UNIVERSITY OF CALGARY

Concurrent effects of landscape context and managed pollinators on wild bee communities

and canola (Brassica napus L.) pollen deposition

by

Lindsay Zink

A THESIS

SUBMITTED TO THE FACULTY OF GRADUATE STUDIES

IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE

DEGREE OF MASTER OF SCIENCE

DEPARTMENT OF BIOLOGICAL SCIENCES

CALGARY, ALBERTA

JANUARY, 2013

© Lindsay Zink 2013

i

ABSTRACT

Both wild and managed bees can provide crop pollination, yet agricultural intensification

has led to an increasing dependence on managed crop pollinators and the conversion of semi-

natural areas to cultivated cropland, both of which affect the floral resources available to wild

bees. Canola fields in southern Alberta were sampled for managed and wild bees, and landscape

within a 3 km radius was classified. Abundance and diversity of wild bees, as well as pollen

deposition in canola varied positively with the area of adjacent semi-natural habitat. Yet the

effects of managed bees and landscape context were found to interact such that abundance of

wild bees in landscapes with more semi-natural land declined more strongly over time with

increased abundance of managed bees. Canola crops in more natural landscapes may have

enhanced yields, but also greater impacts on wild-bee assemblages where managed bees are

employed.

ii

ACKNOWLEDGEMENTS

Most of all I would like to thank my supervisor, Ralph Cartar, who has been an

exceptional mentor to me. Ralph was patient, insightful and always cheerful in his guidance. I

would also like to thank Mark Wonneck, who played an integral role in the development and

implementation of this project, who showed tremendous understanding during a barrage of

destructive incidents in the field, and whose encouragement led me to attend graduate school in

the first place. I am grateful to the other members of my committee, Lawrence Harder and

Darren Bender, for providing valuable feedback on study design, methods and analysis.

This project would not have been possible without cooperation from Monsanto Canada

Inc. (especially Dmytro Guzhva and Shawn Veikle) who provided access to hybrid seed

production canola fields, and from numerous southern Alberta producers (Cor Van Raay Farms

Ltd., Cregg Nichols, Lyndon Nelson, Robert Ober, Bill Torsius, Egbert Schutter, John Nikkel,

Jacob Bueckert, Turin Hutterian Brethren Inc.) who kindly allowed me sample in their

commercial canola fields. Also, I would like to acknowledge Agriculture and Agri-Food Canada,

NSERC-CANPOLIN, and Alberta Conservation Association for providing financial and

logistical support.

For help identifying bee specimens, I thank Jason Gibbs, Lincoln Best and Cory

Sheffield. For enduring long hours of often arduous fieldwork, I thank my assistants, Amanda

Kieswetter and Meaghan Crawford, as well as others who volunteered their time. I thank the

members of the Cartar/Reid lab and other peers in Ecology and Evolutionary Biology at the

University of Calgary for their camaraderie, conversation and helpful advice. And finally, I thank

my parents for their continuing encouragement and support.

iii

TABLE OF CONTENTS

PAGE

Abstract…………………………………………………………………………………….. i

Acknowledgements………………………………………………………………………… ii

Table of Contents…………………………………………………………………………... iii

List of Figures……………………………………………………………………………… v

1: GENERAL INTRODUCTION…………………………………………………………. 1

1.1 Background……..…...…………………………………………………………. 1

1.2 Objectives……..……………………………………………………………….. 2

1.3 Context…………………………………………………………………………. 3

1.3.1 Canola Production in Canada…………………………………………3

1.3.2 Bees in Canola Production…………………………………………… 4

2: CONCURRENT EFFECTS OF MANAGED-BEE ABUNDANCE AND

SEMI-NATURAL LANDSCAPE ON WILD BEES ……………………........................... 6

2.1 Abstract………………………………………………………………………… 6

2.2 Introduction…………………………………………………………………….. 6

2.3 Methods………………………………………………………………………... 9

2.3.1 Study Area, Site Selection and Characterization…………………...... 9

2.3.2 Pollinator Sampling and Floral Resource Assessment………………. 12

2.3.3 Analysis………..…………………………………………………….. 15

2.4 Results………………………………………………………………………….. 18

2.4.1 The Wild-Bee Assemblage…………………………………………... 18

2.4.2 Floral Resource Overlap……………………………………………... 19

iv

2.4.3 Landscape Scale……………………………………………………… 21

2.4.4 Effects of Managed-Bee Abundance and the Proportion of

Semi-Natural Land in the Landscape on Wild Bees……………………….. 21

2.5 Discussion……………………………………………………………………… 26

3: EFFECTS OF WILD AND MANAGED-BEE ABUNDANCE, AND

SEMI-NATURAL LANDSCAPE ON POLLEN DEPOSITION IN CANOLA…………. 30

3.1 Abstract………………………………………………………………………… 30

3.2 Introduction……………………………………………………………………. 30

3.3 Methods………………………………………………………………………... 33

3.3.1 Study Area, Site Selection and Characterization…………………...... 33

3.3.2 Pollinator Sampling………………………………………………….. 34

3.3.3 Stigma Sampling……………………………………………………... 37

3.3.4 Analysis……………………………………………………………… 37

3.4 Results………………………………………………………………………….. 39

3.5 Discussion……………………………………………………………………… 45

4: CONCLUDING DISCUSSION………………………………………………………… 49

Literature Cited…………………………………………………………………………….. 52

Appendices…………………………………………………………………………………. 64

v

LIST OF FIGURES

FIGURE TITLE PAGE

2.1 Map showing the location of study sites in southern Alberta……………… 11

2.2 Placement of field edge and canola transects at a typical

square and circular field………………………………………..................... 13

2.3 Wild and managed-bee abundance over time……………………………… 20

2.4 Correlations between wild-bee abundance, diversity and assemblage

Composition, and the proportion of semi-natural land at a range of 12

landscape radii from 0.25 km to 3 km……………………………………... 22

2.5 Visualisation of the interaction between the effects of ∆managed-

bee abundance and the proportion of semi-natural land in the

landscape on ∆wild-bee abundance, controlling for site…………………... 24

2.6 Effects of managed-bee abundance and the proportion of semi-

natural land in the landscape on wild-bee abundance, Simpson’s

Diversity Index, and sex species Simpson’s Diversity Index……………… 25

3.1 Placement of transects for bee and pollen sampling at a typical

square and circular canola field……………………………………………. 35

3.2 Abundance of wild and managed bees in commercial canola fields

and hybrid seed canola fields………………………………………………. 40

3.3 Comparison of three measures of pollen deposition for commercial

and hybrid seed canola fields…………………………………..................... 41

3.4 Effect of bee abundance and the proportion of semi-natural land

in the landscape context on three measures of pollen deposition

in canola……………………………………………………………………. 43

3.5 Relations of bee abundance and pollen deposition to the proportion

of semi-natural land at a range of 12 landscape radii from 0.25km

to 3km, controlling for canola type………………………………………… 44

1

1: GENERAL INTRODUCTION

1.1 Background

Insects, and particularly bees, provide beneficial pollination services to a wide range of

food crops, making up about one third of the Canadian diet (Richards and Kevan 2002). Some

crops depend entirely on insect pollination for seed and fruit production, whereas others benefit

from higher yields, better quality produce, or more uniform maturation (Delaplane and Mayer

2000). Recent estimates of the total economic value of insect pollination to Canadian agriculture

range from $443 million to $1.2 billion (Richards and Kevan 2002), and global dependence on

insect-pollinated crops is growing rapidly (Aizen and Harder 2009).

Both wild and managed pollinators provide these services, yet agricultural intensification

has increased dependence on managed bees and had widely documented effects on native

biodiversity, including reductions in the diversity and abundance of wild bees (Kremen et al.

2002; Kremen et al. 2004; Ricketts 2004; Tscharntke et al. 2005). Arguably, agricultural

development produces these effects primarily through changes in land use involving the

progressive conversion of semi-natural areas into cultivated cropland. Landscapes with less

semi-natural land exhibit lower abundance and diversity of wild bees, as well as less crop

pollination (Kremen et al. 2002; Morandin and Winston 2006; Morandin et al. 2007; Vergara and

Badano 2009).

Introduction of managed bees to supplement diminished natural pollination services may

further compromise wild bees. The presence of honey bees (Apis mellifera) can reduce wild bee

fitness and consequent abundance through competition for foraging resources (Goulson 2003;

Thomson 2006; Goulson and Sparrow 2009). Other managed bees such as the alfalfa leafcutter

2

bee (Megachile rotundata) may additionally compete with some wild bees for nesting sites

(Donovan 1980). Further, the spread of a number of pathogens and parasites from managed to

wild bees has been documented (Goulson 2003).

Agricultural effects, including habitat loss and managed bees, contribute to general

concerns about wild-bee declines worldwide (Goulson et al. 2008; Steffan-Dewenter and

Westphal 2008; Potts et al. 2010). The need for wild-bee conservation has been further

underscored by recent honey-bee health issues, which have led to substantially increased winter

losses on the Canadian Prairies (Currie et al. 2010). If the services of managed bees are

supplemented or complemented by the additional presence of wild pollinators, habitat provision

in agricultural areas may lead to enhanced yield ( Hoehn et al. 2008; Klein et al. 2012) and

quality of produce (Isaacs and Kirk 2006), as well as stability of pollination services (Garibaldi

et al. 2011), while providing conservation benefits to wild pollinators, native plants (Tscharntke

and Brandl 2004; Van Geert et al. 2010) and numerous other taxa (Tscharntke et al. 2005).

Separately, managed bees and the availability of semi-natural land have been studied

extensively with regard to their effects on wild-bees and their crop pollination (e.g. Steffan-

Dewenter and Tscharntke 2000; Roubik and Wolda 2001; Ricketts 2004; Morandin and Winston

2005; Artz et al. 2011; Klein et al. 2012). Few have explored these effects concurrently

(Bommarco et al. 2012; Kremen et al. 2002), although this is important in determining whether

they may have interactive effects (Didham et al. 2007).

1.2 Objectives

In this thesis, I explore the relationship between wild and managed bees, while

concurrently accounting for the proportion of semi-natural land in surrounding areas. In Chapter

3

2, I evaluate the effects of managed-bee abundance and the proportion of semi-natural land in the

landscape on wild-bee abundance, diversity and assemblage composition. In Chapter 3, I assess

pollination in canola (Brassica napus L.) as it relates to managed and wild-bee abundance, as

well as the proportion of semi-natural land in the landscape. Finally, in Chapter 4, I provide a

general discussion of the results and their implications for wild-bee conservation, as well as

suggesting topics for future research.

1.3 Context

1.3.1 Canola Production in Canada

“Canola” refers to a collection of varieties of rapeseed (also known as ‘oilseed rape’) that

have been bred to produce reduced levels of erucic acid, making the crop more palatable and

nutritious for both humans and livestock (Casséus 2009). Since the initial development of canola

on the Canadian prairies in the 1960s and 1970s, it has expanded to become one of Canada’s

most important crops. In terms of area planted, canola is second only to wheat (Statistics Canada

2012), and economically, canola surpasses wheat as Canada’s most valuable field crop, with

farm cash receipts of 2.5 billion in 2006 (Casséus 2009). Although the traditional terminology

has been maintained in Europe, many of the newer rapeseed varieties grown there also fit the

definition of canola (Canola Council of Canada 2011).

Canola fields provide a suitable context for addressing the concurrent effects of landscape

and managed pollinators on wild bees for two main reasons. First, because canola is so

widespread in Canada, it is cultivated across a range of landscapes, which vary with regard to the

proportion of semi-natural land. With a total cropped area approaching 8 million ha in 2011, and

2.5 million ha in Alberta alone (Statistics Canada 2012), this crop can be found in both highly

4

intensified landscapes dominated by cropland, as well as adjacent to large tracts of semi-natural

land, such as rangelands and river valleys, and in landscapes spanning the range of intensities

between these extremes.

Second, canola fields experience a wide range of managed-bee densities. Two main types

of canola are grown. In the first, male-sterile and male-fertile lines are intercropped to produce

seeds with ‘hybrid vigour’ (Canola Council of Canada 2011). This type of canola cultivation,

termed hybrid seed canola, is usually accompanied by the application of large numbers of

managed bees to transfer pollen between the two lines (Clay 2009). The seeds produced by these

crops are then planted the following season as commercial canola, which is largely self-fertile

and for which producers do not typically employ managed pollinators (Delaplane and Mayer

2000). Seeds produced from this subsequent generation of canola are infertile and are pressed for

oil (Canola Council of Canada 2011).

In southern Alberta, fields of these two types of canola are interspersed. To varying

degrees, commercial canola fields in this region experience managed bees spill-over from hybrid

canola seed production fields, other crops that employ managed pollinators, and local apiaries

(pers. obs.). Wild bees sampled from a combination of hybrid seed production and commercial

canola fields are therefore subject to varying densities of managed bees, as well as varying

amounts of semi-natural land in adjacent areas, presenting an exceptional opportunity to examine

these concurrent effects on wild bees and their crop pollination.

1.3.2 Bees in Canola Production

Canola is the dominant crop for managed bees in Canada. Of the approximately 560,000

colonies of honey bees (Apis mellifera) reported in Canada in 2011 (Statistics Canada 2012),

5

about half harvested nectar from commercial canola crops for honey production, and another

80,000 were employed for pollination of hybrid seed canola (Clay 2009). With each colony

housing approximately 50,000 honey bees (Canadian Honey Council 2012), these numbers

represent a substantial presence of non-native pollinators in agricultural areas, most notably in

southern Alberta, which is home to nearly all of Canada’s hybrid seed canola (Clay 2009).

Furthermore, the alfalfa leafcutter bee (Megachile rotundata), which is also not native to Canada,

has become an established supplementary pollinator in hybrid seed canola fields (Soroka et al.

2001; Clay 2009), and is now typically used in conjunction with honey-bee colonies (pers. obs.).

In 2011, over 355,000 gallons of leafcutter bees and other non-honey-bee pollinators (equivalent

to approximately 355 million bees) were purchased in Canada, and more than half of which were

used in Alberta (Kosinski 2012; Statistics Canada 2012). A diverse assortment of wild bees also

visit canola in southern Alberta, including representatives of at least 19 genera, and five families

(Morandin et al. 2007), and wild bees significantly reduce pollination deficit in commercial

canola fields (Morandin and Winston 2005; Morandin and Winston 2006).

6

2: CONCURRENT EFFECTS OF MANAGED-BEE ABUNDANCE AND SEMI-

NATURAL LANDSCAPE ON WILD BEES

2.1 Abstract

Floral resources are important in structuring bee assemblages. In agricultural areas,

competition with managed pollinators and the amount of semi-natural land present in the

landscape may both affect floral resources available to wild bees. I sampled canola fields in

southern Alberta for managed and wild bees, and classified the landscape within 3 km. The

effects of managed bees and the proportion of semi-natural land within 0.25 km were found to

interact such that increases in managed-bee abundance over time were associated with greater

declines in wild-bee abundance where there was more semi-natural land. The abundance and

diversity of wild bees were positively associated with the availability of semi-natural land, and

wild-bee assemblage composition was associated with both the availability of semi-natural land

and managed-bee abundance. Managed bees and the availability of semi-natural land appear to

act concurrently to affect wild bee populations. Studies evaluating competition between wild and

managed bees may benefit from the use of temporal analyses, which take into account potential

interactions with the availability of semi-natural land.

2.2 Introduction

Bees consume a diet almost exclusively restricted to nectar and pollen (Delaplane and

Mayer 2000). The availability of these floral resources acts is a limiting factor on wild-bee

populations (Roulston and Goodell 2011), and acts to structure bee assemblages at both the local

and landscape level (Heithaus 1979; Potts et al. 2003; Hines and Hendrix 2005). Resource

7

partitioning among co-occurring bee species provides evidence of the importance of competition

for these limited food resources in bee assemblages (Pyke 1982).

The presence of managed bees in an ecosystem has long been suspected to detrimentally

affect wild-bee populations through resource competition (Paini 2004). Managed bees present an

additional, often highly abundant, competitor in the bee assemblage. Furthermore, during periods

of resource scarcity managed bees are often protected from the effects of competition through the

provision of supplemental food sources that are not available to wild bees (Standifer 1980).

The most prominent managed bee, the European honey bee Apis mellifera, is a generalist

forager, visiting nearly 40,000 plant species worldwide (Crane 1990), and therefore has the

potential to share resources with a wide range of wild bees. Indeed, overlapping use of floral

resources used by wild and managed bees has been well-established. Considerable floral

resource overlap exists between wild and managed bees in a variety of environments (Roubik et

al. 1986; Thorp et al. 1994; Wilms and Wiechers 1997; Steffan-Dewenter and Tscharntke 2000),

reaching levels as high as 80-90% (Thomson 2006). However, beyond this preliminary level of

analysis, evidence of competition has proven more difficult to establish. Some studies have

documented reduced wild-bee floral visitation rates (Ginsberg 1983; Gross and Mackay 1998)

and resource harvesting rates (Thomson 2004) in the presence of managed bees, as well as more

direct evidence of competition in the form of reduced bumble bee worker size (Goulson and

Sparrow 2009) and reduced bumble-bee and solitary-bee reproductive success (Thomson 2004;

Paini and Roberts 2005). Numerous other studies however, have had conflicting or non-

significant results, and often suffer from inadequate replication or confounding factors, leading

to the overall conclusion that clear and consistent evidence for competitive impacts on wild-bees

is still lacking (Butz Huryn 1997; Goulson 2003; Paini 2004).

8

The failure of many studies exploring competition between wild and managed bees to

take into account the effects of landscape context may be one reason for the inconsistent

findings. At least for highly-mobile species, landscape-level resources may be even more

important than local resources in determining the abundance and diversity of wild bees (Hines

and Hendrix 2005). The most well-established relationship between wild bees and landscape

context is with semi-natural land, which provides diverse foraging resources, nesting sites and

materials, as well as refuge from pesticides. Widespread evidence indicates wild-bee abundance

and diversity vary positively with both the proximity of semi-natural land (Ricketts 2004;

Ricketts et al. 2008; Carvalheiro 2010), and the proportion of semi-natural land in a landscape

(Morandin and Winston 2006; Morandin et al. 2007).

This relation of bees to the availability of semi-natural land may confound results in two

ways. First, because honey bees are largely a domesticated species cultured by humans, they tend

to be more abundant where human population density is greater, coinciding with the presence of

more highly modified landscapes containing less semi-natural land (Butz Huryn 1997; Paini

2004). Wild bees may therefore respond simultaneously to these two confounded gradients.

Second, managed-bee abundance and the availability of semi-natural land may have interactive

effects on wild bee assemblages. Habitat loss can change the per capita impact of invasive

species on native populations, leading to synergistic effects (Didham et al. 2007), which suggests

wild bees may be more vulnerable to impacts in more highly modified landscapes (Butz-Huryn

1997). For example, if wild bees experience more resource competition when resources are

scarcer, then the impact of managed bees on characteristics of the wild bee assemblage

(abundance, diversity and assemblage composition) may be greater where there is less semi-

natural land.

9

Here, I concurrently address the effects of managed-bee abundance and the availability of

semi-natural land on wild-bee assemblages. Sampling occurs before, during and after major

introductions of managed bees across a large number of sites, as well as at sites where few or no

managed bees were introduced, and sites where managed bees were continually present. This

design allows for characteristics of the wild-bee assemblage to be tested for association both with

managed-bee abundance and change in managed-bee abundance over time, while accounting for

the availability of semi-natural land and testing for interacting effects.

2.3 Methods

2.3.1 Study Area, Site Selection and Characterization

The study area was located in southern Alberta, in an area dominated by irrigation

agriculture and cattle farming. It spanned 126 km east-west and 43 km north-south, and was

roughly located between the towns of Coalhurst and Bow Island (Figure 2.1). I selected 32

canola fields as study sites. Fifteen sites comprised fields planted to hybrid seed canola, and

employed managed pollinators during crop flowering (all used both A. mellifera and M.

rotundata, except for site, which used only A. mellifera). Another fifteen sites comprised fields

planted to commercial canola, and did not employ managed pollinators onsite. The other two

sites were research trial fields planted to hybrid seed canola, but also not employing any

managed pollinators onsite. All site types had varying spill-over of managed bees throughout the

season from bees kept for the pollination of other nearby crops (canola or otherwise) and non-

crop related honey production. I selected sites of each type (commercial, hybrid seed, and

research trial) to maximize variation in the amount of semi-natural land within a 3 km radius,

10

and ensured the different site types were spatially interspersed. I also ensured that all sites were

at least 5 km apart, in an attempt to maintain independence of bee assemblages.

Using aerial photographs and ground-truthing, I classified the land cover around each

study site into 2 categories: 1) semi-natural land, including native and non-native vegetation in

uncultivated, non-cropped areas such as river valleys, pastures, unused pivot corners,

shelterbelts, and ditches, and 2) all other land cover types, including cultivated land, roads,

residential/urban areas and large bodies of water. I then used ArcMap software (version 10.0,

ESRI, Redlands CA) to calculate the proportion of semi-natural land at 12 landscape scales from

0.25 km to 3 km (by increments of 0.25 km). I selected 3 km as the largest landscape scale

because this distance exceeds the recorded typical foraging range for any wild-bee species

(Walther-Hellwig and Frankl 2000; Gathmann and Tscharntke 2002).

I sampled each site three times during the 2011 growing season, coinciding roughly with

the periods: 1) before canola bloom and associated managed pollinator introductions, 2) during

canola bloom, and 3) after canola bloom and the removal of managed pollinators. Initiation of

flowering between sites varied considerably and flowering lasted several weeks, so that some

sites experienced some flowering during the first and last sample set. All fields were flowering

during the second sample set, except for one commercial canola site, which was sampled at both

the start and end of the last sample set (for a total of four samples) to account for unusually late

flowering.

I randomly assigned sites to morning (10:00-12:00), midday (12:30-14:30) or afternoon

(15:00-17:00) sampling times, so that three sites could be sampled each day. I maintained this

assignment across all sampling periods to ensure seasonal changes in the bee assemblage at a site

11

Figure 2.1: Map showing the location of study sites in southern Alberta (ArcMap 10.0).

12

were not confounded with daily preferences in foraging time. I assigned equal numbers of hybrid

and commercial sites to each time of day and randomly assigned the two research trial sites to

morning and midday sampling times. I also randomized the order in which sites were selected for

sampling but this assigned order was frequently altered to account for pesticide spraying, driving

distances between sites, and adverse weather conditions. Whenever possible, I replaced a site

that could not safely or feasibly be sampled with the next randomly selected site.

2.3.2 Pollinator Sampling and Floral Resource Assessment

At each study site, I selected a point of reference. For square canola fields, this was

typically located in the field corner most accessible by vehicle (Figure 2.2a), and for circular

canola fields, the point was along the edge of the pivot area, adjacent to this corner (Figure 2.2b).

I then located six 100 m x 3 m transects within the 250 m radius surrounding this point. Four of

these transects were ‘field edge’ transects, located haphazardly in the best non-crop, flowering

habitat available within this area, as defined by a subjective appraisal of the abundance and

diversity of flowers present at the time of sampling. These field edge habitats included ditches,

field margins, un-cropped pivot corners, and adjacent pastures. The other two, ‘canola’ transects,

I located within the crop itself parallel with a field edge, beginning 25 m and 125 m respectively,

from this edge. I sampled field-edge transects during all three visits to each site, but canola-

transects only when the crop was in flower, defined by at least 1 open canola flower per m2.

I sampled bees along each transect using both aerial and sweep netting. Aerial netting

was important for associating bees with the flowers they visited for floral resource overlap

analyses. However, because all bee sampling methods are subject to sampling biases (Westphal

et al. 2008; Grundel et al. 2011), I also used sweep netting, to gain a more representative

13

Figure 2.2: Placement of field edge (1-4) and canola (5-6) transects at a typical square (a) and

circular field (b). Cardinal direction varied, and placement of field edge transects varied within

the 250 m radius, based on the best available flowering habitat. The intersecting grey lines

represent roads, with adjacent vegetated ditches/field margins in white. The star represents the

reference point of the site for landscape analysis, and the circle represents the smallest scale of

landscape analysis (250 m), within which all transects were located.

14

measure of the overall bee assemblage. To minimize collector biases associated with netting bees

(Westphal et al. 2008), the same collector conducted all sweep-netting samples, and aerial

netting was limited to two collectors.

I conducted aerial netting over a 15-min interval on each transect, targeting only bees

visiting flowers. I identified managed bees in the field without netting, and captured wild bees

for later identification in the lab. I stopped the timer after the capture of a wild bee so that

handling time was not included in the 15-min search time. During aerial netting I also noted the

species of flowering plant being visited by each bee (wild or managed). Sweep netting was

conducted by sweeping once across the vegetation on each meter of transect using a 30 cm

diameter sweep net, for a total of 100 sweeps with an approximately 2 m wide arc. This method

captured managed and wild bees in and around the vegetation, whether they were visiting

flowers or not. Any wild bees that escaped during aerial or sweep netting were recorded and

included in measures of abundance but not diversity, since they could not be identified reliably.

For wild-bee identification, I employed a number of sources (Stephen 1957; Mitchell

1960; Mitchell 1962; Roberts 1973a; Roberts 1973b; Curry 1984; Gibbs 2010; Sheffield et al.

2011; NSERC-CANPOLIN 2012; de Silva and Packer 2012; Discover Life) as well as experts

Jason Gibbs (Cornell University, Ithaca NY), Lincoln Best (York University, Toronto ON) and

Cory Sheffield (Royal Saskatchewan Museum, Regina SK). When possible, wild bees were

identified to species. However for some genera, species-level keys were not available for this

region. I sorted specimens from these genera into morphospecies using distinguishable

morphological traits, especially those indicated in keys from other regions.

Along all transects where bees were sampled, I quantified floral resources by counting

and identifying all open, showy flowers (forbs and shrubs). For composite flowers, I counted

15

inflorescences rather than individual flowers. Flowers from plant species for which no bee visits

were observed over the course of the season I later excluded from the floral counts and any

further analysis.

2.3.3 Analysis

I calculated floral resource overlap in two ways, using methods roughly analogous to

those of Steffan-Dewenter and Tscharntke (2000). First, I determined the percentage of

overlapping plant species (used at least once by both wild and managed bees) and the percentage

of overall floral abundance made up of these overlapping species. Second, I calculated Hurlbert’s

index of niche overlap (1978):

∑

where: = the total abundance of flowers

X and Y = the total abundance of the two bee taxa x and y

xi and yi = the abundances of bee taxa x and y on flowers of plant species

i , respectively

ai = the abundance of flowers of plant species i

This index defines niche overlap as the difference in resource use between species (or in

this case also species groups) from that if each used all resources in proportion to their

abundance. Values greater than 1 indicate that the bee taxa share resource preferences. I used this

16

index to calculate overlap both between the wild-bee assemblage and managed bees

cumulatively, as well as each of the two managed-bee species individually. To determine the

uncertainty in Hurlbert’s index, I bootstrapped each value 1000 times with R software (version

2.13.2, The R Foundation for Statistical Computing, Vienna, Austria).

I evaluated the effects of managed-bee abundance and the proportion of semi-natural

land in the landscape on site-visit-level abundance and diversity of wild bees with least squares

ANCOVA models using JMP software (version 10.0.0, SAS institute Inc., Cary NC). In addition

to including managed-bee abundance, the proportion of semi-natural land, and their interaction

as predictor variables, both models also included site as a random factor.

As a measure of wild bee diversity, I used Simpson’s Diversity Index on all site visit

samples with 2 or more wild-bee specimens:

(∑

)

where = the number of individuals of a particular species

= the total number of individuals of all species

I chose Simpson’s Diversity Index (SDI) because, unlike other measures such as species

richness and Shannon Diversity, it is unbiased with regard to sample size (Lande 1996; Lande et

al. 2000). To account for possible lack of associated between male and female specimens from

morphospecies of the same genus, I calculated diversity in two ways. The first was a standard

calculation of SDI at the species level, which associated male and female morphospecies

17

specimens of the same genus based on their relative abundances at a given site visit. Most

samples did not require these associations, but when they were made diversity could be

underestimated if male and female morphospecies specimens were in fact from different species.

I therefore also calculated a second measure of Simpson’s Diversity using “sex species” in place

of species-level categorization (SSSDI). This measure kept male and female specimens of

morphospecies separate, as well as those from identifiable species, to avoid biasing the diversity

measure toward less identifiable genera.

I evaluated effects on wild-bee assemblage composition through distance-based

redundancy analysis (DISTLM) in PRIMER software (version 6.0, PRIMER-E Ltd., Plymouth

UK). This analysis uses permutational methods to test the relationships between predictor

variables and a response matrix, in this case a ln-transformed Bray Curtis similarity matrix of the

proportional abundances of all species and morphospecies at each site visit. Here I did not

associate male and female morphospecies because keeping them separate avoided erroneous

associations and did not create a bias toward these genera. DISTLM is limited to the use of

sequential sum of squares for hypothesis testing, therefore the order of the predictors was

important. I ordered predictors with the main effects first (semi-natural land, then managed-bee

abundance), followed by the interaction. Also, the DISTLM analysis does not allow for the

inclusion of random effects, therefore to account for repeated sampling of study sites and reduce

the chance of Type I error, I adjusted the Type I error rate according to the procedure developed

by Benjamini and Hochberg (1995; 2000).

I calculated the change in abundance (∆wild-bee abundance, ∆managed-bee abundance)

and diversity (∆wild-bee SDI, ∆wild-bee SSSDI) over time using contrasts of the 1st and 2

nd

sample set, and the 2nd

and 3rd

sample set. I then tested the effects of ∆managed-bee abundance,

18

semi-natural land, and their interaction, on ∆wild-bee abundance, ∆wild-bee SDI, and ∆wild-bee

SSSDI (independently) using least-squares regression models, including site as a random factor.

One study site was an outlier in the ∆wild-bee abundance model, containing very few wild bees

during the second sample set relative to other samples, and was therefore excluded from this

particular analysis to prevent it from having disproportionate influence on the model outcome.

I selected the most predictive landscape scale for each response variable (wild-bee

abundance, ∆wild-bee abundance, wild-bee SDI and ∆wild-bee SDI, wild-bee SSSDI and ∆wild-

bee SSSDI, and Bray Curtis Similarity matrix) through univariate analysis at each of the 12

scales, using the same software and model class as for the relevant full model with all terms. I

then included the scale with the lowest p-value in related full models for each variable.

I performed Mantel tests on the residuals of all ANCOVAs, using package ade4 in R

software (version 2.13.2, The R Foundation for Statistical Computing, Vienna, Austria). None

showed significant spatial autocorrelation. I also tested variance inflation factors to address

potential concerns about collinearity among variables. All were < 2. I applied Box-Cox

transformations as needed to meet assumptions of normality and homoscedasticity of residuals.

Canola type (commercial vs. hybrid seed) was tested as a potential nuisance variable in all

models, but resulted in no significant effects or interactions; therefore it was excluded from the

model statistics presented here.

.

2.4 Results

2.4.1 The Wild Bee Assemblage

In total, I recorded 7523 managed bees and 1549 wild bees. Wild bees represented 54

species and at least 44 morphospecies (a conservative estimate assuming all female

19

morphospecies correspond directly to male morphospecies of the same genus), from 26 genera

(Appendix 1). The most abundant genera were Lasioglossum, Bombus and Andrena, accounting

for 75% of total wild-bee abundance. Overall, wild and managed-bee abundance exhibited

contrasting trends over the season (Figure 2.3). Prior to formal statistical analysis, it is clear that

managed-bee abundance peaked mid-season, which contrasts with a seeming mid-season decline

in wild-bee abundance.

2.4.2 Floral Resource Overlap

On field edge transects, I recorded 3416 bee-flower visits, made by wild bees (31%), A.

mellifera (35%), and M. rotundata (34%). I identified 89 flowering plant species (Appendix 2);

however, I recorded bees foraging on only 51 of these species. Non-native species made up 53%

of the plant species with flowers visited by bees, and 95% of their total field-edge flower

abundance. The plant species that received the most wild-bee visits were Helianthus annuus,

Crepis tectorum, Brassica napus, Sonchus arvensis, and Melilotus officinalis, whereas those

receiving the most managed-bee visits were Medicago sativa, Melilotus officinale, Melilotus

alba, Cirsium arvense, and Brassica napus. Floral resource overlap between wild bees and

managed bees represented 57% of plant species used by bees and 98% of their field edge flower

abundance.

Hurlbert’s index of niche overlap between wild and managed bees collectively was

9.34 (95% CI: 9.09 – 9.59), indicating that the preferences of wild and managed bees were

approximately 9 times higher than if both species groups used all floral resources in proportion

to their abundance. The foraging preferences of A. mellifera overlapped more strongly with the

20

Figure 2.3: Wild and managed-bee abundance over time. Total bee abundances at each study site

are represented once in each third of the graph, corresponding roughly with periods before,

during, and after canola flowering and associated managed pollinator introductions (except one

commercial canola site, which is represented once in each of the first two samples, and twice in

the final sample to account for unusually late flowering). Lines are splines where λ = 0.7.

21

wild-bee assemblage than did those of M. rotundata [12.39 (95% CI: 12.03 – 12.75) vs. 3.22

(95% CI: 3.16 – 3.28)].

2.4.3 Landscape Scale

The proportion of semi-natural land within 3 km of the study sites ranged from <1% to

81%. Wild-bee abundance, diversity and assemblage composition responded differently to

landscape scale (Figure 2.4). The proportion of semi-natural land at a 0.25 km radius was most

predictive of wild-bee abundance (F1,95 = 12.29, P = 0.0007). Wild-bee diversity exhibited two

distinct peaks at landscape radii of 0.75 km and 2 km, with the former being most predictive of

wild-bee SDI (F1,82 = 8.85, P = 0.004), and wild-bee SSSDI (F1,82 = 8.62, P = 0.004).

Assemblage composition was relatively unresponsive to landscape scale, but was most highly

associated with the proportion of semi-natural land at a 2 km radius (Pseudo F1,87 = 1.84, P =

0.016). None of the ∆ variables were significantly associated with the proportion of semi-natural

land at any scale; therefore the same radius was used in these models as for their respective raw

variables.

2.4.4 Effects of Managed-Bee Abundance and the Proportion of Semi-Natural Land in the

Landscape on Wild-Bees

The proportion of semi-natural land within a 0.25 km radius interacted with ∆managed-

bee abundance to affect ∆wild-bee abundance. Wild-bee abundance declined more strongly with

additions of managed bees and increased more quickly with managed bee removals at sites with

more semi-natural land (F1,59 = 4.25, P = 0.048; Figure 2.5). Wild-bee abundance was not subject

to interactive effects (F1,93 = 0.07, P = 0.79), or any association with managed-bee abundance

22

Figure 2.4: Correlations between wild-bee abundance, diversity, and assemblage composition,

and the proportion of semi-natural land at a range of 12 landscape radii from 0.25 km to 3 km.

Triangles correspond to wild-bee abundance, open circles to Simpson’s Diversity Index at the

species level, solid circles to Simpson’s Diversity Index at the sex-species level, and squares to

assemblage composition.)

23

(F1,93 = 0.01, P = 0.92), but was positively associated with the proportion of semi-natural land

(F1,93 = 7.92, P = 0.009; Figure 2.6).

There were no significant interactions between the effects of managed-bee abundance

and the proportion of semi-natural land on any measures of wild-bee diversity (SDI: F1,80 =

0.003, P = 0.95; SSSDI: F1,80 = 0.13, P = 0.72; ∆SDI: F1,48 = 0.01, P = 0.93; ∆SSSDI: F1,48 = 0.32

, P = 0.57). Change in managed-bee abundance had a marginally significant positive effect on

∆SDI (F1,48 = 3.44, P = 0.073), but managed bees were otherwise not associated with diversity

(SDI: F1,80 = 1.40, P = 0.24; SSSDI: F1,80 = 0.93, P = 0.34; ∆SSSDI: F1,48 = 1.37, P = 0.25;

Figure 2.6). Diversity varied positively with the proportion of semi-natural land (SDI: F1,80 =

7.05, P = 0.013; SSSDI: F1,80 = 6.38, P = 0.017), but not changes in diversity over time (∆SDI:

F1,48 = 0.88, P = 0.36; ∆SSSDI: F1,48 = 0.04, P = 0.84). Removing landscape from these models

did not change the overall non-significance of the effect of managed-bee abundance on wild-bee

diversity (SDI: F1,82 = 2.34, P = 0.13; SSSDI: F1,82 = 1.50, P = 0.22; ∆SDI: F1,50 = 3.62, P =

0.066; ∆SSSDI: F1,50 = 1.32, P = 0.26). I excluded 13 of the 97 site visits (13%) from diversity

calculations because they had fewer than 2 wild-bee specimens.

Wild-bee assemblage composition was significantly associated with the proportion of

semi-natural land within 2 km (Pseudo F1,87 = 1.840, P = 0.021) and managed-bee abundance

(Pseudo F1,86 = 1.897, P = 0.013) and at the Benjamin Hochberg-corrected levels (α = 0.033; α =

0.017). After taking into account these main effects, their interaction was not significant (Pseudo

F1,85 = 1.237, P = 0.210). Removing landscape from the model did not affect the level of

significance of managed-bee abundance on assemblage composition (Pseudo F1,87 = 1.924, P =

0.013).

24

Figure 2.5: Visualisation of the interaction between the effects of ∆managed-bee abundance and

the proportion of semi-natural land in the landscape on ∆wild-bee abundance, controlling for

site. Data points show variation in ∆managed-bee abundance, ∆wild-bee abundance, and are

adjusted for the effects of semi-natural land and site (random), but not the interaction between

semi-natural land and ∆managed-bee abundance. Lines represent quintiles of the proportion of

semi-natural land within a 0.25 km radius.

25

Figure 2.6: Effects of managed-bee abundance and the proportion of semi-natural land in the

landscape on wild-bee abundance, Simpson’s Diversity Index, and sex species Simpson’s

Diversity Index. Lines represent significant relation.

26

2.5 Discussion

I found evidence that wild bee populations are subject to interactive effects of changes in

managed bee abundance and the proportion of semi-natural land in the landscape. Contrary to the

interaction I predicted, and speculation by Butz-Huryn (1997) that wild and managed bees

compete more strongly in landscapes that are more highly modified, this study provides evidence

that managed bees have greater negative effects on wild-bee abundance in landscape more

hospitable for wild bees. Change in managed-bee abundance over time interacted with the

proportion of semi-natural land within 0.25 km to affect change in wild-bee abundance. In more

natural landscapes wild bee abundance declined more steeply with addition of managed bees and

increased more quickly after managed bee removal than in landscapes with less semi-natural

land. It appears that landscape with more semi-natural land supported larger populations of

native bees, which were then more detectably affected by managed pollinator introductions.

The abundance of wild bees can reflect the availability of resources in previous years (Le

Féon et al. 2011), so one possible explanation for this finding is that wild bees in landscapes with

less semi-natural land are supressed from greater long-term competition with managed bees.

These landscapes consistently had more cultivated cropland, and may therefore have contained

more crops employing managed pollinators during previous years. In this case, these wild-bee

populations may have suffered declines largely before, rather than during the season in which

sampling took place, when compared to wild bees in more natural landscapes. Similarly, other

effects of managed bees (e.g. disease), or agricultural intensification (e.g. pesticides), could also

suppress wild bee populations in this way. Another possibility is that greater proportions of semi-

natural land in the surrounding landscape allowed wild bees to simply avoid areas immediately

adjacent to crops when managed bees were present (and therefore also avoid my sampling area).

27

However, the small scale at which landscape effects were observed most strongly, combined

with the smaller foraging range of most wild bees compared to honey bees (Beekman and

Ratnieks 2000; Walther-Hellwig and Frankl 2000; Gathmann and Tscharntke 2002) argues

against this explanation for the observed interaction.

Floral resource use by wild and managed bees in field edges overlapped considerably,

indicating that managed bees preferentially depleted flowers that wild bees also favoured. Apis

mellifera overlapped more with the wild bee assemblage than M. rotundata, and exhibited

coinciding floral preferences 12 times greater than if plant species were used according to their

flower abundance. Only about 44% of wild bee taxa were found visiting canola flowers (in the

crop or in field edges), so for most wild bees flowers in these field edges are likely representative

of their core foraging resources. The overall trend in wild-bee abundance over the season (Figure

2.3) also suggested competitive impacts, although similar mid-season declines in wild-bee

abundance are possible without large-scale introductions of managed pollinators (Oertli et al.

2005). Further, within-season declines in wild bee abundance associated with increases in

managed bee density, do not necessarily lead to long-term population-level declines (Roubik and

Wolda 2001).

Findings here are consistent with other evidence that wild-bee species respond differently

to competition with managed bees (Artz et al. 2011). Managed-bee abundance was associated

with wild-bee assemblage composition, and had a marginally positive effect on change in

diversity over time. Species-specific responses to managed bees were not examined here, but

managed bees may compete more strongly with some of the more abundant wild-bee species,

leading them to decline more rapidly than less abundant wild-bee species, thereby increasing

assemblage evenness (and SDI), even as overall wild-bee abundance declined.

28

Even in the highly intensified agricultural landscapes of southern Alberta, semi-natural

land in the form of ditches, field margins, pastures and river valleys, was shown to provide

valuable habitat for wild bees. The abundance and diversity of wild bees were both positively

associated with the proportion of semi-natural land in the landscape, as has been observed

elsewhere (Kremen et al. 2002; Morandin and Winston 2006; Morandin et al. 2007; Carré et al.

2009). Wild-bee assemblage composition was also associated with the proportion of semi-natural

land in surrounding areas, likely due to species-specific levels of dependence on these habitat

types (Greenleaf and Kremen 2006; Kim et al. 2006).

The abundance of wild bees varied strongly with the proportion of semi-natural land in

the local landscape within a 0.25 km radius of the sampling area. This scale is on the lower end

of the range of landscape effects on wild-bee abundance found in the literature, which vary from

0.25 km to 2.4 km (Steffan-Dewenter et al. 2002; Morandin and Winston 2006; Bommarco

2012). Bee foraging range depends strongly on body size (Gathman and Tscharntke 2002;

Greenleaf et al. 2007), so the small scale identified here may reflect the prominence of small-

bodied bees in the bee assemblage in my study area [especially Lasioglossum (Dialictus)], which

are expected to have relatively short foraging ranges. Studies that focused on the landscape

response of solitary bees have found a scale of 0.25 km-0.75 km to be most predictive (Steffan-

Dewenter et al. 2001), whereas others that focus on bumble bees point to the importance of

larger scales of up to 3 km (Steffan-Dewenter et al. 2001; Hines and Hendrix 2005; Westphal et

al. 2006; Knight et al. 2009). Similarly, wild-bee diversity exhibited two peaks in its response to

landscape scale, which may relate to these divergent scales at which bumble bees and smaller-

bodied bees respond. Overall, diversity was most responsive to the amount of semi-natural land

at 0.75 km, which is consistent with findings by Steffan-Dewenter et al. (2002).

29

In summary, although the nature of the interaction between the effects of managed bees

and landscape context was not as predicted, this study still reveals the importance of accounting

for landscape-level drivers of wild-bee abundance in studies of competition between wild and

managed bees. It also supports the concerns of Butz-Huryn (1997) and Paini (2004) that

landscape context may confound studies of this nature. The abundance of managed bees did not

correlate with the proportion of semi-natural land in the landscape in this study area (analysis not

shown). However, where such correlation occurs, my results suggest that the competitive

impacts of managed pollinators on wild-bee populations would be difficult to detect. If managed-

bee abundances tended to be greater where there was less semi-natural land in the landscape,

these confounded interacting effects would act to mask the effects of competition. Managed bees

would have a lesser effect on a per-bee-basis where they were more abundant, and a greater

effect where they were less abundant.

This is not to say that simply accounting for the availability of semi-natural land in

competition studies will resolve all inconsistencies in the field of competition between managed

and wild bees. Even in my study, the effects of managed bees were not visible through

correlation of abundances, but rather only when looking at temporal contrasts that involved

drastic changes in managed-bee abundance at diverse sites. Other studies may also do well to

incorporate temporal comparisons into their study design when possible. Given present concerns

about wild bee conservation, and the increasing use of managed crop pollinators, it is important

to use the most effective methods available to clarify the nature of this relationship.

30

3: EFFECTS OF WILD AND MANAGED-BEE ABUNDANCE, AND SEMI-NATURAL

LANDSCAPE ON POLLEN DEPOSITION IN CANOLA

3.1 Abstract

Promoting semi-natural land in agricultural areas has been suggested as a means to

increase abundance of wild bees and related crop pollination, yet few previous studies have

accounted for the additional presence of managed pollinators. In fields of two types of canola

(commercial and hybrid seed) in southern Alberta, I counted managed and wild bees, collected

stigmas to measure pollen deposition, and classified the surrounding landscape within 3 km.

Abundance of wild bees, but not managed bees, correlated positively with the proportion of

semi-natural land. Pollen deposition was higher in commercial canola fields than in hybrid seed

canola fields, despite lower bee abundance. Managed bees greatly outnumbered wild bees in

both canola types, and pollen deposition correlated positively with the abundance of managed

bees. Despite no detectable correlation between wild-bee abundance and pollen deposition, the

positive relation between semi-natural land and pollen deposition suggests wild pollinators also

contribute significantly to canola pollination. Semi-natural land leads to enhanced crop

pollination even in intensified agricultural landscapes where managed pollinators greatly

outnumber wild bees.

3.2 Introduction

Agricultural intensification has increased dependence on the use of managed bees for

crop pollination, yet wild pollinators also provide this ecosystem service (Kremen et al. 2002;

Breeze et al. 2010; Rader et al. 2012). Enhancement or conservation of semi-natural land in

31

agricultural landscapes, and integration of different agricultural land-use types have been

recommended as ways to increase wild-bee abundance and their ‘free’ crop pollination (Kremen

et al. 2004; Morandin and Winston 2006; Morandin et al. 2007; Ricketts et al. 2008), yet few

have addressed the potential for these strategies in intensified agricultural landscape where

managed bees are abundant (Bommarco et al. 2012).

Canola (Brassica napus L.) is one of Canada’s most widespread crops, as well as its most

economically valuable (Casséus 2009), making it a potentially rewarding candidate for exploring

the role of wild bees and indirectly, semi-natural land, in providing pollination services. Two

main types of canola are produced, each with distinct pollination requirements. The first, termed

hybrid seed canola, is made up of intercropped ‘male’ (pollen donor) and ‘female’ (seed-

producing, not pollen-bearing) plants, and is highly dependent on insect pollination to transfer

pollen between lines (Canola Council of Canada 2011). During flowering of hybrid seed canola,

honey bees (Apis mellifera) are stocked at 3-8 colonies per hectare, and typically supplemented

by managed alfalfa leafcutter bees (Megachile rotundata) (Clay 2009). The second type, termed

commercial canola, does not typically employ managed pollinators because it is considered to be

self-fertile (Delaplane and Mayer 2000). There is some evidence however, that commercial

canola can also benefit from bee-mediated pollination (Morandin and Winston 2005; Morandin

and Winston 2006; Bommarco et al. 2012), with improvements to yield of as much as 46% with

additions of A. mellifera (Sabbahi et al. 2005). These benefits may arise because: a) not all plants

in a population are self-fertile, b) not all self-fertile plants will fully self-pollinate, and c) cross-

pollination can result in greater yields than self-pollination (Free 1993).

In addition to managed bees, a diverse assortment of wild bees are found in canola crops

(Morandin et al. 2007), most of which are native. Wild-bee abundance in canola fields has been

32

found to be positively associated with the proportion of semi-natural land in surrounding areas

(Morandin and Winston 2006; Morandin et al. 2007), and result in increased seed set (Morandin

and Winston 2005; Morandin and Winston 2006), seed weight per plant (Bommarco et al. 2012),

and seed quality (Bommarco et al. 2012), as well as reduced genetic contamination from wind-

carried pollen from other canola fields (Hayter and Cresswell 2006). However, these studies

were implemented exclusively in commercial canola fields, and managed bees are either rare

(Morandin and Winston 2005), ignored (Morandin and Winston 2006; Morandin et al. 2007), or

uniform in their abundance (Bommarco et al. 2012), preventing the separation of their

contributions from those of wild pollinators. In southern Alberta, Canada, managed bees are used

extensively in hybrid seed canola, and also spill-over into commercial canola fields to varying

degrees (pers. obs.), therefore canola fields with a range of managed bee densities can be

selected for sampling.

Here, I ask whether wild bees contribute significantly to pollination of commercial and

hybrid seed canola in an intensified agricultural landscape where they are greatly outnumbered

by managed bees. If in this context, wild-bee pollination services are inconsequential, because

too few wild bees are present to make a significant contribution to crop pollination, or because

they do not forage in the crop with so many managed bees present, then pollen deposition would

not be associated with wild-bee abundance or the proportion of semi-natural land in the

landscape. If their services are redundant, because canola fields do not suffer pollen limitation

with so many managed bees present, any such effects would occur above the level of full

pollination, at which all ovules are expected to develop. However, if wild bees do provide

significant pollination services, which are supplemental to those of managed bees (or even

33

synergistic), pollen deposition should vary positively with both wild bee abundance and the

proportion of semi-natural land in the landscape.

3.3 Methods

3.3.1 Study Area, Site Selection and Characterization

The study area was located in southern Alberta, in an area dominated by irrigation

agriculture and cattle farming. It spanned 126 km east-west and 43 km north-south, and was

roughly located between the towns of Coalhurst and Bow Island (Figure 2.1). I selected 32

canola fields as study sites. Fifteen sites comprised fields planted to hybrid seed canola, and

employed managed pollinators during crop flowering (all used both A. mellifera and M.

rotundata, except for site, which used only A. mellifera). Another fifteen sites comprised fields

planted to commercial canola, and did not employ managed pollinators onsite. The other two

sites were research trial fields planted to hybrid seed canola, but also not employing any

managed pollinators onsite. All site types had varying spill-over of managed bees throughout the

season from bees kept for the pollination of other nearby crops (canola or otherwise) and non-

crop related honey production. I selected sites of each type (commercial, hybrid seed, and

research trial) to maximize variation in the amount of semi-natural land within a 3 km radius,

and ensured the different site types were spatially interspersed. I also ensured that all sites were

at least 5 km apart, in an attempt to maintain independence of bee assemblages.

Using aerial photographs and ground-truthing, I classified the land cover around each

study site into 2 categories: 1) semi-natural land, including uncultivated, non-cropped areas such

as river valleys, pastures, unused pivot corners, shelterbelts, and ditches, and 2) all other land

cover types, including cultivated land, roads, residential areas and large bodies of water. I then

34

used ArcMap software (version 10.0, ESRI, Redlands CA) to calculate the proportion of semi-

natural land at 12 landscape scales from 0.25 km to 3 km (by increments of 0.25 km). I selected

3 km as the largest landscape scale, because this distance exceeds the recorded typical foraging

range for any wild-bee species (Walther-Hellwig and Frankl 2000; Gathmann and Tscharntke

2002).

Sampling occurred between July 9th

and August 15th

, 2011. I sampled each site once, while the

canola crop was in flower. I randomly assigned sites to morning (10:00-12:00), midday (12:30-

14:30) or afternoon (15:00-17:00) sampling time, such that three sites could be sampled each

day. I assigned equal numbers of hybrid and commercial sites to each time of day and randomly

assigned the two research trial sites to morning and midday sampling times. I also randomized

the order in which sites were selected for sampling but this assigned order was frequently altered

to account for pesticide spraying, driving distances between sites, and adverse weather

conditions. Whenever possible, I replaced a site that could not safely or feasibly be sampled with

the next randomly selected site.

3.3.2 Pollinator Sampling

At each study site, I selected a point of reference. For square canola fields, this was

typically located in the field corner most accessible by vehicle (Figure 3.1a), and for circular

fields the point was along the edge of the pivot area adjacent to this corner (Figure 3.1b). This

point served as the central point of landscape analysis and sampling occurred along two 100 m x

3 m transects located within a 0.25 km radius. Transects were oriented parallel with a field edge,

beginning at distances of 25 m and 125 m from this edge, respectively.

35

Figure 3.1: Placement of transects for bee and pollen sampling at a typical square (a) and circular

(b) canola field. Cardinal direction varied. The intersecting grey lines represent roads, with

adjacent vegetated ditches/field margins in white. The star represents the reference point of the

site for landscape analysis, and the circle represents the smallest scale of landscape analysis

(0.25 km), within which both transects were located.

36

Because all bee sampling methods are subject to sampling biases (Grundel et al. 2011;

Westphal et al. 2008), I used both aerial netting and sweep netting on each transect, to gain a

more representative sample of the bee assemblage. To minimize collector biases associated with

netting bees (Westphal et al. 2008), the same collector conducted all samples of a given type. I

conducted aerial netting over 15-min intervals, targeting bees that were visiting flowers. I

identified managed bees in the field without netting, and captured wild bees for later

identification in the lab. I stopped the timer after the capture of a wild bee so that handling time

was not included in the 15-min search time. Sweep netting was conducted by sweeping once

across each meter of transect using a 30 cm diameter sweep net, for a total of 100 sweeps with an

approximately 2 m wide arc. This method captured managed and wild bees in and around the

vegetation, whether they were visiting flowers or not. Any wild bees that escaped during aerial

or sweep netting were recorded and included in measures of abundance but they could not be

identified reliably.

For wild-bee identification, I employed a number of sources (Stephen 1957; Mitchell

1960; Mitchell 1962; Roberts 1973a; Roberts 1973b; Curry 1984; Gibbs 2010; Sheffield et al.

2011; NSERC-CANPOLIN 2012; de Silva and Packer 2012; Discover Life 2012) as well as

experts Jason Gibbs (Cornell University, Ithaca NY), Lincoln Best (York University, Toronto

ON) and Cory Sheffield (Royal Museum of Saskatchewan, Regina SK). When possible, wild

bees were identified to species. However, for some genera species-level keys were not available

for this region. I sorted specimens from these genera into morphospecies using distinguishable

morphological traits, especially those indicated in keys from other regions.

37

3.3.3 Stigma Sampling

I used pollen deposition on stigmas as a direct representation of pollination rather than

seed-based measures which confound the focal effect of pollination with ovule fertilization and

seed development, including potentially yield-limiting factors such as water and nutrient

availability. On each transect, stigmas were harvested every 6.5m, by haphazardly selecting a

canola plant and then removing the stigma from the first open flower above the topmost wilted

flower on a raceme. Care was taken to not transfer pollen to the stigma samples by avoiding

contact with anthers (at commercial sites) and cleaning pollen from forceps between samples. At

hybrid seed sites, stigmas were taken from female plants, as male plants are used only as pollen

donors and not for seed production. This procedure collected 15 canola stigmas per transect, for

a total of 30 stigma samples per study site. However, a number of samples were lost, making the

minimum replication 26 stigmas per site.

I stored stigmas in microcentrifuge tubes with a 70% ethanol solution, and then processed

them in the lab using an acid fuchsin stain before mounting each stigma on a slide (Kearns and

Inouye 1993). I counted only pollen grains that were in direct contact with the stigma, and

excluded those pollen grains that were noticeably misshapen (and therefore likely not viable), or

differed in size and/or stain retention from canola pollen (and therefore likely heterospecific).

For consistency, I counted all stigma samples myself, and randomized the order of stigma sample

processing by transect to minimize the effects of any minor changes in counting over time.

3.3.4 Analysis

I used three site-level measures of pollen deposition: 1) mean pollen deposition per

stigma, 2) the proportion of ‘fully pollinated’ stigmas, based on a threshold of 160 deposited

38

pollen grains, at which all ovules are expected to develop (Mesquida and Renard 1984), and 3)

mean predicted number of seeds per pod, obtained from a type II regression fit to data collected

by Mesquida and Renard (1984; Y = 0.28x – 4.56 , r2 = 0.45, t30 = 4.95, P < 0.0001), and

truncated at a maximum of 30 seeds, corresponding to the approximate number of ovules

(Mesquida and Renard 1984).

I used a two-way ANOVA followed by a Tukey HSD test to compare managed and wild-

bee abundances between canola types (commercial vs. hybrid seed) and t-tests to compare the

total number of bees between canola types as well as measures of pollen deposition. The two

research trial sites were excluded here so hybrid seed canola values represented production fields

only. I used Brown-Forsythe tests to confirm that commercial and hybrid seed fields were of

equivalent variance for each measure of pollen deposition.

To determine the effects of wild and managed-bee abundances on each measure of pollen

deposition, I used ANCOVA that included canola type as a fixed factor (here research trial and

hybrid seed sites were both classified as hybrid seed canola). I similarly used ANCOVA to

determine the effect of the proportion of semi-natural land on wild-bee abundance, managed-bee

abundance, and each measure of pollen deposition, across a range of landscape scales, again

accounting for canola type. The effect of semi-natural land on pollen deposition was expected to

act through an effect on wild bee abundance, therefore bee abundances and semi-natural land

were tested in separate ANCOVAs.

All t-tests and least squares models used JMP software (version 10.0.0, SAS institute

Inc., Cary NC). I performed a Mantel test on the residuals from all ANOVA and ANCOVAs

using package ade4 in R software (version 2.13.2, The R Foundation for Statistical Computing,

Vienna, Austria). None showed significant spatial autocorrelation. I also tested variance inflation

39

factors to address potential concerns about collinearity among variables. All were < 2. I applied

Box-Cox transformations to variables as needed to meet assumptions of normality and

homoscedasticity in the residuals.

3.4 Results

In total, I recorded 3879 bees visiting canola. Most (97%) were managed bees, consisting

of 75% A. mellifera and 25% M. rotundata individuals. Wild bees comprised only 3% of bees

found in canola fields. These 109 wild specimens represented 18 species and at least 6

morphospecies (a conservative estimate assuming all female morphospecies correspond directly

to male morphospecies of the same genus), from 8 genera (Appendix 3). Bombus, Andrena, and

Lasioglossum were the most abundant wild-bee genera, comprising 71% of total wild-bee

abundance.

Based on Tukey HSD tests applied to the significant interaction term (F1,60=15.36,

P=0.004) of the 2-way ANOVA with fixed effects of canola type and bee type, commercial

canola fields had significantly fewer managed bees than hybrid seed canola fields, but the

number of wild bees did not differ between the two canola types (Figure 3.2). Commercial fields

also had fewer total number bees than hybrid seed fields (t28 = 4.49, P ≤ 0.0001). Despite lower

bee abundance, fields of commercial canola had significantly greater mean pollen deposition (t28

= -7.39, P < 0.0001), proportion of stigmas fully pollinated (t28 = -9.27, P < 0.0001) and mean

predicted seeds per pod (t28 = -6.61, P < 0.0001), than fields of hybrid seed canola (Figure 3.3).

There was a significant positive effect of managed-bee abundance on mean pollen deposition (t28

= 2.60, P = 0.015), the proportion of stigmas fully pollinated (t28 = 2.87, P = 0.008), and the

predicted number of seeds per pod (t28 = 2.45, P = 0.021), but no effect of wild-bee abundance

40

Figure 3.2: Abundance of wild and managed bees in commercial canola fields (dark bars) and

hybrid seed canola fields (light bars), sampled by aerial and sweep netting over two 100 m x 3 m

transects. n = 15. Bars represent the standard error of the mean. There was a significant

interaction between canola type and bee type (t61 = -2.96, P = 0.004).

41

Figure 3.3: Comparison of three measures of pollen deposition for commercial and hybrid seed

canola fields. Each point indicates the mean of 15 sites, with 26-30 replicates each. Error bars

represent one SE.

42

on these measures (t28 = 0.10, P = 0.92; t28 = 0.54, P = 0.60; t28 = -0.12, P = 0.91 respectively;

Figure 3.4).

The proportion of semi-natural land within a 3 km radius of the study sites ranged from

<1% to as much as 81%. The three measures of pollen deposition showed similar trends in their

response to landscape scale; however, semi-natural land at a 1.25 km radius was most strongly

associated with mean pollen deposition and the proportion of stigmas fully pollinated, whereas a

0.25 km radius was most strongly associated with the mean predicted number of seeds per pod

(Figure 3.5). At these respective scales, there was a significant positive association between

semi-natural land and the former two measures of pollen deposition (t29 = 2.26, P = 0.031; t29 =

2.35, P = 0.026), but this association was only marginally significant for the mean predicted

number of seeds per pod (t29 = 1.70, P = 0.099; Figures 3.4).

Wild-bee abundance was more responsive to landscape scale than was managed-bee

abundance, and was most strongly associated with a radius of 0.5 km, compared to 1.25 km for

managed bees (Figure 3.5). This positive association was significant for wild-bee abundance (t29

= 2.33, P = 0.027), but not managed-bee abundance (t29 = 1.05, P= 0.30; Figure 3.4). None of the

predictor variables in the ANCOVA models interacted with canola type, therefore I excluded

these interactions from the model statistics presented here.

43

Figure 3.4: Effect of bee abundance and the proportion of semi-natural land in the landscape

three measures of pollen deposition in canola. Solid lines indicate significant relation and dashed

lines indicate marginally significant relation. Two separate ANCOVA models were run for each

measure of pollen deposition, one including both types of bee abundance (left 2 columns), and

the other with the proportion of semi-natural land (right column). Inferential statistics for these

figures are reported in the Results section.

44

Figure 3.5: Relations of a) bee abundance and b) pollen deposition to the proportion of semi-

natural land at a range of 12 landscape radii from 0.25 km to 3 km, controlling for canola type

(commercial or hybrid seed). In Figure a) solid circles correspond to managed-bee abundance

and open circles to wild-bee abundance. In Figure b) solid triangles correspond to mean pollen

deposition, open triangles to the predicted number of seeds per pod, and inverted triangles to the

proportion of stigmas fully pollinated (160 pollen grains or more). n = 32 for each ANCOVA.

45

3.5 Discussion

In this study I found evidence that semi-natural land enhances wild pollinator abundance

in canola crops and crop pollination. This is consistent with other findings (Morandin and

Winston 2006; Morandin et al. 2007), but also extends the scope of these benefits to include

intensified agricultural landscapes where managed pollinators greatly outnumber wild bees. In

contrast, findings by Bommarco et al. (2010), suggested that the abundance of wild bees foraging

in canola was not associated with landscape in this context. Although I did not detect a direct

association of canola pollination to wild-bee abundance, this has been established elsewhere

(Morandin and Winston 2005; Morandin and Winston 2006). Further, in this study, only wild-

bee abundance, and not managed-bee abundance, correlated positively with the amount of

adjacent semi-natural land, discounting the possibility that landscape benefits to crop pollination

in some way acted through effects on managed bees.

I can conceive of no other plausible mechanism that would lead semi-natural land to

benefit canola pollen deposition, other than by enhancing the abundance of wild pollinators

visiting the crop. This relation may be mediated at least partially through effects of semi-natural

land on the abundance of non-bee wild pollinators of canola. Lepidoptera were very rarely seen

visiting canola flowers, but syrphid flies (Diptera) appeared to be abundant in some fields. As

with wild bees, the abundance and diversity of syrphid flies often varies positively with the

amount of adjacent semi-natural land (Hendrickx et al. 2007; Kohler et al. 2008; Bommarco et

al. 2012); however, syrphids are less effective than bees for the pollination of canola and other

Brassicaceae (Sahli and Conner 2007; Jauker et al. 2012). Indeed, increased syrphid density can

reduce pollinator efficiency (Jauker and Wolters 2008), presumably because they act as pollen

thieves, making wild bees somewhat more effective agents of pollen deposition.

46

Overall, wild-bee abundance in the canola crop varied most strongly with the proportion

of semi-natural land in the landscape within 0.5 km, which is comparable to findings of

Morandin and Winston (2006), who determined that 0.75 km was the most influential scale.

Because the foraging range of wild bees varies strongly with body size (Gathman and Tscharntke

2002; Greenleaf et al. 2007), this likely reflects an intermediate scale between that most

important for the small-bodied bees [e.g. Lasioglossum (Dialictus)] and large-bodied bees (e.g.

Bombus) I found visiting canola. Small-bodied bees are expected to have smaller foraging

ranges, and respond to smaller landscape scales compared to bumble bees, which respond at

larger scales of up to 3 km (Steffan-Dewenter et al. 2001; Hines and Hendrix 2005; Westphal et

al. 2006; Knight et al. 2009).

Body size is also positively related to pollen deposition and resulting seed set (Hoehn et

al. 2008; Sahli and Conner 2007). Pollen deposition by bumble bees can average as much as 547

grains on the stigma of a previously unvisited commercial canola flower (Cresswell 1999),

considerably exceeding the 160 grains required for full pollination (Mesquida and Renard 1984).

Larger bees are therefore likely to have a greater effect on mean pollen deposition than smaller

bees, even if a certain amount of this pollen does not translate into increased seed production.

This could explain why my results indicated that mean pollen deposition and the proportion of

stigmas fully pollinated responded to semi-natural land at a larger landscape scale than the

predicted number of seeds per pod, despite these three measures showing similar overall trends

in their response to landscape scale. Larger bees, responding to the landscape at larger scales and

depositing more pollen in a single visit, may have relatively more impact on the former two

measures compared to the latter. In contrast, smaller bees, responding to the landscape more

47

locally and depositing less pollen, are less likely to meet or exceed the threshold of ‘full

pollination’, and will therefore affect the three measures of pollen deposition more equally.

On average, both hybrid seed and commercial canola fields showed a substantial

pollination deficit, indicating that the benefits of insect pollination had not saturated, despite

prolific managed pollinators. Hybrid seed canola fields exhibited an especially low level of

pollination, with only about 9% of stigmas achieving full pollination. Pollen deposition was

lower in hybrid seed canola fields than in commercial fields despite the presence of more

managed bees, and more bees overall, reflecting the difference in availability of pollen between

the two canola types. In commercial canola, all flowers on all plants produce pollen. Without any

bee activity, some pollen can be transferred to a stigma directly from the anthers of the same

flower, or from neighbouring plants through shaking (Free 1993). In contrast, in hybrid seed

fields, pollen is available only on the anthers of ‘male’ plants, which are intercropped with the

‘female’ plants from which I harvested stigmas. No self-pollen transfer can occur, and while a

certain amount of pollen may be transferred from shaking, not all neighbouring plants are pollen-

providing ‘males’. Further, bees deposit more pollen when visiting commercial canola flowers

than hybrid seed canola flowers, because they can transfer self-pollen as well as cross-pollen

(Cresswell 1999). Bees can also be expected to carry more cross pollen in commercial fields than

hybrid seed fields, since all flowers they visit in these fields are pollen-producing.

Pollen deposition was chosen here to measure pollination because it is largely

independent of other potentially yield-limiting factors such as water, sunlight, and nutrient

availability, and was therefore expected to more accurately represent the contributions of

pollinators. At the same time, pollen deposition should not be equated with yield, because these

other growing conditions may determine both the initial number of flowers/pods as well as

48

whether all fertilized seeds develop and to what size (Tayo and Morgan 1979; Williams et al.

1987; Bos et al. 2007). This may explain why some studies do not find a positive effect of bee

abundance on commercial canola yield (Free and Nuttall 1968; Williams et al. 1987; Mesquida

et al. 1988; Steffan-Dewenter 2003). Ultimately, the extent to which pollen quality and quantity

are limiting compared to other growing conditions may determine the extent to which bee

pollination is beneficial to a given field.

Nonetheless, the relation between pollen deposition and the proportion of semi-natural

land in the landscape is noteworthy. Managed pollinators present a substantial cost for canola

hybrid seed producers. Each honey-bee colony is rented seasonally for about $150 (Canadian

Honey Council 2012), for an estimated total cost of $12,000,000 annually in Canada, not

including the cost of managed leafcutter bees. If wild pollinators can provide significant

supplemental pollination services to the production of hybrid canola seed, and enhance yields in

commercial canola as well, producers may find it worthwhile to establish or conserve tracts of

semi-natural land on their farms, with potential conservation benefits to wild bees and other taxa.

49

4: CONCLUDING DISCUSSION

My results support a growing body of literature on the benefits of semi-natural land in

agricultural areas for wild bees and the crop pollination services they provide. I found that semi-

natural land promoted overall wild-bee abundance and diversity, as well as the number of wild

bees foraging in canola crops. Mean pollen deposition on canola stigmas and the proportion of

canola flowers achieving full pollination services were also positively associated with semi-

natural land in the landscape. These results also suggest that the relationship between semi-

natural land and wild bees is robust, because it is evident even in the highly modified agricultural

landscapes of southern Alberta. Here, tracts of semi-natural land contained few native flowering

plants and often bore little resemblance to original landcover types, instead they generally took

the form of mowed ditches and tame pastures, yet this uncultivated land still appeared to provide

valuable habitat for wild bees.

Consistent evidence for the ability of semi-natural land to enhance wild-bee populations

in adjacent crops has led some to promote integrated agricultural land-use, which creates a

mosaic of cultivated fields and semi-natural areas such as livestock pastures, as a means to

leverage these natural pollination services for mass-flowering crops such as canola (Kremen et

al. 2004; Morandin et al. 2007; Ricketts et al. 2008). However, this study presents a novel

concern regarding this practice. When mass-flowering crops employ managed pollinators, as in

the case of hybrid seed canola, integrated land use may actually be more detrimental to wild bees

than the production of such crops in more homogenous landscapes dominated by cultivated

crops. I found that introductions of managed bees for canola pollination reduced wild-bee

abundance more where surrounding landscapes contained more semi-natural land.

50

So while leveraging ecosystem services from semi-natural land for the pollination of

crops has previously been largely conceived of as a “win-win situation”, resulting in benefits to

wild bees and agricultural yields alike, the case of mass-flowering crops that employ managed

pollinator may in fact present a conflict. Pollination services for these crops are still enhanced by

the presence of additional wild pollinators, but because managed pollinators are not restricted to

foraging within the crop, they flood into semi-natural areas where their resource use overlaps

considerably with the wild-bee assemblage and leads to reductions in wild-bee abundance that

exceed those for more intensified landscapes.

From a conservation point of view, managed bees, and crops dependent on them, may

therefore be best situated in more intensified landscapes and not in fields adjacent to semi-natural

land. Alternatively, detrimental impacts on wild bees may be limited by encouraging the use of

less harmful managed pollinators for use in these locations. Managed pollinators with smaller

foraging ranges and lower levels of resource overlap with the wild bee assemblage could be

expected to cause less competitive impacts on wild bees. For hybrid seed canola, this would

mean using only M. rotundata, rather than a combination of M. rotundata and A. mellifera. M.

rotundata were found to have 75% less floral resources overlap with wild bees in canola field

edge habitats than A. mellifera, and were rarely seen away from hybrid seed canola fields.

The relative impact of different managed pollinators is worth exploring further, but this

could be a difficult endeavor since evidence for detrimental effects of managed pollinators on

wild bees has been historically difficult to find. Future studies in this field should focus on those

strategies that have proved successful in this regard. Here, the use of temporal contrasts, using

wild-bee abundances sampled before, during, and after the introduction of large numbers of

managed pollinators, proved to be more powerful than the correlation of raw abundances. My

51

study also indicates that landscape context may confound and interact with the effects of

managed pollinators and should be controlled for, or otherwise taken into account. More

generally, I found evidence that the availability of semi-natural land, and the presence of

managed pollinators act concurrently to structure the wild-bee assemblage in agricultural areas.

Both of these drivers should be considered in conservation initiatives for wild bees, as well as

attempts to leverage their crop pollination services.

52

LITERATURE CITED

Aizen, M. A. and L. D. Harder. 2009. The global stock of domesticated honey bees is growing

slower than agricultural demand from pollination. Current Biology 19:915-918.

Artz, D., Hsu, C. and B. Nault. 2011. Influence of honey bee, Apis mellifera, hives and field size

on foraging activity of native bee species in pumpkin fields. Environmental Entomology

40:1144-1158.

Beekman, M. and F. L. W. Ratnieks. 2000. Long-range foraging by the honey-bee, Apis

mellifera L. Functional Ecology 14:490-496.

Benjamini, Y. and Y. Hochberg. 1995. Controlling the false discovery rate: a practical and

powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B

(Methodological) 57:289-300.

Benjamini, Y. and Y. Hochberg. 2000. On the adaptive control of the false discovery rate in

multiple testing with independent statistics. Journal of Educational and Behavioural

Statistics 25:60-83.

Bommarco, R., Marini, L. and B. Vaissière. 2012. Insect pollination enhances seed yield, quality

and market value in oilseed rape. Oecologia 169:1025-1032.

Bos, M. M., Veddler, D., Bogdanski, A. K., Klein, A.-M., Tscharntke, T., Steffan-Dewenter, I.

and J. M. Tylianakis. 2007. Caveats to quantifying ecosystem services: fruit abortion

blurs benefits from crop pollination. Ecological Applications 17:1841-49.

Breeze, T. D., Bailey, A. P., Balcombe, K. G. and S. G. Potts. 2011. Pollination services in the

UK: How important are honeybees? Agriculture, Ecosystems and Environment 142:137-

143.

53

Butz Huryn, V. M. 1997. Ecological impacts of introduced honey bees. The Quarterly Review of

Biology 72:275-297.

Canadian Honey Council. 2012. Industry Overview; www.honeycouncil.ca. Accessed June 21,

2009.

Canola Council of Canada. 2011. Canola Grower’s Manual: Canola Varieties; www.canola-

council.org. Accessed August 20, 2012.

Carré, G., Roche, P., Chifflet, R., Morison, N., Bommarco, R., Harrison-Cripps, J. and K.

Krewenka. 2009. Landscape context and habitat type as drivers of bee diversity in

European annual crops. Agriculture, Ecosystems and Environment 133:40-47.

Carvalheiro, L. G., Seymour, C. L., Veldtman, R. and S. W. Nicolson. 2010. Pollination services

decline with distance from natural habitat even in biodiversity-rich areas. Journal of

Applied Ecology 47:810-820.

Casséus, L. 2009. Canola: a Canadian Success Story. Canadian Agriculture at a Glance.

Statistics Canada. Catalogue no. 96-325-X.

Clay, H. 2009. Pollinating Hybrid Canola-the Southern Alberta Experience. Hive Lights. pp. 14-

16. Canadian Honey Council, Calgary.

Crane, E. 1990. Bees and Beekeeping: Science, Practice and World Resources. Cornell Univ.

Press/Cornstock, Ithaca.

Cresswell, J. E. 1999. The influence of nectar and pollen availability on pollen transfer by

individual flowers of oil-seed rape (Brassica napus) when pollinated by bumblebees

(Bombus lapidarius). Journal of Ecology 87:670-677.

Currie, R. W., Pernal, S. F. and E. Guzmán-Novoa. 2010. Honey bee colony losses in Canada.

Journal of Apicultural Research 49:104-106.

54

Curry, P. S. 1984. Bumble Bees of Saskatchewan (Hymenoptera: Apidae): A Survey of Their

Geographic Distribution. Natural History Contributions No. 5. Museum of Natural

History, Regina.

Delaplane, K. S. and D. F. Mayer. 2000. Canola Seed (Oilseed Rape). Crop Pollination by Bees.

CABI Publishing, Wallingford.

Didham, R. K., Tylianakis, J. M., Gemmell, N. J., Rand, T. A. and R. M. Ewers. 2007.

Interactive effects of habitat modification and species invasion on native species decline.

TRENDS in Ecology and Evolution 22:489-496.

Discover Life. Bee Genera Identification Guides. www.discoverlife.org. Accessed September

10th

, 2011 – March 1st, 2012.

Donovan, B. J. 1980. Interactions between native and introduced bees in New Zealand. New

Zealand Journal of Ecology 3:104-116.

Free, J. B. 1993. Insect Pollination of Crops. Acadmeic Press, San Diego.

Free, J. B. and P. M. Nuttall. 1968. The pollination of oilseed rape (Brassica napus) and the

behaviour of bees on the crop. Journal of Agricultural Science, Cambridge 71:91-94.

Garibaldi, L. A., Steffan-Dewenter, I., Kremen, C., Morales, J. M., Bommarco, R., Cunningham,

S. A., Carvalheiro, L. G., Chacoff, N. P., Dudenhöffer, J. H., Greenleaf, S. S. et al. 2011.

Stability of pollination services decreases with isolation from natural areas despite honey

bee vists. Ecology Letters 14:1062-1072.

Gathmann, A. and T. Tscharntke. 2002. Foraging ranges of solitary bees. Journal of Animal

Ecology 71:757-764.

Gibbs, J. 2010. Revision of the metallic species of Lasioglossum (Dialictus) in Canada

(Hymenoptera, Halictidae, Halictini). Zootaxa 2591:1-382.

55

Ginsberg, H. S. 1983. Foraging ecology of bees in an old field. Ecology 64:165-175.

Goulson, D. 2003. Effects of introduced bees on native ecosystems. Annual Review of Ecology,

Evolution and Systematics 34:1-26.

Goulson, D., Lye, G. C. and B. Darvill. 2008. Decline and conservation of bumble bees. Annual

Review of Entomology 53:191-208.

Goulson, D. and K. Sparrow. 2009. Evidence for competition between honeybees and

bumblebees; effects on bumblebee worker size. Journal of Insect Conservation 13:177-

181.

Greenleaf, S. S. and C. Kremen. 2006. Wild bee species increase tomato production and respond

differently to surrounding land use in Northern California. Biological Conservation

133:81-87.

Greenleaf, S. S., Williams, N. M., Winfree, R. and C. Kremen. 2007. Bee foraging ranges and

their relationship to body size. Oecologia 153:589-596.

Gross, C. L. and D. Mackay. 1998. Honeybees reduce fitness in the pioneer shrub Melastoma

affine (Melastomataceae). Biological Conservation 86:169-178.

Grundel, R., Frohnapple, K. J., Jean, R. P. and N. B. Pavlovic. 2011. Effectiveness of bowl

trapping and netting for inventory of a bee community. Environmental Entomology

40:374-380.

Hayter, K. E. and J. E. Cresswell. 2006. The influence of pollinator abundance on the dynamics

and efficiency of pollination in agricultural Brassica napus: implications for landscape-

scale gene dispersal. Journal of Applied Ecology 43:1196-1202.

Heithaus, E. R. 1979. Community structure of neotropical flower visiting bees and wasps:

diversity and phenology. Journal of Ecology 60:190-202.

56

Hendrickx, F., Maelfait, J.-P., Van Wingerden, W., Schweiger, O., Speelmans, M., Aviron, S.,

Augenstein, I., Billeter, R., Bailey, D., Bukacek, R. et al. 2007. How landscape structure,

land-use intensity and habitat diversity affect components of total arthropod diversity in

agricultural landscapes. Journal of Applied Ecology 44:340-351.

Hines, H. M. and S. D. Hendrix. 2005. Bumble bee (Hymenoptera: Apidae) diversity and

abundance in tallgrass prairie patches: effects of local and landscape floral resources.

Environmental Entomology 34:1477-1484.

Hoehn, P., Tscharntke, T., Tylianakis, J. M. and I. Steffan-Dewenter. 2008. Functional group

diversity of bee pollinators increases crop yield. Proceedings of the Royal Society B

275:2283-2291.

Hurlbert, S. H. 1978. The measurement of niche overlap and some relatives. Ecology

59:67-77.

Isaacs, R. and A. K. Kirk. 2010. Pollination services provided to small and large highbush

blueberry fields by wild and managed bees. Journal of Applied Ecology 47:841-849.

Jauker, F., Bondarenko, B., Beckert, H. C. and I. Steffan-Dewenter. 2012. Pollination efficiency

of wild bees and hoverflies provided to oilseed rape. Agricultural and Forest Entomology

14:81-87.

Jauker, F. and V. Wolters. 2008. Hover flies are efficient pollinators of oilseed rape. Oecologia

156:819-823.

Kearns, C. A. and D. W. Inouye. 1993. Techniques for Pollination Biologists. University Press of

Colorado, Niwot.

57

Kim, J., Williams, N. and C. Kremen. 2006. Effects of cultivation and proximity to natural

habitat on ground-nesting native bees in California sunflower fields. Journal of the

Kansas Entomological Society 79:309-320.

Klein, A.-M., Brittain, C., Hendrix, S. D., Thorp, R., Williams, N. and C. Kremen. 2012. Wild

pollination services to California almond rely on semi-natural habitat. Journal of Applied

Ecology 49:723-732.

Knight, M. E., Osborne, J. L., Sanderson, R. A., Hale, R. J., Martin, A. P. and D. Goulson. 2009.

Bumblebee nest density and the scale of available forage in arable landscapes. Insect

Conservation and Diversity 2:116-124.

Kohler, F., Verhulst, J., van Klink, R. and D. Kleijn. 2008. At what spatial scale do high-quality

habitats enhance the diversity of forbs and pollinators in intensively farmed landscapes?

Journal of Applied Ecology 45:753-762.

Kosinski, S. 2012. The Value of Alberta’s Forage Industry: A Multi-Level Analysis, pp 10.

Alberta Agriculture and Rural Development.

Kremen, C., Williams, N. M., Bugg, R. L., Fay, J. P. and R. W. Thorp. 2004. The area

requirements of an ecosystem service: crop pollination by native bee communities in

California. Ecology Letters 7:1109-1119.

Kremen, C., Williams, N. W. and R. W. Thorp. 2002. Crop pollination from native bees at risk

from agricultural intensification. Proceedings of the National Academy of Sciences

99:16812-16816.

Lande, R. 1996. Statistics and partitioning of species diversity, and similarity among multiple

communities. Oikos 76:5-13.

58

Lande, R., DeVries, P. J. and T. R. Walla. 2000. When species accumulation curves intersect:

implications for ranking diversity using small samples. Oikos 89:601-605.

Le Féon, V., Burel, F., Chifflet, R., Henry, M., Ricroch, A., Vaissière, B. E. and J. Baudry. 2011.

Solitary bee abundance and species richness in dynamic agricultural landscapes.

Agriculture, Ecosystems and Environment, In Press.

Mesquida, J. and M. Renard. 1984. Etude des quantités de pollen deposés sur les stigmates dans

différentes conditions de pollinisation; influence sur la production de graines chez le

colza d’hiver mâle-fertile. Vème Symposium International sur la Pollinisation. Versailles,

27–30 septembre 1983. pp. 351-356. INRA Publishing, Le Rheu.

Mesquida, J., Renard, M. and J-S. Pierre. 1988. Rapeseed (Brassica napus L.) productivity: the

effect of honeybees (Apis mellifera L.) and different pollination conditions in cage and

field tests. Apidologie 19:51-72.

Mitchell, T. B. 1960. Bees of the Eastern United States. I. Family Andrenidae; Colletidae;

Halictidae. Technical Bulletin 141. pp. 23-521. North Carolina Agricultural Experiment

Station, Raleigh.

Mitchell, T. B. 1962. Bees of the Eastern United States II. Family Megachilidae; Apidae.

Technical Bulletin 152. pp. 5-232; 513-546. North Carolina Agricultural Experiment

Station, Raleigh.

Morandin, L. A. and M. L. Winston. 2005. Wild bee abundance and seed production in

conventional, organic, and genetically modified canola. Ecological Applications 15:871-

881.

Morandin, L. A. and M. L. Winston. 2006. Pollinators provide economic incentive to preserve

natural land in ecosystems. Agriculture, Ecosystems and Environment 116:289-292.

59

Morandin, L. A., Winston, M. L., Abbott, V. A. and M. T. Franklin. 2007. Can pastureland

increase wild bee abundance in agriculturally intense areas? Basic and Applied Ecology

8:117-124.

NSERC-CANPOLIN. 2012. Bee Course 2012: Key to Bee Genera in Canada. Unpublished.

Oertli, S., Müller, A. and S. Dorn. 2005. Ecological and seasonal patterns in the diversity of a

species-rich bee assemblage (Hymenoptera: Apoidea: Apiformes). European Journal of

Entomology 102:53-63.

Paini, D. 2004. Impact of the introduced honey bee (Apis mellifera) (Hymenoptera: Apidae) on

native bees: a review. Austral Ecology 29:399-407.

Paini, D. and J. Roberts. 2005. Commercial honey bees (Apis mellifera) reduce the fecundity of

an Australian native bee (Hylaeus alcyoneus). Biological Conservation 123:103-112.

Potts, S. G., Biesmeijer, J. C., Kremen, C., Neumann, P., Schweiger, O. and W. E. Kunin. 2010.

Global pollinator declines: trends, impacts and drivers. Trends in Ecology and Evolution

25:345-353.

Potts, S. G., Vulliamy, B., Dafni, A., Ne’eman, G. and P. Willmer. 2003. Linking bees and

flowers: how do floral communities structure pollinator communities? Ecology 84:2628-

2642.

Pyke, G. H. 1982. Local geographic distributions of bumblebees near Crested Butte, Colorado:

competition and community structure. Ecology 63:555-573.

Rader, R., Howlett, B. G., Cunningham, S. A., Westcott, D. A. and W. Edwards. 2012. Spatial

and temporal variation in pollinator effectiveness: do unmanaged insects provide

consistent pollination services to mass flowering crops? Journal of Applied Ecology

49:126-134.

60

Richards, K. W. and P. G. Kevan. 2002. Aspects of bee biodiversity, crop pollination, and

conservation in Canada. Pollinating Bees: The Conservation Link Between Agriculture

and Nature (eds P. G. Kevan and V. L. Imperatriz Fonseca). pp. 77-94. Ministry of

Environment, Brasilia.

Ricketts, T. H. 2004. Tropical forest fragments enhance pollinator activity in nearby coffee

crops. Conservation Biology 18:1262-1271.

Ricketts, T. H., Regetz, J., Steffan-Dewenter, I., Cunningham S. A., Kremen, C., Bogdanski, A.,

Gemmill-Herren, B., Greenleaf, S. S., Klein A. M., Mayfield, M. M. et al. 2008.

Landscape effects on crop pollination services: are there general patterns? Ecology

Letters 11:499-515.

Roberts, R. B. 1973a. Bees of Northwestern America: Agapostemon (Hymenoptera: Halictidae).

Technical Bulletin 125. pp. 4-5. Oregon State University, Corvallis.

Roberts, R. B. 1973b. Bees of Northwestern America: Halictus (Hymenoptera: Halictidae).

Technical Bulletin 126. pp. 5-7. Oregon State University, Corvallis.

Roubik, D. W., Moreno, J. E., Vergara, C. and D. Wittman. 1986. Sporadic food competition

with the African honey bee: projected impact on neotropical social bees. Journal of

Tropical Ecology 2:97-111.

Roubik, D. W. and H. Wolda. 2001. Do competing honey bees matter? Dynamics and abundance

of native bees before and after honey bee invasion. Population Ecology 43:53-62.

Roulston, T. H. and K. Goodell. 2011. The role of resources and risks in regulating wild bee

populations. Annual Review of Entomology 56:293-312.

61

Sabbahi, R., de Oliviera, D. and J. Marceau. 2005. Influence of honey bee (Hymenoptera:

Apidae) density on the production of canola (Crucifera: Brassicacae). Journal of

Economic Entomology 98:367-372.

Sahli, H. F. and J. K. Conner. 2007. Visitation, effectiveness, and efficiency of 15 genera of

visitors to wild radish, Raphanus raphanistrum (Brassicaceae). American Journal of

Botany 94:203-209.

Sheffield, C. S., Ratti, C., Packer, L. and T. Griswold. 2011. Leafcutter and mason bees of the

genus Megachile Latreille (Hymenoptera: Megachilidae) in Canada and Alaska.

Canadian Journal of Arthropod Identification 18:1-107.

de Silva, N. and L. Packer. 2012. A revision of the cleptoparasitic bee genus Coelioxys

(Hymenoptera: Megachilidae) in Canada. Unpublished.

Soroka, J. J., Goerzen, D. W., Falk, K. C. and K. E. Bett. 2001. Alfalfa leafcutting bee

(Hymenoptera: Megachilidae) pollination of oilseed rape (Brassica napus L.) under

isolation tents for hybrid seed production. Canadian Journal of Plant Science 81:199-204.

Standifer, L. N. 1980. Honey Bee Nutrition and Supplemental Feeding. pp. 39-45. Beekeeping in

the United States. Agriculture Handbook Number 33. United States Department of

Agriculture.

Statistics Canada. 2012. 2011 Census of Agriculture. Farm and Farm Operator Data.

www.statcan.gc.ca/ca-ra2011/index-eng.htm. Accessed September 21, 2012.

Steffan-Dewenter, I. 2003. Seed set of male-sterile and male-fertile oilseed rape (Brassica

napus) in relation to pollinator density. Apidologie 34:227-235.

Steffan-Dewenter, I., Münzenberg, U., Bürger, C., Thies, C. and T. Tscharntke. 2002. Scale-

dependent effects of landscape context on three pollinator guilds. Ecology 83:1421-1432.

62

Steffan-Dewenter, I., Münzenberg, U., and T. Tscharntke. 2001. Pollination, seed set and seed

predation on a landscape scale. Proceedings of the Royal Society B 268:1685-1690.

Steffan-Dewenter, I. and T. Tscharntke. 2000. Resource overlap and possible competition

between honey bees and wild bees in central Europe. Oecologia 122:288-296.

Steffan-Dewenter, I. and C. Westphal. 2008. The interplay of pollinator diversity, pollination

services and landscape change. Journal of Applied Ecology 45:737-741.

Stephen, W. P. 1957. Bumble Bees of Western America (Hymenoptera: Apoidea). Technical

Bulletin 40. Agriculture Experiment Station. Oregon State College, Corvallis.

Tayo, T. O. and D. G. Morgan. 1979. Factors influencing flower and pod development in oil-

seed rape (Brassica napus L.) The Journal of Agricultural Science 92:363-373.

Thomson, D. 2004. Competitive interactions between the invasive European honey bee and

native bumble bees. Ecology 85:458-470.

Thomson, D. 2006. Detecting the effects of introduced species: a case study of competition

between Apis and Bombus. Oikos 114:407-418.

Thorp, R. W., Wenner, A. M. and J. F. Barthell. 1994. Flowers visited by honey bees and native

bees on Santa Cruz Island. The Fourth California Islands Symposium: Update on the

Status of Resources (eds W.L. Halvorson & G.J. Maender). pp. 351-365. Santa Barbara

Museum of Natural History, Santa Barbara.

Tscharntke, T. and R. Brandl. 2004. Plant-insect interactions in fragmented landscapes. Annual

Review of Entomology 49:405-430.

Tscharntke, T., Klein, A. M., Kreuss, A., Steffan-Dewenter, I. and C. Thies. 2005. Landscape

perspectives on agricultural intensification and biodiversity – ecosystem service

management. Ecology Letters 8:857-874.

63

Van Geert, A., Van Rossum, F. and L. Triest. 2010. Do linear landscape elements in farmland act

as biological corridors for pollen dispersal? Journal of Ecology 98:178-187.

Vergara, C. H. and E. I. Badano. 2009. Pollinator diversity increases fruit production in Mexican

coffee plantations: the importance of rustic management systems. Agriculture,

Ecosystems and Environment 129:117-123.

Walther-Hellwig, K., and R. Frankl. 2000. Foraging habitats and foraging distances of

bumblebees, Bombus spp. (Hym., Apidae), in an agricultural landscape. Journal of

Applied Entomology 124:299-306.

Westphal, C., Bommarco, R., Carré, G., Lamborn, E., Morison, N., Petanidou, T., Potts, S. G.,

Roberts, S. P. M., Szentgyörgyi, H., Tscheulin, T. et al. 2008. Measuring bee diversity in

different European habitats and biogeographical regions. Ecological Monographs 78:653-

671.

Westphal, C., Steffan-Dewenter, I. and T. Tscharntke. 2006. Bumblebees experience landscapes

at different spatial scales: possible implications for coexistence. Oecologia 149:289-300.

Williams, I. H., Martin, A. P. and R. P. White. 1987. The effect of insect pollination on plant

development and seed production in winter oil-seed rape (Brassica napus L.). Journal of

Agricultural Science, Cambridge 109:135-139.

Wilms, W. and B. Wiechers. 1997. Floral resource partitioning between native Melipona bees

and the introduced Africanized honey bee in the Brazilian Atlantic rain forest. Apidologie

28:339-355

64

APPENDICES

Appendix 1: Wild bee species/morphospecies caught through aerial and sweep netting in canola

fields and field edges at 32 sites. When numbers in brackets are accompanied by a “+”, this

indicates a conservative estimate of the total number of morphospecies for a given genus that

assumes all female morphospecies correspond directly to male morphospecies.

Family Genus, species Abundance

Andrenidae 245

Andrena 230

accepta 1

peckhami 5

prunorum 23

morphospecies (10+) 200

Panurginus 1

polytrichus 1

Perdita 3

morphospecies (2) 3

Pseudopanurgus 12

morphospecies (2+) 12

Apidae 391

Bombus 320

borealis 23

centralis 21

fervidus 6

huntii 66

insularis 23

nevadensis 15

occidentalis 5

rufocinctus 136

ternarius 25

Ceratina 1

nanula 1

Diadasia 4

diminuta 2

rinconis 2

Epeolus 6

morphospecies (2+) 6

Eucera 2

morphospecies (1) 2

65

Family Genus, species Abundance

Melissodes 33

morphospecies (6+) 33

Nomada 23

morphospecies (5+) 23

Triepeolus 2

balteatus 1

morphospecies (1) 1

Colletidae 44

Colletes 31

morphospecies (3+) 31

Hylaeus 13

morphospecies (2+) 13

Halictidae 686

Agapostemon 27

angelicus 19

virescens 8

Dufourea 53

marginata 53

Halictus 98

confusus 44

ligatus 20

rubicundus 34

Lasioglossum 499

aberrans 1

aff. lineatulum 3

albipenne 120

albohirtum 18

cooleyi 2

glabriventre 2

laevissimum 6

leucozonium 53

novascotiae 4

occidentale 25

parforbesii 12

pectorale 3

perpunctatum 7

pictum 4

prasinogaster 47

pruinosum 26

pulveris 4

ruidosense 88

sagax 18

66

Family Genus, species Abundance

semicaeruleum 46

succipenne 2

zephyrum 4

zonulum 4

Sphecodes 9

morphospecies (4+) 9

Megachilidae 30

Atoposmia 2

mophospecies (1) 2

Calliopsis 1

morphospecies (1) 1

Coelioxys 3

rufitarsis 3

Hoplitis 3

pilosifrons 2

morphospecies (1) 1

Megachile 15

centuncularis 1

frigida 2

melanophae 2

perihirta 10

Osmia 4

morphospecies (4) 4

Stelis 2

labiata 1

lateralis 1

67

Appendix 2: Flower species recorded on field edges transects (total length 38 800 m, 3 m width),

and visitation by wild and managed bees. Flower species that were visited by both wild and

managed bees are in bold, and those visited by neither are in grey. Asterisks mark species for

which inflorescences were counted rather than individual flowers.

Flower species Total field-edge

abundance

Wild bee

visits

Apis

mellifera

visits

Megachile

rotundata

visits

Achillea millefolium 46565 6 0 2

Allium textile 514 0 0 0

Arctium minus* 92 6 4 0

Asclepias speciosa 119 0 5 0

Astragalus bisulcatus 4692 3 2 0

Astragalus pectinatus 1142 0 0 0

Astragalus striatus 1050 3 0 0

Atriplex hortensis 150 0 0 0

Brassica napus 64456 115 125 15

Capsella bursa-pastoria 235341 13 4 91

Cirsium arvense* 24442 64 96 82

Cirsium undulatum* 39 10 3 0

Cirsium vulgare* 52 4 0 0

Cleome serrulata 50 0 0 0

Convolvulus arvensis 181 0 0 0

Convolvulus sepium 2191 4 6 0

Conyza canadensis* 15903 27 0 0

Coreopsis tinctoria* 16 0 0 0

Crepis tectorum* 14599 147 11 38

Cynoglossum officinale 8 0 0 0

Descurainia sophia 556965 27 1 27

Epilobium angustifolium 80 0 0 3

Epilobium ciliatum 4251 8 0 0

Erigeron caespitosus* 241 0 0 0

Erigeron compositus* 622 0 0 0

Eriogonum flavum 70 0 0 0

Erodium cicutarium 559 0 0 0

68

Flower species Total field-edge

abundance

Wild bee

visits

Apis

mellifera

visits

Megachile

rotundata

visits

Erysimum cheiranthoides 37 0 0 0

Gaillardia aristata* 40 2 0 0

Gaura coccinea 535 0 0 0

Geum triflorum 96 0 0 0

Glycyrrhiza lepidota 5957 1 0 0

Grindelia squarrosa* 1512 3 0 0

Gutierrezia sarothrae* 5444 3 0 0

Helianthus annuus* 6988 148 20 1

Heterotheca villosa* 25390 4 0 32

Hymenoxys acaulis* 129 0 0 0

Kochia scoparia 3070 0 5 0

Lactuca pulchella* 2136 13 11 2

Lactuca serriola* 15959 3 0 0

Lappula squarrosa 70 0 0 0

Liatris punctate 13008 20 0 1

Linum lewisii 262 6 0 0

Lygodesmia juncea 235 0 0 0

Malva rotundifolia 8121 0 0 0

Matricaria matricarioides* 70 0 0 0

Matricaria perforate 68 0 0 0

Medicago falcate 1077 0 0 0

Medicago lupulina 78718 0 0 4

Medicago sativa 2390848 45 472 558

Melilotus alba 165101 53 135 99

Melilotus officinalis 1042877 72 137 122

Oenothera nuttallii 14 0 0 0

Pentstemon albidus 53 0 0 0

Petalostemon purpureum* 1190 0 0 0

Pisum sativum 26510 0 5 1

Polygonum arenastrum 79043 0 0 0

Polygonum convolvulus 5755 6 0 0

Polygonum persicaria 784 0 0 0

Polygonum scabrum 10818 0 0 0

Potentilla norvegica 4989 6 0 0

69

Flower species Total field-edge

abundance

Wild bee

visits

Apis

mellifera

visits

Megachile

rotundata

visits

Potentilla pensylvanica 54 2 0 0

Ratibida columnifera* 3548 0 0 0

Rosa arkansana 123 1 0 0

Sagittaria cuneata 5 0 0 0

Senecio vulgaris* 756 1 0 5

Sinapis arvensis 11958 31 26 31

Sisymbrium altissimum 8411 11 0 15

Sisyrinchium montanum 9 0 0 0

Solanum ptychanthum 32 0 0 0

Solidago canadensis* 46140 1 0 0

Solidago rigida* 432 0 0 0

Sonchus arvensis* 7870 84 6 12

Sonchus asper* 1911 3 5 6

Sonchus oleraceus* 16 3 0 0

Sphaeralcea coccinea 11 0 0 0

Symphoricarpos occidentalis 18639 8 76 0

Symphyotrichum ericoides* 305 0 0 0

Symphyotrichum falcatum* 298 0 0 0

Symphyotrichum leave* 27 0 0 0

Taraxacum officinale* 5103 51 21 20

Thermopsis rhombifolia 69 0 0 0

Thlaspi arvense 65468 4 0 1

Tragopogon dubius* 931 18 0 3

Trifolium hybridum 19392 1 7 0

Verbena bracteata 52370 7 0 0

Vicia americana 2586 7 0 0

Vicia cracca 5464 1 5 0

Viola nuttalli 1 0 0 0

70

Appendix 3: Wild bee species found on aerial and sweep net transects conducted in canola crops.

When numbers in brackets are accompanied by a “+”, this indicates a conservative estimate of

the total number of morphospecies for a given genus that assumes all female morphospecies

correspond directly to male morphospecies.

Family Genus, species Abundance

Andrenidae 26

Andrena 26

prunorum 6

morphospecies (3+) 20

Apidae 28

Bombus 28

borealis 2

fervidus 1

huntii 9

nevadensis 3

rufocinctus 10

unknown 3

Colletidae 5

Colletes 4

morphospecies (2+) 4

Halictidae 44

Agapostemon 9

angelicus 6

virescens 1

unknown 2

Halictus 11

confusus 3

rubicundus 8

Lasioglossum 23

albipenne 8

albohirtum 3

prasinogaster 2

pulveris 1

ruidosense 1

sagax 1

semicaeruleum 7

Sphecodes 1

morphospecies (1) 1

71

Family Genus, species Abundance

Megachilidae 2

Megachile 1

perihirta 1

Unknown 6

Unknown 6

unknown 6