is also exhibited by wild bees. Indeed, some
B. terrestrisworkers observed damaging plants
mayalsohavecomefromwildcolonies,al-
thoughthesecouldnotbereliablydistin-
guished from bees from our experimental
colonies. It is also notable that only bumble
bees were observed to land on flowerless plants
within our experimental patches, although other
bees, including honey bees (Apis mellifera)and
common furry bees (Lasioglossum calceatum),were frequent visitors to the nearby patch of
plants in flower (fig. S3).
In 2019, we conducted a second outdoor
study that differed fromthe 2018 study in sev-
eral respects. This second study used youngPashalidouet al.,Science 368 , 881–884 (2020) 22 May 2020 3of4
Fig. 3. Leaf damage by pollen-satiated and pollen-
deprived bumble bees.(A) Experimental setup:
Two colonies were used in parallel for each trial,
one of which was initially“pollen satiated”(PS) and then
switched to“pollen deprived”(PD), while the other
experienced the same treatments in the reverse order.
Asterisks indicate days (1 and 8) when hives were
weighed and new diet treatments were implemented.
Adjustment (“A”) periods (days 2 to 4 and days 9 to 11)
allowed colonies to acclimate to treatments. (B) Bees
from pollen-deprived colonies inflicted higher levels
of damage on plant leaves [Fvalue (F) = 258, df = 32,
R^2 = 0.8801,P< 0.001, GLM], regardless of the order of
the treatment. Additionally, the observed effects did
not differ significantly among colonies (F= 258, df = 32,
P= 0.732, GLM). Boxes represent the first to third
quartile of the interquartile range, horizontal lines within
boxes are medians, and the whiskers are the minimum
and maximum values. Six hives, three replicates,
and 72 plants. ***P≤0.001; ns, not significant.
*** ***ns020406080Pollen Satiated−Pollen Deprived Pollen Deprived−Pollen Satiated
Diet Plan% damage onB.nigraplantsABPS-PDPD-PSDay 1 26543 7891011121314Pollen satiated (PS)Pollen deprived (PD)*Treatment implemented B. nigra***AAAAAAAAAAAAAAdjustment dayFig. 4. Daily measurements of damage (total
number of new leaf holes) byB. terrestrisworkers
in two outdoor experiments.(A) 2018 experiment
with queenless microcolonies. In phase 1, only
flowerless plants were present locally. In phase 2,
100 plants in flower were placed adjacent to the
focal patch of flowerless plants. Colors indicate
individual bee colonies. Damage by bees was signifi-
cantly higher in phase 1 [generalized linear mixed
model fit by maximum likelihood, term = phase, df = 1,
F= 712.70,P<0.001, Akaike information criterion
(AIC) 1660.9]. Dots indicate days on which weather
prevented data collection. The black line represents a
7-day centered moving average of new bee holes
per day. Six flowerless plant species, 36 flowerless plants
per bee colony, and five bee colonies (180 flowerless
plants in total). (B) 2019 experiment with queenright
colonies. Colonies in both treatments were placed
adjacent to focal patches of flowerless plants. Colonies
in the flowering treatment also had access to
30 plants in flower, placed adjacent to the focal patch,
as well as a rooftop garden planted with wildflowers.
On 2 days, indicated by triangles, all colonies were
given access to sugar solution for 24 hours to mitigate
effects of adverse weather conditions. Damage by bees
was significantly higher in the flowerless treatment
(GLM fit by maximum likelihood, term = treatment, df = 1,F= 5217.7,P< 0.001, AIC 2049.3). Damage levels on roof 2 increased significantly in the month after the wildflower
garden was mown (df = 30,P≤0.001, AIC 2960.4, GLM). Seven flowerless plant species, 600 flowerless plants per treatment, and eight bee colonies per treatment.
Phase 1 Phase 2(^0) ** ***
50
100
150
200
Mar 26 Apr 2Apr 9 Apr 16 Apr 23 Apr 30 May 7 May 14 May 21 May 28 Jun 4 Jun 11 Jun 18 Jun 25Jul 2 Jul 9 Jul 16Jul 23
Date
New holes
1
2
3
4
5
(^0) *
250
500
750
1000
Jun 3 Jun 10 Jun 17 Jun 24 Jul 1 Jul 8
Date
New holes
Treatment
nonflowering
flowering
Hive
A
B
Flowering plants added
Rooftop garden mown
2018 queenless colonies
2019 queenright colonies (8 per treatment)
RESEARCH | REPORT