UC Berkeley
UC Berkeley Previously Published Works
Title
Hedgerow restoration promotes pollinator populations and exports native bees to
adjacent fields.
Permalink
https://escholarship.org/uc/item/4tr1z08r
Journal
Ecological applications : a publication of the Ecological Society of America, 23(4)
ISSN
1051-0761
Authors
Morandin, Lora A
Kremen, Claire
Publication Date
2013-06-01
DOI
10.1890/12-1051.1
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Ecological Applications, 23(4), 2013, pp. 829–839
Ó 2013 by the Ecological Society of America
Hedgerow restoration promotes pollinator populations
and exports native bees to adjacent fields
LORA A. MORANDIN
1
AND CLAIRE KREMEN
Department of Environmental Science, Policy and Management, University of California, Berkeley, 130 Mulford Hall,
Berkeley, California 94720 USA
Abstract. In intensive agricultural landscapes, restoration within farms could enhance
biodiversity and ecosystem services such as pollination by native pollinators. Although
governments and conservation groups are promoting small-scale restoration on working
farms, there are few studies that assess whether these practices enhance pollinator communities
in restored areas. Further, there is no information on whether floral enhancements will deplete
pollinators in adjacent fields by concentrating ambient populations or whether they result in a
net increase in abundance in adjacent farm fields. We investigated whether field edges restored
with native perennial plants in California’s Central Valley agricultural region increased floral
abundance and potential bee nesting sites, and native bee and syrphid fly abundance and
diversity, in comparison to relatively unmanaged edges. Native bees and syrphid flies collected
from flowers were more abundant, species-rich, and diverse at hedgerow sites than in weedy,
unmanaged edges. Abundance of bees collected passively in pan traps was negatively
correlated with floral abundance, was significantly different from communities captured by net
sampling from flowers, and did not distinguish between site types; we therefore focused on the
results of net samples and visual observations. Uncommon species of native bees were
sevenfold more abundant on hedgerow flowers than on flowers at weedy, unmanaged edges.
Of the species on flowers at hedgerows, 40% were exclusive to hedgerow sites, but there were
no species exclusively found on flowers at control sites. Hedgerows were especially important
for supporting less-common species of native bees in our intensive agricultural landscape.
Hedgerows did not concentrate ambient native bee, honey bee, or syphid fly populations, and
they acted as net exporters of native bees into adjacent fields. Within-farm habitat restoration
such as hedgerow creation may be essential for enhancing native pollinator abundance and
diversity, and for pollination services to adjacent crops.
Key words: biodiversity; Central Valley of California, USA; crop; ecosystem services; hedgerows;
intensive agricultural landscape; native bees; pollination; restoration; syrphid flies.
INTRODUCTION
Habitat enhancement within farms is thought to be an
important component for restoring ecosystem services in
intensive agricultural landscapes. Growers have little or
no control over the surrounding landscape, but can
implement within-farm enhancements. However, wheth-
er restoration on a field scale can provide benefits to
agricultural production, and thereby to growers, is
largely unknown. This lack of information is hindering
widespread adoption of within-farm habitat enhance-
ment (see Griffiths et al. 2008, Brodt et al. 2009).
Loss of biodiversity in intensive agricultural land-
scapes has led to a reduction in ecosystem services that
are essential for ensuring sustainable food production
(Millennium Ecosystem Assessment 2005, Zhang et al.
2007). Managed honey bees now provide pollination
services for the majority of global food crops that
require insect-mediated pollen transfer (Klein et al.
2007). However, reliance on honey bees is becoming
increasingly expensive and risky as managed honey bee
colonies continue to decline in numbers in both North
America and Europe (see Potts et al. 2010), coinciding
with an increase in the proportion of crops that rely on
insect pollination (Aizen and Harder 2009). Increasing-
ly, growers and scientists are recognizing the value of
conserving and/or restoring native bee populations as an
alternative to such heavy reliance on honey bees for
global crop pollination (Winfree 2010, Menz et al.
2011).
Numerous studies have shown that when crops are
grown within a matrix of natural or uncultivated land,
native bees are more abundant and diverse than in more
homogenous crop areas (Morandin et al. 2007, Ricketts
et al. 2008, Garibaldi et al. 2011). Further, in such
situations, native bees can often provide adequate
pollination services to crops without the aid of managed
honey bees (Kremen et al. 2004, Winfree et al. 2008).
However, intensive agricultural landscapes (for example,
those with .80% of land devoted to rotational crops)
Manuscript received 20 June 2012; revised 25 October 2012;
accepted 1 November 2012; final version received 26 November
2012. Corresponding Editor: A. K. Brody.
1
E-mail: lora_morandin@berkeley.edu
829
dominate many parts of the world (e.g., National
Agricultural Statistics Service, CropScape 2010, avail-
able online).
2
Restoring healthy communities of native
pollinators in these intensive agricultural environments
may prove problematic because large areas of natural
and seminatural land are not available and are not likely
to be created. Restoration of small areas on farms could
counter the lack of large natural habitat areas in
intensive agricultural landscapes.
With this goal of bringing biodiversity and ecosystem
services into intensive agricultural areas, some growers
and landowners are utilizing government incentive
programs, which compensate farmers for enhancing
environments on their land. Small-scale restorations,
such as hedgerows, can use little or no arable land and
are relatively easy for landowners to install, offering
exciting potential as a means of integrating agricultural
production with conservation of biodiversity and
ecosystem services. However, there is a surprising lack
of information on how hedgerow and other within-farm
enhancements impact biodiversity and ecosystem servic-
es, especially considering the large amounts of money
spent annually on habitat restoration in the European
Union and United States (Kleijn et al. 2006, Winfree
2010).
Field edge enhancements with flowering plants may
support a greater abundance and diversity of bumble
bees (Carvell et al. 2007, Pywell et al. 2011) and other
native bee species (Hopwood 2008, Batary et al. 2011).
Flowering hedgerows can attract bees that are uncom-
mon in the landscape (Hannon and Sisk 2009) and
potentially increase biodiversity and native bee abun-
dance in depauperate agricultural landscapes. Yet, little
is known about how restoration of field edges will
impact entire pollinator communities and how restored
areas will impact biodiv ersi ty and abundan ce of
pollinators in adjacent crop fields (Winfree 2010).
If restored areas increase only forage resources, these
areas could act as concentrators of ambient pollinator
populations, potentially diminishing or adding no net
diversity or abundance of pollinators to adjacent crops.
Few studies have examined whether enhancing floral
resources on crop edges concentrates or exports polli-
nating insects to adjacent fields, a crucial question for
population restoration and long-term ecosystem service
delivery.
We assessed pollinator communities (native bees,
native syrphid flies, managed honey bees) in hedgerows
of native flowering shrubs in the Central Valley of
California over two years (see Plate 1). We compared
floral and nesting characteristics and populations of
pollinators between restored native perennial plant
hedgerows and weedy, relatively unmanaged field
margins. We assessed abundance, diversity, and com-
munity composition of pollinators both in edges of
hedgerow and control sites, and at designated distances
into crop fields. We hypothesized that: (1) hedgerow
sites would provide more nestin g opportunities for
native bees and more abundant, diverse, and continuous
floral resources for pollinators than control margins; (2)
native pollinators would be more diverse and abundant
in hedgerows, and differ in composition between
hedgerows and control edges; (3) hedgerows would
enhance both common and less-common pollinator
species; and (4) perennial hedgerows would act as net
exporters of pollinators to adjacent crop fields rather
than concentrating ambient populations from the
surrounding landscape.
M
ATERIALS AND METHODS
Study design
The study was conducted in California’s Central
Valley in the summers of 2009 and 2010. The study area
was primarily comprised of rotational field crops with
regions of seminatural oak woodland, grassland, and
riparian gallery forests to the west of some sites (Fig. 1).
All sites were surrounded by at least 85% intensively
managed cropland in a 1500 m radius. Four native plant
hedgerow sites were selected each year, with two of them
being the same in 2009 and 2010. Hedgerows were at
least 10 years of age and had a row of perennial shrubs
bordered by a stand of perennial grasses and ranged in
length from 305 m to 550 m (for species composition, see
Bugg et al. 1998, Long et al. 1998). Hedgerow plants
were chosen so that there was successive and overlap-
ping bloom from early spring to late fall.
Within each year we chose hedgerows that were
adjacent to processing tomato fields, one of the most
common crops in the region, in order to ensure that sites
shared similar crop backgrounds. For each hedgerow
site, we selected a matching control site with a weedy,
relatively unmanaged edge. We chose to compare the
hedgerows to weedy field edges because it is the most
prevalent edge type for crops in our region. Control sites
were located a minimum of 1 km and a maximum of 3
km from corresponding hedgerow sites (Fig. 1). Our
design insured independence of bee communities at
hedgerow and control sites, while allowing both
treatments to span the same environmental conditions
across the region.
Pollinators were assessed in hedgerow and control
sites (‘‘sites’’ herein refers to edges and adjacent fields)
four times (sample rounds) during each summer, with
approximately one month between sample rounds, from
early May until early August. This time frame spans the
summer crop bloom in our region. Samples were only
done on days when the temperature was at least 188C,
the wind below 2.5 m/s, and the conditions partly cloudy
to sunny for the duration of the sampling time. Because
pollinator activity is very sensitive to weather condi-
tions, collections were made at a hedgerow site and its
corresponding control site on the same day.
2
http://nassgeodata.gmu.edu/CropScape
LORA A. MORANDIN AND CLAIRE KREMEN830
Ecological Applications
Vol. 23, No. 4
Floral, nesting, and pollinator assessment
At each sample round, floral cover was assessed by
placing 50 1-m
2
quadrats along the hedgerow or control
edge, ;8 m apart. Plants in bloom were identified and
floral cover per species was estimated using seven bins
for percent cover scores. During the final sample round
each year, bee nesting habitat was assessed in each of the
50 quadrats, following Potts et al. (2005). We quantified
potential nesting resources as the percentage of quadrats
with dead wood, hollow stems, bare ground, cracked
ground, land slope, and soil hardness (using three
measurements with a penetrometer per quadrat, at the
two closest corners and the quadrat center). In addition,
we counted small (,2 cm) and large (.2 cm) cavities in
the ground, which could indicate ground-nesting bee
tunnels.
In each sampling round, pollinators were assessed
using three methods in edges and two methods in fields.
In edges, we placed a total of 21 pan traps consisting of
seven each of yellow, blue, and white traps made from
spray painted bowls (6-ounce [;177 mL] Solo plastic
bowls painted with fluorescent yellow and blue paint or
left white) containing water and a small amount of
detergent to reduce surface tension (Westphal et al.
2008). Pans were placed out in the morning, ;18 m
apart on the ground along the hedgerow or control edge,
in an alternating color pattern. Within fields, we placed
three pan traps (one of each color) at each of three
distances (10, 100, and 200 m from edges) along each of
two transects into fields. Pans were left out for five to six
hours before being collected.
We conducted timed aerial netting, capturing bees
(Apoidea) and syrphid flies (Syrphidae) visiting flowers
in edges. The collector checked every flower for the
presence of a bee or syrphid fly. If a bee or syrphid fly
was observed touching the reproductive parts of a
flower, then it was collected in the net and put into a
labeled vial specific to that plant species. The timer was
stopped after the insect was captured in the net, until the
collector was ready to recommence flower observations,
so that total observation time was standardized among
collections. Net collections were not done in fields
because of the potential damage the net could cause to
tomato flowers.
To further quantify abundance and diversity of flower
visitors at our sites, we conducted visual observations in
1-m
3
areas. At three locations along edges, bees and
syrphid flies were recorded as either landing on
FIG. 1. Hedgerow and control sites from 2009 and 2010 in Yolo County, California, USA. Four hedgerows were matched with
four control sites each year. Two hedgerow edges and two control edges were the same in 2009 and 2010, but due to rotation of
tomato crops, the other sites were different between years.
June 2013 831HEDGEROWS AND POLLINATORS IN AGRICULTURE
reproductive parts of flowers or flying through quadrat
areas. Two 4-min visual observations of flower-visiting
insects were made at each of the edge locations and one
4-min visual observation was conducted at each of the
six in-field locations previously described. Bees were
identified as either honey bees or within categories for
native bees, as defined in Kremen et al. (2011). We did
not attempt to categorize syrphid flies during visual
observations and only recorded their numbers.
D
ATA ANALYSIS
Site characteristics
Floral cover bin scores were translated into percent
cover by selecting the midpoint of each bin. Cover and
flower species richness were compared between hedge-
row and control sites using a mixed-model ANOVA
(SAS 1999) with site type as a main effect, sample round
as a repeated factor, and site nested within site type and
year, and year as random effects. We used the same
model, but excluded sample round, to compare nesting
variables between control and hedgerow edges.
Pollinator communities, abundance, and diversity
We analyzed edge and field pollinator data separately.
Throughout, native bee and syrphid fly data were
analyzed separately due to fundamental ecological
differe nces in their nesting and foraging stra tegies.
Female bees are central-place foragers, with nest sites
that they return to between foraging trips. Conversely,
syrphid flies are ubiquitous foragers and do not return to
nest sites.
We first conducted analyses of similarities between
communities collected using pans vs. nets in order to
assess whether data collected using these methods
should be analyzed separately. We used a multi-response
permutation procedure (MRPP) PC-ORD (McCune
and Mefford 2006) and found that pan and net
collections captured significantly different communities
of native bees, regardless of site type (P , 0.0001). We
therefore analyzed net and pan data separately when
comparing communities of native bees. Another reason
for analyzing pan and net data separately is the
likelihood that floral resources were competing with
pan traps for pollinating insects. Syrphid fly communi-
ties were not distinguishable by colle ction method;
however, to keep analyses consistent between syrphid
flies and native bees, we also analyzed syrphid fly
community composition separately for pan- and net-
collected specimens. Because visual data were resolved
to category for bees and to abundance only for syrphid
flies, they were analyzed separately from other data.
Bee and syrphid communities were compared statis-
tically between site types using MRPP, with nonmetric
multidimensional scaling for visual representation
(McCune and Mefford 2006). MRPP is a nonparametric
test of the null hypothesis of no difference between
species composition between two or more groups. To
compare pollinator abundance, richness, and diversity
(Shannon index) between site types, we used mixed-
model ANOVAs (SAS 1999) with site type as a fixed
effect, sample round as a repeated factor, and site nested
within year and treatment, and year as random effects.
We also examined the influence of hedgerow vs. control
on the abundance of pollinators, controlling for total
abundance of each species, using an ANCOVA analysis
for net and pan data. Site type and species we re
categorical main factors, and total abundance of each
species (total collected in either net or pan from all sites)
was the continuous variable, with a negative binomial
distribution for over-dispersion and a log link function.
Kleijn et al. (2006) examined biodiversity benefits of
agri-environment schemes in the European Union and
assessed their benefit to uncommon species by specif-
ically analyzing abundance of species (within species
groups) that were found at ,5% of sites in each country.
We did not have enough sites to model our data in that
way; hence, we first calculated species that made up
,5% of total abundance and found that 81 of the 83
species of native bees were present at ,5%. This was
because of the large predominance of two species of
native bees, Lasioglossum incompletum (Crawford) and
Halictus tripartitus (Cockerelle), which made up 83% of
our samples (64% and 19%, respectively). Syrphid flies
samples were also dominated by a small number of
species, Toxomerus marginatus (Meigen), Eupeodes
fumipennis (Thomson), and Syrphus opinator (Osten
Sacken) (60%,10%, and 7%, respectively). We therefore
adjusted our criteria for ‘‘uncommon’’ to species that
made up ,1% of the total individuals collected. We
conducted ANOVA analyses of abundance of uncom-
mon species of native bees and syrphid flies in hedgerow
and control sites, using the same model outlined
previously for abundance.
For analyses into fields (pan and visual data only), we
added distance from edge (herein ‘‘distance’’; 0, 10, 100,
200 m) and distance 3 site type interaction as fixed
effects. Abundance and richness data were over-dis-
persed and we used a log link function with a Poisson,
negative binomial, or gamma distribution, whichever
normalized the over-dispersion best for that response
variable.
We assessed whether there were ‘‘indicator’’ species
and genera of hedgerow or control sites (McCune and
Mefford 2006). The analysis contrasts individual species
performance across two or more treatments (in our case,
hedgerow and control sites). A perfect indicator species
(or genus) is both always present and exclusive to that
treatment. Based on these criteria, indicator values were
generated and tested for significance using a randomi-
zation (Monte Carlo) technique.
Collection method
We hypothesized that pollinators might be more
attracted to floral resources than they were to pan
traps. If so, abundance in pan traps should be negatively
correlated with floral cover (Baum and Wallen 2011).
LORA A. MORANDIN AND CLAIRE KREMEN832
Ecological Applications
Vol. 23, No. 4
