POLLINATORS
Bumble bees damage plant leaves and accelerate
flower production when pollen is scarce
Foteini G. Pashalidou1,2, Harriet Lambert^1 , Thomas Peybernes^1 ,
Mark C. Mescher^1 †, Consuelo M. De Moraes^1 †
Maintaining phenological synchrony with flowers is a key ecological challenge for pollinators that may be
exacerbated by ongoing environmental change. Here, we show that bumble bee workers facing pollen
scarcity damage leaves of flowerless plants and thereby accelerate flower production. Laboratory
studies revealed that leaf-damaging behavior is strongly influenced by pollen availability and that
bee-damaged plants flower significantly earlier than undamaged or mechanically damaged controls.
Subsequent outdoor experiments showed that the intensity of damage inflicted varies with local flower
availability; furthermore, workers from wild colonies of two additional bumble bee species were also
observed to damage plant leaves. These findings elucidate a feature of bumble bee worker behavior that
can influence the local availability of floral resources.
O
ngoing environmental change is exac-
erbating phenological mismatches be-
tween plants and pollinators ( 1 – 9 ), with
the greatest detrimental effects on pol-
linator populations occurring in early
spring ( 10 , 11 ). Such mismatches may be par-
ticularly challenging for bumble bees, which
establish new colonies each spring ( 7 , 12 – 14 ).
One recent study found that the timely avail-
ability of floral resources within 1 km of found-
ing colonies strongly influenced colony survival
and performance ( 9 ), while another showed
that resource shortfalls during a critical pe-
riod of early colony development have pro-
found effects on later reproductive success
( 14 ). The fitness implications of phenological
mismatches presumably favor adaptations that
ensure synchrony ( 7 , 13 – 20 ), and previous re-
search on social bees has documented changes
in life history and social organization associated
with variation in the seasonal availability of
floral resources ( 13 , 15 – 22 ). This study explores
whether bumble bee workers may themselves
influence local flower availability by damag-
ing the leaves of flowerless plants and thereby
accelerating flower production.
Initial behavioral observations with four plant
species (Brassica oleracea,B. nigra,Solanum
elaeagnifolium,andS. melongena) revealed
that bumble bee (Bombus terrestris)workers
use their proboscises and mandibles to cut
distinctively shaped holes in plant leaves, with
each damage event taking only a few seconds
(Fig. 1 and movie S1). However, we saw no
clear evidence that bees were actively feeding
on leaves or transporting leaf material back to
the hive. We therefore hypothesized that dam-
age inflicted on plant leaves might influence
subsequent flower production. Although stress
is known to influence flowering time in many
plant species ( 23 ), most previous work has
focused on abiotic stressors, and few studies
have explored effects of leaf damage inflicted
by insects ( 23 , 24 ).
To determine whether damage inflicted by
B. terrestrisworkers influences flowering, we
conducted experiments with two distantly re-
lated plants,S. lycopersicumandB. nigra,and
compared the flowering time of bee-damaged
plants to that of undamaged plants and plants
subjected to similar amounts of damage in-
flicted mechanically. Plants given bee-damage
treatments wereindividually placed in mesh
cages with pollen-deprivedB. terrestriscolo-
nies and removed when worker bees had made
5 to 10 leaf holes. Each damaged plant was
then paired with a plant given the mechanical-
damage treatment, in which we replicated the
pattern of bee-inflicted damage as closely as
possible using metal forceps and a razor (see
supplementary materials for further method-
ological details). These experiments revealed
highly significant treatment effects on flower-
ing time in both species, with bee-damaged
plants flowering earlier than those that under-
went the other treatments (Fig. 2). Further-
more, the acceleration of flowering elicited
by bee-inflicted damage was substantial: In
S. lycopersicum, the average flowering time
of bee-damaged plants was 30 days earlier
than that of undamaged plants and 25 days
earlier than that of mechanically damaged
plants. InB. nigra, the same comparisons
yielded differences of 16 and 8 days, respec-
tively (Fig. 2). Pairwise treatment comparisons
[via a generalized linear model (GLM)] con-
firmed highly significant effects of bee damage
compared with other treatments for both plant
species (fig. S1). The mean flowering time of
mechanically damaged plants was interme-
diate between bee-damaged and undamagedtreatments in both species but differed sig-
nificantly from that of undamaged plants only
forB. nigra(fig. S1).
Initial observations of laboratory colonies
suggested that damaging behavior was in-
fluenced by the availability of pollen, the only
food given to bumble bee larvae and an im-
portant protein source for workers ( 25 ). To
test this hypothesis, we conducted a labora-
tory assay in whichB. terrestrismicrocolonies
were either given abundant pollen resources
(within the hive) or deprived of pollen. After
subjecting colonies to these treatments for
several days (Fig. 3A), we presented them with
flowerlessB. nigraplants within mesh enclo-
sures then replaced plants daily and quantified
the proportion of damaged leaves. Trials with
paired pollen-satiated and pollen-deprived co-
lonies were conducted in parallel, and the
treatment for each colony was then reversed
(Fig. 3A). The results of this experiment re-
vealed that rates of damaging behavior were
markedly higher under pollen limitation re-
gardless of the order in which treatments were
presented(Fig.3B).Thesefindingsarere-
inforced by further observations showing
that workers from pollen-satiated colonies
consistently inflict only minor amounts of leaf
damage (table S2), whereas higher levels of
damaging behavior can consistently be eli-
cited by depriving microcolonies of pollen for
~3 days, as in our flowering-time experiments.
The laboratory studies described above dem-
onstrate that the leaf-damaging behavior of
B. terrestrisworkers accelerates flowering and
is itself influenced by the availability of floral
resources. However, these results do not exclude
the possibility that the observed behavior might
be an artifact of our experimental conditions, in
which bees were confined in cages that pre-
vented them from going farther afield to forage
for floral resources. To address this possibility,
we next conducted a series of seminatural out-
door experiments by establishing experimental
B. terrestriscolonies and plant patches on
rooftops at the Zentrum campus of ETH Zürich
(Zürich, Switzerland) in 2018 and 2019.
The primary goal of our 2018 study was to
determine whether bees would damage flow-
erless plants near the hive when they had the
option of foraging farther afield. A secondary
goal was to test the prediction that the fre-
quency of damaging behavior would decline
over time as floral resources became more
abundant in the surrounding environment.
This study was carried out in two phases, the
first beginning on 26 March 2018, slightly
earlier than the time at which wild bumble
beequeens would likely be terminating dia-
pause ( 19 ), and continuing to 25 May. Phase 1
thus extended over a period during which
surrounding flower resources increased from
initially low levels ( 26 , 27 )(fig.S2).Duringthis
phase, we positioned aB. terrestrismicrocolonyRESEARCH
Pashalidouet al.,Science 368 , 881–884 (2020) 22 May 2020 1of4
(^1) Department of Environmental Systems Sciences, ETH Zürich,
8092 Zürich, Switzerland.^2 UMR Agronomie, INRA, AgroParisTech,
Universite Paris-Saclay, 78850 Thiverval- Grignon, France.
*These authors contributed equally to this work.
†Corresponding author. Email: [email protected]
(C.M.D.M.); [email protected] (M.C.M.)