UC Berkeley
UC Berkeley Previously Published Works
Title
Bee Preference for Native versus Exotic Plants in Restored Agricultural Hedgerows
Permalink
https://escholarship.org/uc/item/2df1b5ss
Journal
Restoration Ecology, 21(1)
ISSN
1061-2971
Authors
Morandin, LA
Kremen, C
Publication Date
2013
DOI
10.1111/j.1526-100X.2012.00876.x
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
RESEARCH ARTICLE
Bee Preference for Native versus Exotic Plants
in Restored Agricultural Hedgerows
Lora A. Morandin
1,2
and Claire Kremen
1
Abstract
Habitat restoration to promote wild pollinator populations
is becoming increasingly common in agricultural lands.
Yet, little is known about how wild bees, globally the most
important wild pollinators, use resources in restored habi-
tats. We compared bee use of native and exotic plants
in two types of restored native plant hedgerows: mature
hedgerows (>10 years from establishment) designed for
natural enemy enhancement and new hedgerows (≤2 years
from establishment) designed to enhance bee populations.
Bees were collected from flowers using timed aerial net-
ting and flowering plant cover was estimated by species
using cover classes. At mature hedgerow sites, wild bee
abundance, richness, and diversity were greater on native
plants than exotic plants. At new sites, where native plants
were small and had limited floral display, abundance of
bees was greater on native plants than exotic plants; but,
controlling for floral cover, there was no difference in bee
diversity and richness between the two plant types. At both
mature and new hedgerows, wild bees preferred to forage
from native plants than exotic plants. Honey bees, which
were from managed colonies, also preferred native plants
at mature hedgerow sites but exhibited no preference at
new sites. Our study shows that wild bees, and managed
bees in some cases, prefer to forage on native plants in
hedgerows over co-occurring weedy, exotic plants. Semi-
quantitative ranking identified which native plants were
most preferred. Hedgerow restoration with native plants
may help enhance wild bee abundance and diversity, and
maintain honey bee health, in agricultural areas.
Key words: agriculture, Apoidea, ecosystem services,
pollinators.
Introduction
Seventy-five percent of the leading food crops and 35% of
global food production is dependent on pollinators (Klein
et al. 2007). Bees are the primary crop pollinators, with man-
aged honey bees being the most important pollinator globally
(Watanabe 1994). However, recent problems with honey bee
colony health (Neumann & Carreck 2010) and a greater than
300% increase in area devoted to pollinator-dependent crops
in the last 50 years, has made reliance on managed honey bees
a risky proposition (Aizen & Harder 2009). Native pollinators
supply a significant amount of pollination to many agricultural
crops. In areas with large amounts of natural or semi-natural
land, native bees can fully meet pollination requirements of
a crop without the need for managed honey bees (Kremen
et al. 2004; Morandin & Winston 2006; Winfree et al. 2007).
However, intensive agricultural systems often are lacking in
native pollinators (Winfree et al. 2009) and hence native polli-
nation services to crops (Kremen et al. 2002; Klein et al. 2003;
Morandin & Winston 2006; Ricketts et al. 2008).
1
Department of Environmental Science, Policy & Management, University of
California, Berkeley, 130 Mulford Hall, Berkeley, CA 94720, U.S.A.
2
Address correspondence to L. A. Morandin, email lora_morandin@berkeley.edu
© 2012 Society for Ecological Restoration
doi: 10.1111/j.1526-100X.2012.00876.x
While growers have little or no control over amounts of
natural habitat in their region, they are able to implement farm-
scale habitat enhancements. Restoration of weedy field edges
with native shrubs and forbs is a feasible, and increasingly
popular, method for increasing semi-natural land and thus
native pollinator diversity, abundance, and possibly pollination
services in working farmlands (Pywell et al. 2005; Hopwood
2008; Hannon & Sisk 2009). While restoration initiatives to
enhance native pollinators in agricultural lands increasingly
are being promoted through incentive or outreach programs,
very little is known about resource use by pollinators within
such restored areas (Winfree 2010).
Habitat enhancements for wild pollinators generally aim to
enhance the abundance and diversity of floral resources in
order to provide a consistent supply over the flight season
(Vaughan et al. 2007; Menz et al. 2011). Often, non-native
annual plants are recommended for pest-control enhancement
on farm sites, despite the fact that native plants are more
suitable for conservation efforts that intend to also preserve
native plants and the beneficial insects associated with them
(see Tuell et al. 2008 and references therein). Most studies
on bee use of native and exotic plants in disturbed habitats
have found that exotic plants receive more visits than native
plants by native bees because of greater attraction and rewards
(Brown et al. 2002). In a meta-analysis of 40 studies that
examined effect of exotic plants on native plant pollination and
Restoration Ecology 1
Bee Preference in Restored Hedgerows
reproductive success, Morales and Traveset (2009) showed a
negative impact on native pollination and reproductive success
in the presence of exotic plants. Vila et al. (2009) examined
plant–pollinator interactions in invaded (presence of exotic
species) and un-invaded networks and found that pollinators
interacted more with exotic species than they did with native
species in invaded areas, but found no clear effect on whether
the greater visits to exotic plants negatively impacted visits
to native plants. Williams et al. (2011), working in disturbed
agricultural plots in California, found that wild bees did not
exhibit a preference for either native or exotic plants, but uti-
lized both in proportion to their availability. These studies beg
the question as to whether native plantings provide important
resources for wild bees and honey bees in agricultural settings,
where exotics often are numerically dominant; yet, there is lit-
tle information on whether pollinators preferentially choose to
forage from native plant restoration plantings rather than co-
occurring exotic species. Despite increasing funds and effort
to restore agricultural areas with native plants, we know of
no studies that examine pollinator use of native versus exotic
plants in agricultural restorations.
We quantitatively assessed native bee preference for exotic
and native plants in new and mature native plant hedgerows
in an agriculturally intense area of Northern California. We
asked the questions, (1) Do abundance, richness, and diversity
of bee species foraging on exotic versus native plant species
differ? (2) Do native bees preferentially forage on native as
opposed to exotic plants in agricultural hedgerow restoration
sites? (3) Is bee utilization of exotic and native plants different
in mature hedgerows, where native plants dominate, versus
newly planted hedgerows, where exotic plants still dominate?
and, (4) Within the native plant species, are there species
that are preferred or avoided relative to other native plants?
We expected that native and exotic plants would be used in
proportion to their availability in new and mature hedgerow
sites, and would only be an important resource if availability
was high.
Methods
Hedgerow Sites and Plantings
Newly established hedgerow sites were planted between 2007
and 2008 for the purpose of enhancing native bee populations.
They were approximately 350 m long, located adjacent to
natural or man-made sloughs, and contained a mix of native
shrubs and forbs.
Mature hedgerow sites were established in 1996 and were
comprised of a row of perennial shrubs, 305–550 m long, bor-
dered by a stand of perennial grasses. While mature hedgerows
were planted to promote natural enemy and reduce pest insect
abundances (Bugg et al. 1998), the majority of the native flow-
ering plant species also are part of the native planting palette
that was used for the new hedgerow sites in this study.
Composition varied among sites in native plants due to dif-
ferences in service focus (pollination enhancement at new
hedgerows, pest control at mature hedgerows), differences
in species choices made by the land owner, and differential
survival among sites. All new and mature hedgerows con-
tained varying ratios of Ceanothus griseus (California lilac),
Eriogonum fasciculatum (California buckwheat), Rosa califor-
nica (California wild rose), Rhamnus californica (California
coffeeberry), Baccharis pilularis (coyote brush), Sambucus
mexicana (Mexican elderberry), and Heteromeles arbutifolia
(toyon). Most, but not all sites contained Salvia spp. (sage),
Eschscholzia californica (California poppy), Grindelia campo-
rum (gumplant), Achillea millefolium (yarrow), and Atriplex
lentiformis (quail bush).
The primary herbaceous exotic weeds found in both new and
mature hedgerows were Brassica spp. (mustard), Convolvulus
arvensis (field bindweed), Malva parviflora and neglecta
(mallow), and Picris echioides (bristly oxtongue). Hedgerows
were planted adjacent to rotational field crops of approximately
32 ha, that included primarily wheat, processing tomatoes, and
alfalfa, which is typical of crop production in this region.
Study Design
We examined four mature hedgerows and four new hedgerow
sites in Yolo County, CA, in 2009. There was a mini-
mum of 1.5 km distance among hedgerows to ensure spatial
independence.
New hedgerow sites were sampled three times over the
course of the season, from late April until early August, with
approximately 1.5 months between sample rounds. Mature
hedgerow sites were sampled four times during the growing
season, from early May until late July, with approximately
1 month between sample rounds. At each sample round, flow-
ering vegetation was quantified using fifty 1 m
2
quadrats
evenly spaced along the hedgerow. Within the quadrats, all
plants with mature flowers were identified to species and
flower cover of each species was estimated using a Braun-
Blaunquet cover scale. Cover estimates were made by envi-
sioning all flowers of a species within the vertical plane of
the quadrat as a two-dimensional flat surface. To standard-
ize estimates among collectors, all observers used an example
“score-sheet” and were trained collectively to standardize and
score in a consistent manner.
Bee communities were quantified using timed aerial net-
ting. At mature hedgerow sites, bees were collected off of
flowers for 30 minutes and at new hedgerows for 1 hour. The
shorter amount of time at mature hedgerows was due to addi-
tional sampling protocols not reported here. As the collector
slowly walked along the hedgerow, they checked every flower
for the presence of a bee. If a bee was observed touching
the reproductive parts of a flower it was netted and put into
a labeled vial specific to that plant species. The timer was
stopped after the bee was captured in the net, until the collec-
tor was ready to recommence flower observations, so that total
observation time was standardized among collections. Because
the observer examined every flower along the path for the pres-
ence of bees, time viewing each plant species was in proportion
to its floral cover. Collected bees were pinned for later species
identification, which was conducted by Prof. E. R. W. Thorp
2 Restoration Ecology
Bee Preference in Restored Hedgerows
(U. C. Davis, Henry B. Laidlaw Bee Biology Center). The
non-native, but naturalized solitary bee species, Megachile api-
calis and Ceratina dallatoreana were included in our native
bee dataset; the combined total of these two naturalized bee
species made up less than 2% of the total non-Apis bees in
our samples.
In our study region, honey bees (Apis mellifera) p rimar-
ily come from managed colonies. Because high recent colony
losses of honey bees (Neumann & Carreck 2010) may par-
tially be due to nutritional deficiencies that make bees more
susceptible to disease (vanEngelsdorp et al. 2009), enhanc-
ing quality, abundance, diversity, and continuity of foraging
resources may aid managed honey bee colony health. There-
fore, we also examined honey bee use and preference of exotic
and native plants in hedgerows. Because honey bees could
reliably be identified to species in the field, honey bees were
captured, recorded, and then released.
Data Analyses
Each dataset (mature sites and new hedgerow sites) was
analyzed separately due to the slightly different methodologies
employed. Floral cover from each quadrat, at each sample
round, was summed to get a total cover score for that plant type
(exotic or native) at that site for that sample round. Therefore
for new hedgerow sites, there were four sites and three sample
rounds, resulting in 12 records for native plants and 12 for
exotic plants and at mature hedgerow sites there were four
sites and four sample rounds, resulting in 16 records each for
native and exotic plants. Mean floral cover of native and exotic
plants was compared using a mixed model analysis of variance
with a poisson distribution and log link function (GLIMMIX
procedure, SAS 1999). Sample round nested within site was
included as a repeated factor and site as a random factor.
We compared the response variables native bee abundance,
richness, and diversity (Shannon index) on native and exotic
plants (fixed factor) using a mixed model analysis of covari-
ance with a log link function and poisson or negative binomial
distribution (the poisson distribution was tried first and if the
over-dispersion was not corrected than the negative binomial
distribution was used) for bee abundance and richness data,
and a normal distribution for diversity data. We included sam-
ple round nested within site as a repeated factor, site as a ran-
dom factor, and floral cover as the covariate. We first included
the interaction between cover and plant type to test the assump-
tion of homogeneity of regression slopes. If the interaction was
non-significant, we removed the interaction term from the anal-
yses and report on the test of fixed effects for plant type. If the
interaction was significant, we kept the interaction term and
report least squares mean difference and the region(s) of sig-
nificance between native and exotic plants along values of the
covariate, flower cover (Johnson & Neyman 1936; Milliken
& Johnson 2002). Including flower cover as a covariate acts
to standardize for effort as flowers were observed in propor-
tion to their cover. For bee abundance on native and exotic
plants, standardizing for floral cover of natives and exotics
additionally gives a measure of “preference” (Johnson 1980).
That is, controlling for floral cover o f plant type, preference
is established if bees are more abundant on one plant type
than the other (Alldredge & Ratti 1992). For preference anal-
yses, we excluded sites that had less than 10 bees collected
at that site and sample round, because sites with less than 10
bees collected would not have enough replication on exotic
or native plants to give meaningful information on preference.
Therefore, there are different results from “abundance” anal-
yses controlling for cover (where all sites are included) and
“preference” analyses where only a subset of sites are included.
We additionally utilized the reduced dataset to examine
preference by native bees among native plant species semi-
quantitatively at mature hedgerow sites (where there was
enough cover of mature plants to permit meaningful analyses
among species). The data were not amenable to statistical
analyses such as Chi-square because of a large number of
samples under five for each plant species. Therefore, we used
a ranking system at each site calculated by number of bees
found on each plant species at a site and sample round divided
by the flower cover of that plant species. We ranked species
according to this ratio, with higher numbers getting higher
ranks (i.e. relatively more attractive).
Results
Floral Cover
At new hedgerow sites, cover of exotic flowers was greater
than native flowers (F
[1,19]
= 11.81, p = 0.005) with mean
total cover score (SE) of 35.5 (6.8) and 10.1 (6.8), respectively.
At mature hedgerow sites, there was no difference in floral
cover between exotic and native plants (F
[1,26]
= 0.34, p =
0.57) with mean total cover score (SE) of 19.1 (7.3) and 25.8
(7.1), respectively.
Bee Abundance, Richness, and Diversity
Of the 23 species of native bees netted on flowers at the new
hedgerow sites, 7 species were observed only on exotic plants
and 7 species were observed only on native plants. Of the 30
species of native bees netted on flowers at mature hedgerow
sites, 23 bee species were observed only on native plant species
and only 1 bee species was found only on exotic plant species.
There was an interaction between floral cover and native bee
abundance at new hedgerow sites (F
[1,17]
= 8.08, p = 0.01).
We found significantly more native bees on native plants than
exotic plants (t
[17]
=−3.32, p = 0.004; Fig. 1) and p<0.05
for all floral cover values greater than 15. At new hedgerow
sites, there was no difference in native bee species richness and
diversity between native and exotic plants (richness: F
[1,18]
=
0.83, p = 0.37, diversity: F
[1,18]
= 0.17, p = 0.68; Fig. 2).
At mature hedgerow sites, there was greater abundance
(Fig. 1), richness, and diversity (Fig. 2) of native bees on
native plants than exotic plants (abundance: F
[1,25]
= 19.22,
p = 0.0002, richness: F
[1,25]
= 13.07, p = 0.001, diversity:
F
[1,25]
= 10.00, p = 0.004). Honey bee abundance was the
same on native and exotic plants at new hedgerow sites
Restoration Ecology 3
Bee Preference in Restored Hedgerows
Figure 1. Mean number of native bees on exotic versus native plants
from each site at each sample round. *Above bars indicates the response
variable is different between native and exotic plants at p<0.05.
Figure 2. Mean native bee species richness and diversity (Shannon’s
diversity index) from each site and sample round at new hedgerow sites
and mature hedgerow sites. * Above bars indicates the response variable
is different between native and exotic plants at p<0.05.
(F
[1,17]
= 0.38, p = 0.55). At mature sites, there was an inter-
action between floral cover and plant type on honey bee abun-
dance (F
[1,24]
= 105.2, p<0.0001). Closer inspection of the
region of significance revealed that honey bee abundance was
greater on native plants at all cover levels and that the mag-
nitude of difference between honey bee abundance on native
versus exotic plants increased with increasing cover score.
Preference
At new hedgerow sites, regression slopes were significantly
different for native bee abundance on native versus exotic
plants (floral cover by plant type interaction: F
[1,11]
= 10.64,
p = 0.008). Mean bee abundance on native plants was greater
than abundance on exotic plants (t
[11]
=−5.44, p = 0.0002).
At floral cover scores greater than 10, native bees showed a
preference for native plants. As floral cover score increased,
the difference between bee abundance on native and exotic
plants increased (Fig. 3). Native bees preferred native plants
over exotic plants at mature hedgerow sites at all cover
amounts (F
[1,13]
= 39.08, p<0.0001) (Fig. 3).
Honey bees exhibited no preference for exotic or native
plants in new hedgerow sites (F
[1,16]
= 0.01, p = 0.93). In
mature hedgerow sites, however, honey bees preferentially
selected native plant species (F
[1,12]
= 102.81, p<0.0001)
(Fig. 4).
Native Species Ranking
Semi-quantitative ranking of preference among native plant
species showed that when present, Eriogonum fasciculatum
and Salvia spp. were the most preferred native plant species
(Table 1). Other species within the top preferred native plants
for bee forage were Eschscholzia californica, Rhamnus califor-
nica,andGrindelia camporum. Heteromeles arbutifolia, Achil-
lea millefolium,andAtriplex lentiformis had mixed results in
terms of preference. Sambucus mexicana and Rosa califor-
nica were consistently less preferred by native bees when
other native species were available. However, large numbers
of syrphid flies (Family Syrphidae), which also can be impor-
tant native pollinators of agricultural crops (Jauker & Wolters
2008), were caught on elderberry (L. Morandin & C. Kremen,
unpublished data). Of the native bee species that were repre-
sented by greater than two individuals (17 species), four bee
species were found on only one species of native plant in this
subset of samples.
Discussion
These data indicate that native bees prefer to forage on native
plants in both new and mature hedgerow sites. In addition,
we found that bee abundance was greater on native plants
in both new and mature hedgerows and bee richness and
diversity were greater on native plants than exotic plants
in mature hedgerows. Strikingly, 77% of bee species at
mature hedgerows were only found on native plant species.
These results indicate that in intense agricultural landscapes
native plants are important for sustaining both abundance and
diversity of native bee species.
Our finding of greater preference by native bees for native
plants is contrary to some bee preference studies in natural or
semi-natural areas that found greater or equal preference for
exotic plants (Vila et al. 2009; Williams et al. 2011). At our
new hedgerow sites, native shrubs were less than 3 years, and
cover of exotic plants was greater at most sites; yet, we found
that native bees preferentially chose native plants even when
relative abundance of native plants was low.
Despite preference for native plants, at new hedgerow sites
where native plants were sparse, substantial proportions of
native bee (45%) and honey bee (66%) collections were on
exotic plants. In contrast, at mature hedgerow sites, where
native flowers were as abundant as exotic flowers, exotic plants
4 Restoration Ecology
