol. Lett. 10, 299 (2007).
28. R. Winfree, N. M. Williams, J. Dushoff, C. Kremen,
Ecol. Lett. 10, 1105 (2007).
29. A. Aebi et al., Trends Ecol. Evol. 27, 142 (2012).
30. S. D. Wratten, M. Gillespie, A. Decourtye, E. Mader,
N. Desneux, Agric. Ecosyst. Environ. 159, 112 (2012).
Acknowledgments: Funding acknowledgments and author
contributions are listed in the supplementary materials.
The data used in the primary analyses are available in the
supplementary materials, including tables S1 and S2.
Materials and Methods
Figs. S1 to S13
Tables S1 to S5
14 September 2012; accepted 5 February 2013
Published online 28 February 2013;
over 120 Years: Loss of Species,
Co-Occurrence, and Function
Laura A. Burkle,
*John C. Marlin,
Tiffany M. Knight
Using historic data sets, we quantified the degree to which global change over 120 years
disrupted plant-pollinator interactions in a temperate forest understory community in Illinois,
USA. We found degradation of interaction network structure and function and extirpation of
50% of bee species. Network changes can be attributed to shifts in forb and bee phenologies
resulting in temporal mismatches, nonrandom species extinctions, and loss of spatial
co-occurrences between extant species in modified landscapes. Quantity and quality of pollination
services have declined through time. The historic network showed flexibility in response to
disturbance; however, our data suggest that networks will be less resilient to future changes.
Almost 90% of flowering plant species,
including many important crop species
(1), rely on animal pollinators (2). Plant-
pollinator interaction networks may be particu-
larly susceptible to anthropogenic changes, owing
to their sensitivity to the phenology, behavior,
physiology, and relative abundances of multiple
species (3). Alternatively, the overall structure of
plant-pollinator networks might be robust to per-
turbations because of a high degree of nestedness
and redundancy in interactions (4).
Several authors have speculated about how
changes in biodiversity (5) and phenology (6–8)
might translate into changes in the structure (9,10)
and stability (11)ofcomplexinteraction networks.
However, there has been a lack of historical data
on plant-pollinator networks and phenologies for
both plants and insects in the same community.
By using an extensive and unique data set, we
were able to examine changes in plant-pollinator
network structure and phenologies of forbs and
bees across more than a century of anthropogenic
In the late 1800s, Charles Robertson metic-
ulously collected and categorized insect visitors
to plants, as well as plant and insect phenolo-
gies, in natural habitats near Carlinville, Illinois,
USA (12–14). Over the next century, this region
experienced severe habitat alteration, including
conversion of most forests and prairies to agri-
culture, and moderate climatic warming of 2°C in
winter and spring. In 2009 and 2010, we revisited
the area studied by Robertson and re-collected
data on the phenologies and structure of a subset
of this network—26 spring-blooming forest un-
derstory forbs and their 109 pollinating bees
(15). Hence, we could quantify changes in net-
work structure, local bee diversity, and phenol-
ogies of forbs and bees. Further analyses and a
null model determined the degree to which changes
in network structure and bee diversity were at-
tributed to species’traits, phenological mismatches,
and land-use factors that spatially separate inter-
acting species. To examine shifts in the quantity
of pollinator services, we used a second histor-
ical data set from Carlinville collected in the
early 1970s (16), examining the diversity and
visitation rate of bees to the most important floral
resource in this network (Claytonia virginica).
Washington University, Department of Biology, St. Louis, MO
Montana State University, Department of Ecol-
ogy, Bozeman, MT 59717, USA.
University of Illinois, Illinois
Sustainable Technology Center, Champaign, IL 61820, USA.
*Corresponding author. E-mail: email@example.com
www.sciencemag.org SCIENCE VOL 339 29 MARCH 2013 1611
on April 1, 2013www.sciencemag.orgDownloaded from