INSIGHTS | PERSPECTIVES
substitute for honey bees and often provide
superior services (9). Of 20,000 bee species
worldwide, 12% are estimated to contribute
to crop-pollination services (6).
Winfree et al. elegantly disentangle the effects of species dominance from turnover in
how different native bee species contribute
to pollination of three crops (watermelon,
blueberry, and cranberry) in the U.S. mid-Atlantic region. High functional dominance
exists in the system, with just a few species
supplying a large proportion of the pollination overall (10). The authors determine
this by measuring both the typical amount
of pollen that each species delivers on a
single visit and the visit rate of each species
at a given farm site, and then estimating
the contribution of each species to the total
amount of pollen delivered. However, as the
team considered a larger and larger array
of sites, they found that many more species
were required to reach a threshold level of
pollination (25, 50, or 75% of the mean total
pollination per site).
Further, the authors quantified the relative importance of species turnover and
species dominance with a null model that
removed the effect of dominance. In this
model, they effectively apportioned the pollination provided at a site evenly among all
species present. The surprising result was
that the effects of species turnover were,
on average, 14 times more important than
dominance effects for pollination function (at the largest scale of analysis), even
though high levels of dominance occurred
at individual sites and dominant species
were widespread. Thus, achieving the 50%
pollination threshold at a single farm site
required, on average, 5.5 bee species, but 55
species were needed across the entire study
region. At the 75% threshold, most bee
species in the pool of crop-visiting species
would be needed across all sites.
Why are these results so different from
previous studies, including studies of crop
pollination, which concluded that only a
few dominant species are needed to supply
ecosystem functions? A partial answer is
that this is the first study to disentangle the
contrasting effects of species dominance
Equally important is how exactly the
service was measured. Previous studies,
including those by Winfree and co-workers
(6, 10), looked at the relative contributions
of functionally dominant and nondominant
species to ecosystem function without considering the actual amount of pollination
needed by farmers to reach critical pollination thresholds. In the current study, Winfree et al. instead looked at the magnitude
of the service and whether a given threshold (25, 50, or 75%) is achieved at each site
on the basis of the bee-community composition. Critically, at sites where the dominant,
widespread pollinators are low in abundance, almost all or even all pollinator species may be needed (see the figure). At such
sites, relatively rare species provide essential contributions to pollination function.
Species turnover among such sites, then, is
the reason why so many species are needed,
regionally, to provide pollination.
Even though rare species make a small
contribution overall across sites, identifying
their contributions to reaching a threshold
on a farm-by-farm basis shows how important they are. Arguably, understanding how
much pollination a farmer would get at any
point in the landscape is the relevant metric for assessing ecosystem services to real
people in real landscapes (11).
The growing chorus, both from plot-
based experimental studies (3) and now
from a large-scale natural experiment (2),
strongly supports the importance of main-
taining a large amount of biodiversity to
support human well-being sustainably. But
maintaining this biodiversity in agricultural
landscapes, both for pollination services
and for other ecosystem functions and ser-
vices that support crop production, is likely
to require substantial changes in manage-
ment. Specifically, it will require moving
away from monocultures and fencerow-to-
fencerow farming that rely extensively on
external inputs of pesticides and fertilizers,
as well as managed honey bees that may
compete with wild bee species (12), and
toward farms that generate much of the
needed pest and disease control, soil fertil-
ity, and pollination services through crop
and noncrop diversification and “ecological
intensification” (increasing crop produc-
tivity through management practices that
promote the organisms producing ecosys-
tem services, rather than through increased
use of pesticides and fertilizers) (13, 14). For
example, planting diverse crops, flowering
strips, and hedgerows can restore wild pol-
linator populations, enhance species turn-
over, and supply pollination services (15).
Winfree et al.’s study helps to show that
there may be much more alignment than
previously thought between ecosystem-service arguments for biodiversity conservation
and intrinsic-value arguments (conserving
biodiversity for its own sake). Given the key
role of biodiversity for human well-being and
sustainability, it is crucial that human societies better protect and restore biodiversity. j
1. G. S. Daily et al., Issues Ecol. 2, 1 (1997).
2. R.Winfree et al., Science359, 791(2018).
3. D.Tilman, F.Isbell, J.M.Cowles, Annu. Rev. Ecol. Evol. Syst.
45, 471 (2014).
4. P. B. Reich et al., Science 336, 589 (2012).
5. F. Isbell et al., Nature 546, 65 (2017).
6. D. Kleijn etal. , Nat.Commun. 6, 7414 (2015).
7. M. Lohbeck, F. Bongers, M. Martinez-Ramos, L. Poorter,
Ecology 97, 2772 (2016).
8. H. Tallis, J. Lubchenco, Nature 515, 27 (2014).
9. L. A. Garibaldi et al., Science 339, 1608 (2013).
10. R. Winfree,J. W.Fox,N.M. Williams, J.R.Reilly,D.P.
Cariveau, Ecol. Lett. 18, 626 (2015).
11. C. Kremen, R. S. Ostfeld, Front. Ecol. Environ. 3, 540
12. J. Geldmann, J. P. González-Varo, Science 359, 392 (2018).
13. R. Bommarco, D. Kleijn, S. G. Potts, Trends Ecol. Evol. 28,
14. C.Kremen, A.Miles, Ecol.Soc. 17,40(2012).
15. A.Kovács-Hostyánszki et al., Ecol. Lett. 20,673(2017).
Site 2 Site 3 Site 4 Site 1
Number of species
742 16 FEBRUARY 2018 • VOL 359 ISSUE 6377
Bee diversity needed for pollination
Pollinating species vary from site to site; numbers of individuals indicate abundances at the site for each species type.
Dominant species contribute most
to pollination function at sites 1
and 2, and only one or two species,
respectively, are needed to surpass
the threshold required for full
pollination. Dominant species occur
at all sites, but because of their low
abundance at sites 3 and 4, most
species are needed for pollination
function. Species turnover between
such sites means that most species
in the species pool are needed to
supply pollination function across
the entire array of sites.