1
Wild pollinators provide the majority of crop visitation across
2
land use gradients in New Jersey and Pennsylvania
3
Rachael Winfree1, 5, Neal Williams 2, Hannah Gaines 3, John Ascher4, and Claire
4
Kremen5
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1
Corresponding author
Department of Ecology and Evolutionary Biology
Princeton University
Princeton, NJ 08544
609 258 7313 (phone) 609 258 1334 (FAX) rwinfree@princeton.edu
2
Department of Biology
Bryn Mawr College
101 N. Marion Ave.
Bryn Mawr, PA 19101
3
Department of Entomology
University of Wisconsin
1630 Linden Dr.
Madison, WI 53706
4
Division of Invertebrate Zoology
American Museum of Natural History
Central Park West at 79th St.
New York, NY 10024-5192
5
Department of Environmental Science, Policy and Management
University of California, Berkeley
137 Mulford Hall
Berkeley, CA 94720-3114
32
Running head: Crop visitation by wild pollinators
33
Word count: XX needs to be <8000; right now Summary – Supp Material info, it’s 8180
34
Yellow highlight in text is new since last version XX
35
1
36
Summary
37
1. Concern about a global decline in wild pollinators has led to increased interest in how
38
pollinators are affected by human land use, and how this in turn affects the ecosystem
39
services wild pollinators provide to crops.
40
2. We measured wild bee visitation to four summer vegetable crops, and investigated
41
associations between flower visitation rates and land use intensity at local and landscape
42
scales. We studied 29 farms in New Jersey and Pennsylvania, U.S.A. Over two years we
43
recorded >7400 bee visits to crop flowers and identified 54 species of wild bees visiting
44
crops.
45
3. Wild bees were the dominant flower visitors at three of the four crops we studied;
46
domestic honey bees Apis mellifera L. provided the remainder of visits. We estimated the
47
upper bound value of insect pollination in New Jersey and Pennsylvania for a subset of
48
crops as 107 - 221 million U.S. dollars year-1, suggesting that pollination by wild bees is
49
in the tens of millions year-1.
50
4. Ordination of the two best-studied crops showed that the wild bee species visiting
51
tomato Solanum lycopersicon L. were distinct from those visiting watermelon Citrullus
52
lanatus (Thunb.) Matsum. & Nakai.
53
5. Crop visitation by wild bees was not associated with organic farming, nor with natural
54
habitat cover at either the local or the landscape scales. Organic, as compared to
55
conventional, farming in our system is not associated with smaller field sizes, greater
56
crop diversity, or greater flowering weed abundance, features which may be more
57
important than organic farming per se in increasing wild bee abundance. Native habitat
2
58
cover may not have been an important factor in our study system where many bees may
59
utilize anthropogenic habitats more than native woodlands.
60
7. Synthesis and applications. Our findings contrast with other studies of crop visitation
61
by wild bees, most of which have found negative effects of intensifying agricultural land
62
use. Management recommendations developed in other types of ecosystems may not be
63
appropriate for temperate woodland-agricultural mosaics. The extent of crop visitation by
64
wild bees observed in this study is among the highest recorded and its value merits
65
inclusion in comprehensive assessments of ecosystem services.
66
67
Key-words: agroecosystems, bee, biodiversity-ecosystem function, crop pollination,
68
ecosystem services, forest fragmentation, landscape mosaic, organic farming
69
70
Introduction
71
Wild insects provide important services to agriculture, including pest control and
72
pollination of crops (Tscharntke et al., 2005; Losey & Vaughan, 2006). Agricultural
73
landscapes, when appropriately managed, can in turn provide habitat for many insect
74
species (Tscharntke et al., 2005). However, insect biodiversity is threatened by increasing
75
agricultural intensification, which includes the loss of natural and semi-natural habitats,
76
extensive monoculture plantings, and increased pesticide and herbicide use (Krebs et al.,
77
1999; Tscharntke et al., 2005). A better understanding of the benefits provided by insects,
78
and of land management practices that support insect biodiversity, could provide for
79
better cost-benefit analyses of land use alternatives, and help to make agricultural
80
production more sustainable.
3
81
Pollination is an important ecosystem service because 85% or more of 107 globally
82
traded crops, accounting for 35% of the global plant-based food supply, either require or
83
benefit from animal-mediated pollination (Klein et al., 2007). Bees (Hymenoptera:
84
Apoidea) are the most important crop pollinators (Delaplane & Mayer, 2000). Most
85
farmers rely on managed bees, primarily the honey bee, to provide crop pollination.
86
Despite the honey bee’s effectiveness as a pollinator for many crops, the risks associated
87
with reliance on a single, managed pollinator species have become evident over the past
88
15 years as North American honey bee populations have declined due to the parasitic
89
mite Varroa destructor Anderson & Trueman and other diseases (National Research
90
Council, 2006). Many wild bee species also contribute substantially to pollination of such
91
crops as coffee Coffea spp. (e.g. Klein, Steffan-Dewenter & Tscharntke, 2003b),
92
watermelon (Kremen, Williams & Thorp, 2002), tomato (Greenleaf & Kremen, 2006a),
93
blueberry Vaccinium spp. (Cane, 1997), sunflower (Greenleaf & Kremen, 2006b) and
94
canola Brassica spp. (Morandin & Winston, 2005).
95
Concern about a global decline in pollinators (Kearns, Inouye & Waser, 1998;
96
Biesmeijer et al., 2006) in combination with the recognition that few large-scale data sets
97
on pollinator populations exist (Ghazoul, 2005), has led to increased interest in how wild
98
pollinators are affected by human activity. Most previous studies suggest that bees, the
99
main pollinator taxon in most ecosystems (Axelrod, 1960), are sensitive to the loss of
100
natural and semi-natural habitats (Kremen & Chaplin-Kramer, In press). For example,
101
bee abundance and/or species richness decreases with increasing isolation from natural
102
habitats in such ecosystems as subtropical dry forest (Aizen & Feinsinger, 1994),
103
chaparral (Kremen, Williams & Thorp, 2002), tropical moist forest (Klein, Steffan-
4
104
Dewenter & Tscharntke, 2003a; Klein, Steffan-Dewenter & Tscharntke, 2003b; Ricketts,
105
2004), and subtropical premontane forest (Chacoff & Aizen, 2006). However, in some
106
situations wild bees are more abundant and/or species-rich in human-disturbed areas, and
107
particularly in agricultural habitats (Eltz et al., 2002; Klein et al., 2002; Westphal,
108
Steffan-Dewenter & Tscharntke, 2003; Winfree, Kremen & Griswold, 2007). A better
109
understanding of how pollinators respond to land use change would contribute to more
110
effective land management strategies as well as basic ecological knowledge. For
111
management purposes, it is especially important to understand the relative importance of
112
local- versus landscape-scale land use on pollinator communities (Tscharntke et al.,
113
2005).
114
In this study we investigated wild bees visiting flowers of summer vegetable crops in
115
a temperate forested ecosystem. We focused on three questions: 1) To what extent are
116
crops visited by wild, predominantly native bees, as compared to managed honey bees?,
117
2) Are different crops visited by different wild bee species?, and 3) Is wild bee visit
118
frequency to crops associated with land use intensity at the local and/or landscape scales?
119
We defined local land use as farm management (conventional farming versus the less
120
intensive organic strategy ), and landscape-scale land use as the proportion of natural
121
habitat (woodland) at various radii (0.5 – 3 km) around each farm. We expected that
122
honey bees would be the dominant visitors to crops because many farmers rent hives for
123
pollination, that crops with different floral structures would be visited by different wild
124
bee species which are known to have species-specific floral preferences (Wcislo & Cane,
125
1996), and that visitation by wild bees would be positively associated with both organic
126
farming and the extent of natural habitat in the surrounding landscape. As our outcome
5
127
variable, we focus on bee visit frequency to flowers, which along with per-visit pollen
128
deposition, determines the total pollination services contributed by the different pollinator
129
taxa (e.g., Kremen, Williams & Thorp, 2002; Morris, 2003). While the two are not
130
synonymous, a recent meta-analysis has shown that visit frequency is a stronger predictor
131
of total pollination than is per visit pollen deposition (Vázquez, Morris & Jordano, 2005).
132
Based on this relationship, we also sought a rough upper-bound estimate of the economic
133
value of crop pollination by wild bees in our study system.
134
135
Materials and methods
136
STUDY SYSTEM AND DESIGN
137
Our study system was a 90 by 60 km area of central New Jersey and eastern
138
Pennsylvania, U.S.A. This is a temperate, deciduous forest ecosystem dominated by oaks
139
Quercus spp., maples Acer spp, and hickories Carya spp., and in the understory by shrubs
140
such as viburnums Viburnum spp. and spicebush Lindera benzoin L. In 2004, we studied
141
22 farms, of which six were no-spray (i.e., growers do not spray either pesticides or
142
herbicides) or U.S. Department of Agriculture certified organic, and 16 were
143
conventional. In 2005, we used 16 of the previously-selected farms and added seven new
144
farms for a total of 23, of which seven were no-spray or organic and 16 were
145
conventional. Not all crops were grown at each farm, so sample sizes varied by crop
146
(Table 1). The study was designed to minimize spatial autocorrelation and co-linearity
147
between local- and landscape-scale variables in several ways. Farms of a given
148
management type (organic or conventional) and landscape setting (wooded or not) were
149
interspersed throughout the study area (Fig. 1), and farms were selected so that these
6
150
local- and landscape-scale variables were uncorrelated. Farms were also at least 1 km
151
apart, which is beyond the typical foraging movement distance of all but the largest bees
152
in the study (Greenleaf et al., Accepted).
153
In 2004, we surveyed pollinator visitation to four crops: tomato, bell pepper
154
Capsicum annuum L., muskmelon Cucumis melo L., and watermelon. In 2005, we
155
focused on two crops, cherry tomatoes (including plum and grape varieties; hereafter
156
cherry tomato) and watermelon, and collected more detailed data on each. Tomato and
157
pepper can self-pollinate without insect vectors; however, fruit number and quality
158
increase with bee pollination (Delaplane & Mayer, 2000; Greenleaf & Kremen, 2006a).
159
Tomato flowers do not produce nectar, and the anthers require sonication for full release
160
of pollen. Sonication behavior is performed by many wild bees, but not by honey bees;
161
we therefore expected tomato to be visited predominantly by wild bees. Pepper produces
162
nectar and pollen that does not require sonication, but is relatively unattractive to bees
163
(Delaplane & Mayer, 2000). Watermelon and muskmelon require insect vectors to
164
produce fruit because they are monoecious (watermelon) or functionally monoecious
165
(muskmelon; (Delaplane & Mayer, 2000). Flowers produce abundant nectar (females)
166
and easily-accessible pollen as well as nectar (males), and are attractive to bees. Their
167
flowers are active for only one day, opening at daybreak and closing by early afternoon.
168
We therefore collected our data in the morning (see below).
169
Sixty-four percent of the farmers in our study use hived honey bees for crop
170
pollination. Feral honey bees were virtually eliminated in our study region during the
171
1990s by Varroa mites and other problems (Paul Raybold, New Jersey State Apiarist,
172
pers. comm.). Therefore, most honey bees recorded in our study would have been
7
173
managed. We define “wild bees” as either native to our study area, or non-native but
174
unmanaged. The latter category accounted for only two species and 0.002% of specimens
175
collected.
176
177
DATA COLLECTION
178
Our data collection protocol in 2004 was a simplified version of that used in 2005, so we
179
first describe the latter. For each crop on each farm, we established a 50-m transect of
180
crop row where all data were collected. Transects began at the edge of the farm field in
181
order to standardize edge effects. We measured pollinator visitation rate to flowers during
182
45-second scans of flowers at 40 equally spaced points along the transect. We counted
183
and then observed visits to as many flowers as we could view simultaneously within an
184
approximately one by one meter area. Visitation rate data are therefore in units of bee
185
visits flower-1 time-1. We censused each transect at standard times of day, beginning at
186
8:30 and 10:30 hrs for tomato and 8:00, 9:30, and 11:00 hrs for watermelon. After
187
collecting the visitation rate data, we netted bees from crop flowers for 30 minutes and
188
used these specimens for species richness analyses.
189
We measured two co-variates related to floral resource availability: the row length of
190
flowering, bee-attractive crops for each crop in the farm field, and the abundance and
191
species richness of weedy flowers growing between the crop rows and in the fallow field
192
margins. The line intercept method (Kercher, Frieswyk & Zedler, 2003) was used to
193
measure weed flowers.
194
Data were only collected on days that were sunny, partly cloudy, or bright overcast,
195
with wind speeds of < 2.5 m/s. We collected data at each farm when the target crop was
8
196
at or near its peak period of bloom, as determined by repeated site visits prior to data
197
collection. Cherry tomato data were collected between 30 June and 14 July, and
198
watermelon data between 5 July and 10 August. We surveyed cherry tomato once and
199
watermelon twice at each farm.
200
In 2004, each crop at each farm was visited only once. Flower visitation rate data
201
were collected only once in the morning between 8:30 and 12:00 hours. We collected
202
bees by net at each farm, but as collection effort was limited, we do not use these data in
203
analyses. The abundance of flowering weedy vegetation in the fallow areas surrounding
204
the farm field was assessed visually on a scale of one to five, and crop flower abundance
205
within the farm field was not measured. All data were collected in July 2004.
206
We compiled information on nesting specialization and sociality of bees from the
207
published literature (e.g., from references in Krombien, Hurd & Smith, 1979; Michener,
208
2000) and used this for analyses of trait-specific responses of crop pollinators. We
209
defined wood-nesting bees as those species obligatorily nesting in rotting wood, or using
210
cavities in wood or twigs, but not those nesting in soft-pithed stems. We considered both
211
primitively eusocial and advanced eusocial species to be “eusocial,” and solitary and
212
communal species to be “solitary.”
213
214
GIS ANALYSES
215
We mapped the center of each data collection transect, and the perimeter of each farm
216
field, with a Trimble GeoExplorer Global Positioning System (GPS; Trimble Navigation,
217
Sunnyvale, CA) corrected to ± 10 m accuracy with GPS Pathfinder Office (version 2.9,
218
Touch Vision, Cypress, CA). GIS (Geographical Information Systems) land-cover data
9
219
for New Jersey were provided by the New Jersey State Department of Environmental
220
Protection, and were based on aerial photographs taken in 2002 and subsequently
221
classified to 61 land-cover types. GIS data for Pennsylvania were provided by the
222
Delaware Valley Regional Planning Commission, and were based on aerial photographs
223
taken in 2000 and subsequently classified to 27 land-cover types.
224
In our study system, patches of remaining woodland are irregularly shaped and often
225
interconnected by riparian corridors or other narrow strips of vegetation. Therefore we
226
did not attempt to define fragment size. Rather, We analyzed landscape cover as the
227
proportion of woodland at various radii surrounding each site (see below). The proportion
228
of habitat cover is highly correlated with other area-based indices of habitat proximity,
229
and is appropriate for analyzing community-scale response variables such as species
230
richness (Winfree et al., 2005). We used ArcGIS 9.0 (Environmental Systems Research
231
Institute, Redlands, CA) to determine the proportion of the area surrounding each farm
232
that consisted of woodland, the native vegetation type. As a local-scale measure of
233
isolation from natural habitat, we measured the linear distance from the center of our data
234
collection transect to the nearest patch of woodland.
235
236
PRELIMINARY ANALYSES
237
We tested for and found no spatial autocorrelation among sites at any scale in terms of
238
either wild bee visitation rate or bee species richness (Moran’s I , all P ≥ 0.07, R-
239
Package; http://www.bio.umontreal.ca/Casgrain/en/labo/R/v4/index.html; we used our
240
larger 2005 data sets for this analysis). To identify the scale at which surrounding land
241
cover had the most explanatory power, we regressed bee visitation rate against the
10
242
proportion of woodland surrounding study sites at radii of 500, 1000, 1500, 2000, 2500,
243
and 3000 m. We then compared the resulting r2 values, and used the scale with the
244
highest r2 value in all subsequent analyses (Holland, Bert & Fahrig, 2004).. We did this
245
separately for each crop, because the crops can have distinct pollinator communities (see
246
Results) which may respond to the surrounding landscape at different scales (e.g.,
247
Steffan-Dewenter et al., 2002) and because difference sites were used for different crops
248
and years. For watermelon and tomato in 2004, we used the scale of analysis determined
249
for the larger and more detailed 2005 data sets (Table 1). We repeated this analysis using
250
the 2005 watermelon data set of collected specimens to determine the scale of analysis
251
for the wood-nesting, eusocial, and solitary bee species (Table 1).
252
253
STATISTICAL ANALYSES
254
To compare flower visitation rates (bee visits flower-1 time-1) on a given crop between
255
wild bees and honey bees, we used paired t-tests, or when normality assumptions were
256
not met, Wilcoxon sign-rank tests.
257
We compared differences in species composition of the wild bee communities
258
visiting watermelon and tomato using nonmetric multidimensional scaling (NMS; PC-
259
ORD version 4, MjM Software, Glendeden Beach, OR). Sampling sites were ordinated
260
using the relative Sørensen index, which is based on the proportional rather than the
261
absolute abundance of each species. This metric avoids separating the two crops on the
262
basis of absolute bee abundance, because watermelon is more bee-attractive than tomato.
263
For NMDS analysis we used only the 14 farms for which we had data from both
264
watermelon and tomato (2005 data set), so as to control for differences among farms and
11
265
isolate the effect of crop type. We used only the watermelon data from the first day we
266
visited each farm in order to equalize sampling effort on the two crops. To assess the
267
significance of the difference between the two crops, we used multi-response permutation
268
procedures (MRPP) for blocked (paired) data in PC-ORD.
269
To assess the strength of local- versus landscape-scale environmental variables in
270
predicting bee visitation rate and species richness at crops, we defined six a priori
271
general linear statistical models, based on local- and landscape-scale factors that might
272
influence wild bee communities (Table 2). For the 2004 data sets, we used a reduced set
273
of models (Table 2). We compared among these models using the adjusted r2 values from
274
regression, ANOVA, and ANCOVA (JMP version 5.1, SAS Institute, Cary, NC). This
275
least-squares method of model selection will produce the same results as AIC (Akaike’s
276
Information Criterion) model selection, provided that model errors are normally
277
distributed (Hilborn & Mangel, 1997), which was the case for our data. To improve
278
normality and homoskedasticity, we used the following transformations of the outcome
279
variables: for all data sets except watermelon (2004 and 2005), flower visitation rate was
280
arcsin square root transformed; counts of wood-nesting bees were transformed as ln (x +
281
1); and counts of solitary bees were transformed as square root (x + 0.5) (Sokal & Rohlf,
282
1995).
283
To analyze the 2005 watermelon data set, for which we made repeat visits to each
284
farm, we used “robust cluster” regression (Stata version 7.0, Stata Corp., College Station,
285
TX). Robust clustering is a bootstrap procedure for sampling from data that are not
286
independent within a cluster (in our case, sample dates within a farm) but that are
287
independent across clusters (in our case, across farms).
12
288
For the analyses of how particular species groups that share a life history trait respond
289
to land use, our measure of bee abundance was the number of bee specimens collected
290
rather than visitation rate, because these analyses require data resolved to the species
291
level. We limited analysis to the 2005 watermelon data set which contained 70% of the
292
collected specimens. For eusocial and solitary bee species, we used the six models in
293
Table 2. For wood-nesting bee species, we investigated only associations with the two
294
predictor variables related to woodland (distance to nearest woodland patch, and
295
proportion of woodland in the surrounding landscape), which are the main variables of
296
interest for this group. We were unable to investigate the full set of models with this data
297
set which had many zero values and did not meet the models’ statistical assumptions.
298
We estimated the annual dollar value of crop pollination by insects (wild bees and
299
honey bees combined, plus a smaller contribution by flies) for a subset of crops by
300
multiplying the statewide crop production value in 2005 (USDA-NASS, 2006b, 2006a)
301
by the crop’s dependence on insect pollination (Klein et al., 2007). This valuation method
302
provides an upper-bound estimate representing the current “subsidy from nature” (Losey
303
& Vaughan, 2006). Production value data are only available for a subset of the most
304
economically important crops in each state, which is why we considered only this subset.
305
306
Results
307
Over the two years of the study, we observed 7434 bee visits (4592 by wild bees and
308
2842 by honey bees) to 32,171 crop flowers, and collected 1738 wild bee specimens,
309
which were identified to 54 species (see Appendix S1 in Supplementary Material).
13
310
In 2004, we found that tomato and pepper were visited significantly more frequently
311
by wild bees than by honey bees (for tomato, Wilcoxon sign-rank test with n = 17, W = 0,
312
P < 0.001; for pepper, paired t-test with df = 21, t = - 4.22, P < 0.001), whereas for
313
watermelon and muskmelon wild bee and honey bee visitation rates did not differ
314
significantly (for watermelon, paired t-test with df = 11, t = - 0.02, P = 0.98; for
315
muskmelon, Wilcoxon sign-rank test with n = 14, W = 28, P > 0.10). In our more
316
extensive data sets from 2005, tomato and watermelon were both visited significantly
317
more often by wild bees (for tomato, Wilcoxon sign-rank test with n = 13, W = 0, P <
318
0.001; for watermelon, paired t-test with df = 22, t = - 2.45, P = 0.02; Fig. 2). Crop flower
319
visitation rates by honey bees and wild bees were not significantly negatively correlated
320
across farms for any of the crops in the study.
321
The wild bee communities visiting watermelon and tomato were clearly separable in
322
ordination analysis (Fig. 3), and were significantly different by the MMRP test (A = 0.09,
323
P = 0.004). The NMS ordination reliably represented the true pairwise similarities among
324
sites: the r2 value, which measures the proportion of variance explained by the rank
325
correlation between the true pairwise similarity values (in our case, the relative Sørensen
326
index) and the physical distances between sites on the ordination plot, was 0.76.
327
Neither local- nor landscape-scale land use variables strongly predicted wild bee
328
visitation rate or species richness at crops (Table 3). Even after selecting the best model
329
for each data set, only two of the eight models were statistically significant at the P =
330
0.05 level, and none of the models was significant after adjusting the critical P-value for
331
multiple comparisons (adjusted P = 0.006, Dunn-Šidák correction). Similarly, no
332
individual variables were strongly or consistently predictive (Table 3).
14
333
The abundance of eusocial bees showed little association with land use variables. The
334
best model was number six (Table 2; whole model r2 = 0.06, P = 0.13), which included
335
only the proportion of woodland cover in the surrounding landscape. The abundance of
336
solitary bees showed little association with land use variables, but were positively
337
associated with the abundance of weedy flowers in the farm field. The best model was
338
number one (Table 2; whole model r2 = 0.20, P = 0.01), and the model’s overall
339
significance was driven by the variable weedy flower abundance (t = 3.11, P = 0.005).
340
For wood-nesting bees, we examined only the two variables related to woodland cover.
341
The abundance of wood-nesting bees was weakly associated with distance to the nearest
342
woodland patch (t = - 1.39, P = 0.18) and was not associated with the proportion of
343
woodland in the surrounding landscape at a 1000 m radius (t = 0.42, P = 0.68).
344
The value of crop pollination by insects for a subset of crops in New Jersey and
345
Pennsylvania was estimated as being between 107 and 221 million U.S. dollars per year
346
(see Appendix S2). The low estimate results from using the lower bound on the increase
347
in fruit production attributable to insect pollination as reported in Klein et al. (2007),
348
whereas the high estimate results from using the upper bound.
349
350
Discussion
351
HUMAN LAND USE AND WILD BEE VISITATION TO CROPS
352
Contrary to our starting expectations, we found no strong associations between land
353
use intensity at either the local or the landscape scales, and wild bee visitation to crops
354
(Table 3). The best models were statistically significant for only two of our eight data
355
sets, and none was significant after correcting for multiple comparisons.
15
356
Natural history traits such as sociality and nesting requirements can determine how
357
bees respond to land use (Klein, Steffan-Dewenter & Tscharntke, 2003a; Steffan-
358
Dewenter et al., 2006) , and variation in traits among species could lead to findings of no
359
significant effect for the bee community as a whole. However, our analyses by natural
360
history trait reveal only one pattern not found at the community scale: solitary species
361
were positively associated with the abundance of weedy flowers in the farm field. This
362
suggests that farm management that allows flowering weeds to remain might increase bee
363
visitation to crops, at least for bee-attractive crops such as watermelon. As found for the
364
community-scale analyses, neither solitary, eusocial, nor wood-nesting species showed
365
associations with woodland cover. Eusocial species showed little association with any
366
land use variable. Surprisingly, obligatorily wood-nesting bees did not have strong
367
associations with woodland cover at either the local or the landscape scales.
368
Most previous studies of crop visitation by wild pollinators have found negative
369
effects of human land use at the local and/or landscape scales (reviewed in Kremen &
370
Chaplin-Kramer, In press). Why then did land use intensity have such weak effects in our
371
study? We suggest five possible reasons, which are not mutually exclusive. First, we
372
considered the possibility that we were unable to detect the true effects of human
373
disturbance because our data sets were too small and/or variable, especially given the
374
known variability of wild bee populations (Williams, Minckley & Silveira, 2001). This
375
may have been the case for our smaller data sets from 2004. However, the statistical
376
models for the more extensive 2005 data sets showed even less association between wild
377
bees and land use than those from 2004. In particular our 2005 watermelon data set
378
included 23 sites, 3828 wild bee visits to 15,888 watermelon flowers, and 1221 collected
16
379
specimens, yet all variables relating to land use intensity were highly non-significant. In
380
contrast, many studies finding significant effects of land use on wild bees in other
381
systems had smaller or similar sample sizes, e.g. 104 wild bee visits to grapefruit in
382
Argentina (Chacoff & Aizen, 2006), 615 wild bee visits to coffee in Costa Rica (Ricketts,
383
2004), and 3349 wild bee visits to watermelon in California (Kremen, Williams & Thorp,
384
2002; Kremen et al., 2004). Furthermore, in our 2005 watermelon study the range of
385
values included within the 95% confidence intervals for flower visitation rates versus
386
natural habitat cover are all sufficient for full pollination (this approach is preferable to
387
power analysis; Colegrave & Ruxton, 2005).
388
Second, we may have found few differences between organic and conventional farm
389
management because in our system many of the features often associated with organic
390
farming are common to both management categories (Hole et al., 2005). In our study
391
system, both organic and conventional farms are family operated and grow mixed
392
vegetable crops; none supports large monoculture plantings. Farm field size did not differ
393
between organic and conventional farms (t-test, t = 0.92, P = 0.37), nor did the density of
394
flowering crop “species” (Wilcoxon test, 2 = 1.0, P = 0.32), weedy flower density
395
(Wilcoxon test, 2 = 0.45, P = 0.50), or the density of weedy flower species (t-test, t =
396
0.50, P = 0.62). In our system, organic and conventional farms differ primarily in the
397
extent of pesticide, herbicide, and fertilizer use. Synthetic pesticide, herbicide, and
398
fertilizers are used only on the conventional farms, and some of the synthetic pesticides
399
used in our area are known to be highly toxic to bees (e.g., Malathion and permethrin).
400
However, organic farms are also permitted to use certain non-synthetic pesticides.
401
Overall, our findings are consistent with the suggestion of Benton, Vickery & Wilson
17
402
(2003), that the habitat heterogeneity resulting from factors such as weediness and field
403
size may be more important than organic farming per se in supporting biodiversity. In
404
Europe, the agri-environment scheme, a low-intensity designation similar to the organic
405
designation in the U.S., appears to have mixed though generally positive effects on
406
biodiversity (Kleijn et al., 2006).
407
Third, the lack of sensitivity to natural habitat loss that we found in this study may be
408
a result of working in a temperate forested ecosystem. Previous studies of forest loss and
409
wild bee visitation to crops, which found mostly negative effects of forest loss, were done
410
in tropical ecosystems (Klein, Steffan-Dewenter & Tscharntke, 2003a; Klein, Steffan-
411
Dewenter & Tscharntke, 2003b; Ricketts, 2004; Chacoff & Aizen, 2006). We cannot
412
account for why bee communities in tropical forests might have different responses from
413
those in temperate forests. However, several studies of wild bee populations per se
414
(unrelated to crop visitation ) in temperate forest ecosystems suggest that although forests
415
support certain bee species, they support lower bee densities than human-disturbed
416
habitats (Banaszak, 1992, 1996; Klemm, 1996; Steffan-Dewenter et al., 2002; Winfree,
417
Kremen & Griswold, 2007). Some of these anthropogenic habitats are now considered a
418
critical habitat for insect conservation — for example, central European calcareous
419
grasslands, which are grazed meadows that were deforested in historical times for
420
pasturing livestock (Steffan-Dewenter & Tscharntke, 2002).
421
Fourth, we speculate that high habitat heterogeneity within the foraging ranges of
422
bees may allow landscapes to support diverse and numerous bees even when the total
423
amount of natural habitat remaining is low. Land cover in our system is heterogeneous,
424
with small patches of trees dispersed throughout even the most intensively used
18
425
landscapes. This heterogeneity is reflected in the high range of variation in our data sets
426
in terms of the proportion of woodland cover in the surrounding landscape (3 – 60%), as
427
compared to the small range of variation in the distance to the nearest patch of woodland
428
(18 – 343 m). Habitat heterogeneity is associated with increased species richness in other
429
taxa (Rosenzweig, 1995; Ricklefs & Lovette, 1999), and bees may be particularly
430
benefited by habitat heterogeneity because their foraging, nesting, and over-wintering
431
resources are often located in different habitat types (Westrich, 1996). Furthermore, the
432
grain or scale (Levin, 1992) at which many bees experience the landscape may be small,
433
given their small body size and typical daily movement distances of a few hundred
434
meters (Greenleaf et al., Accepted). In contrast to findings for larger-bodied taxa, small
435
habitat fragments may support high insect diversity; on calcareous grasslands in central
436
Europe, the cumulative number of insect species found in a series of small fragments is
437
greater than the number of species found in a single large fragment of equivalent area
438
(although bees were not included in this study; Tscharntke et al., 2002).
439
Fifth, the phenology of floral resources in our study system may be complementary
440
between woodlands and human-disturbed habitats (Heinrich, 1976). In early spring,
441
flowers and pollinators are abundant in woodlands, but by the time the forest canopy
442
closes in June, there is little bloom and few pollinators (RW, unpublished data). In
443
contrast, by June many species, particularly exotics, are flowering in agricultural areas,
444
old fields, and gardens, and such habitats can continue to provide floral resources through
445
the fall. Our data set was dominated by bee species with long flight seasons (e.g.,
446
Bombus, Lasioglossum (Dialictus), Halictus, Ceratina) which could take advantage of
447
this complementarity over time. Indeed, because we studied bees visiting summer
19
448
vegetable crops, our data were probably biased towards including such long-flying
449
species.
450
451
CROP VISITATION BY MANAGED VERSUS UNMANAGED BEES
452
Most of the pollinator visitation to summer vegetable crops in our study region is by wild
453
bee species. This is a striking finding from a farm management perspective, because most
454
of the farmers in our region rent honey bees to pollinate their crops, and few are aware of
455
the pollination provided by wild species. The upper-bound value of insect pollination
456
services to a subset of crops in New Jersey and Pennsylvania is 107 - 221 million U.S.
457
dollars per year. If the four crops we studied are representative of other crop species, it
458
seems likely that wild bees are providing at least half of that value.
459
Our study contributes to the growing literature demonstrating the importance of crop
460
pollination by wild bees (Kremen & Chaplin-Kramer, In press), and is notable for finding
461
that wild, native bees provide more visitation than managed or feral Apis mellifera, being
462
responsible for 62% of the flower visits over the entire study system. Previous studies of
463
canola (Morandin & Winston, 2005) and coffee in Indonesia (Klein, Steffan-Dewenter &
464
Tscharntke, 2003a; Klein, Steffan-Dewenter & Tscharntke, 2003b) found that wild bees,
465
including wild, native Apis species for coffee, contributed >98% of the flower visits. All
466
other studies have found that wild bees contribute a smaller proportion of visits than we
467
found, when summed over all sites in the system (although particular sites can have
468
higher visitation than the means presented here). Wild bees accounted for 59% of visits to
469
coffee in Costa Rica (Ricketts, 2004), 51% of visits to longan in Australia (Blanche,
470
Ludwig & Cunningham, 2006), about 37% (Heard & Exley, 1994) or 0% (Blanche,
20
471
Ludwig & Cunningham, 2006) of visits to macadamia in Australia, 34% of visits to
472
watermelon in California (Kremen, Williams & Thorp, 2002; Kremen et al., 2004), 28%
473
of visits to sunflower (Greenleaf & Kremen, 2006b) in California, 2-30% of visits to a
474
variety of crops in Poland (Banaszak, 1996), and 3% of the visits to grapefruit in
475
Argentina (Chacoff & Aizen, 2006).
476
Although flower visitation rate is not equivalent to pollination, which also depends on
477
per-visit pollen deposition, visitation is the most important predictor of actual pollination
478
(Vázquez, Morris & Jordano, 2005). Some wild bees surpass Apis mellifera in per-visit
479
pollen deposition for some crops (e.g., Delaplane & Mayer, 2000). In a study of pollen
480
deposition at watermelon in our system, we found that wild bees bracket honey bees in
481
per-visit pollen deposition, and that wild bee visitation rate at a farm is highly correlated
482
with estimated day-long pollen deposition by wild bees at that farm (Pearson’s r = 0.81;
483
RW, unpublished data).
484
Lastly, the wild bee communities visiting watermelon and tomato crops were distinct
485
(Fig. 3), even when we controlled for farm and for the differential attractiveness of the
486
two crops. Our result lends credence to the argument that conservation of diverse
487
communities is necessary in order to retain the full range of ecosystem services (e.g.,
488
Tilman, 1999).
489
490
CONCLUSIONS AND MANAGEMENT RECOMMENDATIONS
491
In contrast to previous studies of crop pollination by wild bees, in our system neither
492
local- and landscape-scale land use affected wild bee visit frequency to crop flowers.
493
Contrasts between our study system and others suggest possible reasons for these
21
494
differences. First, organic farming in our system in our system is not distinct from
495
conventional farming in field size, crop diversity, or flowering weed abundance. These
496
co-variates may be more important to wild bee populations than is organic farming per
497
se, and warrant further work in developing policy and management recommendations
498
(Benton, Vickery & Wilson, 2003). Second, we worked in a temperate forested
499
ecosystem, where many bees may use anthropogenic habitats more than they use the
500
native woodlands (Winfree, Kremen & Griswold, 2007). This could account for the lack
501
of significant effects of native woodland in our study. Management recommendations for
502
increasing wild bee contributions to crop pollination that were developed in other types
503
of ecosystems may not be appropriate for temperate woodlands.
504
Wild bees were responsible for the majority of crop flower visitation to three of four
505
summer vegetable crops. Different bee communities were found on different crops,
506
suggesting that maintaining diverse bee communities leads to more complete pollination
507
services. The value of crop pollination by wild bees in New Jersey and Pennsylvania is
508
probably in the tens of millions U.S. dollars year-1 and adds to the growing evidence for
509
the global value of ecosystem services.
510
22
511
Acknowledgements
512
We thank the many landowners who participated in the study; Sam Droege and Terry
513
Griswold for help with bee species identification; and Christina Locke, Rosemary Malfi,
514
Daniela Miteva, and Becky Tonietto for field and lab work. Funding was provided by the
515
National Fish and Wildlife Foundation, the Bryn Mawr College Summer Science
516
program, and NSF (grant #’s DEB-05-54790 and DEB-05-16205 to CK, NW and RW).
23
517
References
518 National Research Council of the National Academies. (2006) Status of Pollinators in
519
North America National Academy Press, Washington, D.C.
520 Aizen, M.A. & Feinsinger, P. (1994) Habitat fragmentation, native insect pollinators, and
521
feral honey bees in Argentine "chaco serrano". Ecological Applications, 4, 378-92.
522 Axelrod, D.I. (1960). The evolution of flowering plants. In Evolution After Darwin,
523
Volume I: The Evolution of Life (ed S. Tax), pp. 227-305. University of Chicago Press,
524
Chicago, IL.
525 Banaszak, J. (1992) Strategy for conservation of wild bees in an agricultural landscape.
526
Agriculture, Ecosystems and Environment, 40, 179-92.
527 Banaszak, J. (1996). Ecological bases of conservation of wild bees. In The Conservation of
528
Bees. Linnean Society Symposium Series 18 (eds A. Matheson, S.L. Buchmann, C.
529
O'Toole, P. Westrich & I.H. Williams), pp. 55-62. Academic Press, London, UK.
530 Benton, T.G., Vickery, J.A. & Wilson, J.D. (2003) Farmland biodiversity: is habitat
531
heterogeneity the key? Trends in Ecology and Evolution, 18, 182-88.
532 Biesmeijer, J.C., Roberts, S.P.M., Reemer, M., Oholemuller, R., Edwards, M., Peeters, T.,
533
Schaffers, A.P., Potts, S.G., Kleukers, R., Thomas, C.D., Settele, J. & Kunin, W.E.
534
(2006) Parallel declines in pollinators and insect-pollinated plants in Britain and the
535
Netherlands. Science, 313, 351-54.
536 Blanche, K.R., Ludwig, J.A. & Cunningham, S.A. (2006) Proximity to rainforest enhances
537
pollination and fruit set in macadamia and longan orchards in north Queensland,
538
Australia. Journal of Applied Ecology, 43, 1182-87.
24
539 Cane, J.H. (1997) Lifetime monetary value of individual polilnators: the bee Habropoda
540
laboriosa at rabbiteye blueberry (Vaccinium ashei Reade). Acta Horticultura, 446, 67-70.
541 Chacoff, N.P. & Aizen, M.A. (2006) Edge effects on flower-viisting insects in grapefruit
542
plantations bordering premontane subtropical forest. Journal of Applied Ecology, 43, 18-
543
27.
544 Colegrave, N. & Ruxton, G.D. (2005) Confidence intervals are a more useful complement
545
to nonsignificant tests than are power calculations. Behavioral Ecology, 14, 446-50.
546 Delaplane, K.S. & Mayer, D.F. (2000) Crop Pollination by Bees CABI Publishing, New
547
York, NY.
548 Eltz, T., Bruhl, C.A., van der Kaars, S. & Linsenmair, K.E. (2002) Determinants of
549
stingless bee nest density in lowland dipterocarp forests of Sabah, Malaysia. Oecologia,
550
131(1), 27-34.
551 Ghazoul, J. (2005) Buzziness as usual? Questioning the global pollination crisis. Trends in
552
Ecology and Evolution, 20, 367-73.
553 Greenleaf, S.S. & Kremen, C. (2006a) Wild bee species increase tomato production but
554
respond differently to surrounding land use in Northern California. Biological
555
Conservation, 133, 81-87.
556 Greenleaf, S.S. & Kremen, C. (2006b) Wild bees enhance honey bees' pollination of hybrid
557
sunflower. Proceedings of the National Academy of Sciences, 103, 13890-95.
558 Greenleaf, S.S., Williams, N., Winfree, R. & Kremen, C. (Accepted) Bee foraging ranges
559
and their relationship to body size. Oecologia.
560 Heard, T.A. & Exley, E.M. (1994) Diversity, abundance, and distribution of insect visitors
561
to macadamia flowers. Environmental Entomology, 23(1), 91-100.
25
562 Heinrich, B. (1976) Flowering phenologies: bog, woodland, and disturbed habitats.
563
Ecology, 57, 890-99.
564 Hilborn, R. & Mangel, M. (1997) The Ecological Detective: Confronting Models with Data
565
Princeton University Press, Princeton, NJ.
566 Hole, D.G., Perkins, A.J., Wilson, J.D., Alexander, I.H., Grice, P.V. & Evans, A.D. (2005)
567
Does organic farming benefit biodiversity? Biological Conservation, 122, 113-30.
568 Holland, J.D., Bert, D.G. & Fahrig, L. (2004) Determining the spatial scale of species'
569
response to habitat. Bioscience, 54, 227-33.
570 Kearns, C.A., Inouye, D.W. & Waser, N.M. (1998) Endangered mutualisms: the
571
conservation of plant-pollinator interactions. Annual Review of Ecology and Systematics,
572
29, 83-112.
573 Kercher, S.M., Frieswyk, C.B. & Zedler, J.B. (2003) Effects of sampling teams and
574
estimation methods on the assessment of plant cover. Journal of Vegetation Science, 14,
575
899-906.
576 Kleijn, D., Baquero, R.A., Clough, Y., Diaz, M., De Esteban, J., Fernandez, F. & Others
577
(2006) Mixed biodiversity benefits of agri-environment schemes in five European
578
countries. Ecology Letters, 9, 243-54.
579 Klein, A.-M., Steffan-Dewenter, I., Buchori, D. & Tscharntke, T. (2002) Effects of land580
use intensity in tropical agroforestry systems on coffee flower-visiting and trap-nesting
581
bees and wasps. Conservation Biology, 16, 1003-14.
582 Klein, A.-M., Steffan-Dewenter, I. & Tscharntke, T. (2003a) Fruit set of highland coffee
583
increases with the diversity of pollinating bees. Proceedings of the Royal Society of
584
London Series B Biological Sciences, 270, 955-61.
26
585 Klein, A.M., Steffan-Dewenter, I. & Tscharntke, T. (2003b) Pollination of Coffea
586
canephora in relation to local and regional agroforestry managment. Journal of Applied
587
Ecology, 40, 837-45.
588 Klein, A.M., Vaissière, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen,
589
C. & Tscharntke, T. (2007) Importance of pollinators in changing landscapes for world
590
crops. Proceedings of the Royal Society of London, Series B, 274, 303-13.
591 Klemm, M. (1996). Man-made bee habitats in the anthropogenous landscape of central
592
Europe: substitutes for threatened or destroyed riverine habitats? In The Conservation of
593
Bees (eds A. Matheson, S.L. Buchmann, C. O'Toole, P. Westrich & I.H. Williams).
594
Academic Press, London, UK.
595 Krebs, J.R., Wilson, J.D., Bradbury, R.B. & Siriwardena, G.M. (1999) The second Silent
596
Spring? Nature, 400, 611-12.
597 Kremen, C. & Chaplin-Kramer, R. (In press). Insects as providers of ecosystem services:
598
crop pollination and pest control. In Insect Conservation Biology: Proceedings of the
599
Royal Entomological Society's 23rd Symposium (eds A.J.A. Stewart, T.R. New & O.T.
600
Lewis). CABI Publishing, Wallingford, UK.
601 Kremen, C., Williams, N.M., Bugg, R.L., Fay, J.P. & Thorp, R.W. (2004) The area
602
requirements of an ecosystem service: crop pollination by native bee communities in
603
California. Ecology Letters, 7, 1109-19.
604 Kremen, C., Williams, N.M. & Thorp, R.W. (2002) Crop pollination from native bees at
605
risk from agricultural intensification. Proceedings of the National Academy of Sciences,
606
99, 16812-16.
27
607 Krombien, K.V., Hurd, P.D., Jr. & Smith, D.R. (1979) Catalog of Hymenoptera in
608
American North of Mexico Smithsonian Institution Press, Washington, DC.
609 Levin, S. (1992) The problem of pattern and scale in ecology. Ecology, 73, 1943-67.
610 Losey, J.E. & Vaughan, M. (2006) The economic value of ecological services provided by
611
insects. Bioscience, 56(4), 311-23.
612 Michener, C. (2000) The Bees of the World Johns Hopkins University Press, Baltimore and
613
London.
614 Morandin, L.A. & Winston, M.L. (2005) Wild bee abundance and seed production in
615
conventional, organic, and genetically modified canola. Ecological Applications, 15, 871-
616
81.
617 Morris, W. (2003). Which mutualists are most essential? In The importance of species:
618
perspectives on expendability and triage (eds P. Karieva & S. Levin). Princeton
619
University Press, Princeton, NJ.
620 Ricketts, T.H. (2004) Tropical forest fragments enhance pollinator activity in nearby coffee
621
crops. Conservation Biology, 18(5), 1262-71.
622 Ricklefs, R.E. & Lovette, I.J. (1999) The roles of island area per se and habitat diversity in
623
the species-area relationships of four Lesser Antillean faunal groups. Journal of Animal
624
Ecology, 68, 1142-60.
625 Rosenzweig, M.L. (1995) Species diversity in space and time Cambridge University Press,
626
New York, NY.
627 Sokal, R.R. & Rohlf, F.J. (1995) Biometry: the Principles and Practice of Statistics in
628
Biological Research, 3rd edn. W H Freeman, New York, NY.
28
629 Steffan-Dewenter, I., Klein, A.M., Gaebele, V., Alfert, T. & Tscharntke, T. (2006). Bee
630
diversity and plant-pollinator interactions in fragmented landscapes. In Plant-Pollinator
631
Interactions: From Specialization to Generalization (eds N. Waser & J. Ollerton), pp.
632
387-407. University of Chicago Press, Chicago.
633 Steffan-Dewenter, I., Münzenberg, U., Bürger, C., Thies, C. & Tscharntke, T. (2002)
634
Scale-dependent effects of landscape context on three pollinator guilds. Ecology, 83,
635
1421-32.
636 Steffan-Dewenter, I. & Tscharntke, T. (2002) Insect communities and biotic interactions on
637
fragmented calcareous grasslands: a mini review. Biological Conservation, 104, 275-84.
638 Tilman, D. (1999) The ecological consequences of changes in biodiversity: a search for
639
general principles. Ecology, 80, 1455-74.
640 Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I. & Thies, C. (2005)
641
Landscape perspectives on agricultural intensification and biodiversity - ecosystem
642
service management. Ecology Letters, 8, 857-74.
643 Tscharntke, T., Steffan-Dewenter, I., Kruess, A. & Thies, C. (2002) Contribution of small
644
habitat fragments to conservation of insect communities of grassland-cropland
645
landscapes. Ecological Applications, 12, 354-63.
646 USDA-NASS. (2006a). Non-citrus fruits and nuts: 2005 summary. United States
647
Department of Agriculture, National Agricultural Statistics Service, Washington, D.C.
648 USDA-NASS. (2006b). Vegetables: 2005 Summary. United States Department of
649
Agriculture, National Agricultural Statistics Service, Washington, D.C.
650 Vázquez, D.P., Morris, W.F. & Jordano, P. (2005) Interaction frequency as a surrogate for
651
the total effect of animal mutualists on plants. Ecology Letters, 8, 1088-94.
29
652 Wcislo, W.T. & Cane, J.H. (1996) Floral resource utilization by solitary bees
653
(Hymenoptera: Apoidea) and exploitation of their stored foods by natural enemies.
654
Annual Review of Ecology and Systematics (Annual Review of Entomology, 41, 257-86.
655 Westphal, C., Steffan-Dewenter, I. & Tscharntke, T. (2003) Mass flowering crops enhance
656
pollinator densities at a landscape scale. Ecology Letters, 6, 961-65.
657 Westrich, P. (1996). Habitat requirements of central European bees and the problems of
658
partial habitats. In The Conservation of Bees (eds A. Matheson, S.L. Buchmann, C.
659
O'Toole, P. Westrich & I.H. Williams), pp. 1-16. Academic Press for the Linnean Society
660
of London and IBRA, London, UK.
661 Williams, N.M., Minckley, R.L. & Silveira, F.A. (2001) Variation in native bee faunas and
662
its implications for detecting community changes. Conservation Ecology, 5, [online]
663
URL: http://www.consecol.org/vol5/iss1/art7.
664 Winfree, R., Dushoff, J., Crone, E., Schultz, C.B., Williams, N. & Kremen, C. (2005)
665
Testing simple indices of habitat proximity. American Naturalist, 165, 707-17.
666 Winfree, R., Kremen, C. & Griswold, T. (In Press 2007) Effect of human disturbance on
667
bee communities in a forested ecosystem. Conservation Biology.
668
669
670
30
671
Tables
672
Table 1: The six data sets analyzed, with the radius (scale) used in analyses of landscape
673
cover, and the range of variation in the proportion of woodland at that scale. Separate
674
analyses were done for particular species groups in the 2005 watermelon data set.
Crop
Number
Radius with
Range of variation in
of farms
highest r2 (m)
proportion woodland (%)
23
2000
8 - 60
Wood-nesting (7 species)
1000
5 - 53
Eusocial (22 species)
30001
10 - 64
Solitary (18 species)
1000
5 - 53
Watermelon
Year
2005
Tomato
2005
13
500
3 - 38
Watermelon
2004
12
2000
8 - 59
Tomato
2004
17
500
3 - 42
Muskmelon
2004
14
1000
5 - 51
Pepper
2004
22
500
3 - 59
675
1
676
bees didn’t have a true maximum at a larger radius.
Because the 3000 m was the largest radius we examined, we cannot be sure that eusocial
677
31
678
Table 2: The statistical models compared to determine the relative importance of local-
679
and landscape-scale variables in predicting wild bee visitation rate and species richness at
680
crop flowers.
Models for 2005 data sets
1: local floral resources
Included variables
weedy flower density (cm / 300m), weedy
flower species (number of species / 300m),
flowering crop density (m / 1000m2); flowering
crop “species” (number of species 1000m-2)
2: local farm management
organic / conventional
3. local isolation from natural habitat
distance from transect center to nearest
woodland (m)
4: local, all variables
weedy flower density, weedy flower species,
flowering crop density; flowering crop
“species”; organic / conventional; distance from
transect center to nearest woodland
5. landscape cover by natural habitat
proportion woodland at most explanatory radius
(see Table 1)
6. all local and landscape variables
weedy flower density, weedy flower species,
flowering crop density; flowering crop
“species”; organic / conventional; distance from
transect center to nearest woodland; proportion
woodland
32
Models for 2004 data sets
1: local
Included variables
weedy flower abundance (ranked 1-5), organic /
conventional1
2. landscape
proportion of woodland at most explanatory
radius (see Table 1)
3. local and landscape
weedy flower abundance, organic /
conventional1, proportion of woodland
681
682
1
683
only one and two organic farms, respectively.
This variable was not used for the muskmelon or watermelon data sets, which included
33
684
Table 3: The most explanatory model for each data set, and the slope coefficient B (or t for categorical variables), and SE (or P-
685
value for categorical variables) of each included variable. See Table 2 for model definitions.
Data set
Watermelon
Most explanatory model1
Visitation: local model 1
Whole
Whole model
model r2
P-value
0.19
0.02
2005
Species richness: landscape
B2
SE3
weedy flower density
0.0018
0.0009
weedy flower species
0.0106
0.0122
flowering crop density
0.0000
0.0002
flowering crop species
0.0147
0.0139
-0.0600
0.0371
0.842
0.423
Model variables
0.05
0.12
% woodland 2000 m
Visitation: local model 2
0.06
0.42
organic
Species richness: local model 3
0.16
0.18
distance to woodland
0.0060
0.0041
Visitation: landscape model 1
0.04
0.51
% woodland 2000 m
0.0015
0.0022
Visitation: local model 1
0.12
0.42
weedy flower abundance
-0.3281
1.2122
model 5
Tomato
2005
Watermelon
2004
Tomato
34
2004
Muskmelon
2004
Pepper
2004
1.342
0.203
weedy flower abundance
2.2794
1.6295
% woodland 1000 m
0.2802
0.1214
weedy flower abundance
-0.2458
1.8087
% woodland 500 m
0.2056
0.1536
1.732
0.103
organic
Visitation: local and landscape
0.49
0.03
model 3
Visitation: local and landscape
0.32
model 3
0.07
organic
686
687
1
The 2004 and 2005 data sets use different models; see Table 2.
688
2
t is reported for categorical variables
689
3
P-value is reported for categorical variables
690
35
691
Figures
Fig. 1: Map of the study area, with natural woodland habitat shaded gray, and
the study farms used in 2005 indicated.
692
36
Fig. 2: Total number of visits to crop flowers by honey bees versus wild bees at each study farm. Points
37
above the 45º line indicate that wild bee visitation exceeded honey bee visitation.
693
38
694
Fig. 3: Nonmetric multidimensional scaling ordination of
study farms and crops according to bee species composition.
The ordination is based on the relative Sørensen index,
which separates sites according to proportional rather than
absolute species composition. Cumulative r2 = 0.76, stress =
21.8, instability = 0.005.
695
696
697
39
698
Supplementary Material
699
The following supplementary material is available online from www.Blackwell-
700
Synergy.com:
701
702
Appendix S1 List of bee species collected on watermelon Citrullus lanatus (Thunb.)
703
Matsum. & Nakai, tomato Solanum lycopersicon L., muskmelon Cucumis melo L., and
704
bell pepper Capsicum annuum L. crops.
705
706
Appendix S2 The estimated U.S. dollar value of insect pollination for selected crops in
707
New Jersey (NJ) and Pennsylvania (PA) in 2005.
708
40
709
Appendix S1: List of bee species collected on watermelon Citrullus lanatus (Thunb.)
710
Matsum. & Nakai, tomato Solanum lycopersicon L., muskmelon Cucumis melo L., and
711
bell pepper Capsicum annuum L. crops.
712
Genus species1
Watermelon
Agapostemon sericeus
Tomato
Muskmelon
Pepper
x
Agapostemon texanus
x
Agapostemon virescens
x
Andrena wilkella
x
x
Augochlora pura
x
x
x
x
Augochlorella aurata
x
x
x
x
Augochloropsis metallica
x
x
Bombus auricomus
x
Bombus bimaculatus
x
Bombus fervidus
x
Bombus griseocollis
x
x
x
x
Bombus impatiens
x
Bombus perplexus
x
Bombus vagans
x
Calliopsis andreniformis
x
Ceratina calcarata or dupla2
x
Ceratina strenua
x
41
x
x
x
x
x
Halictus confusus
x
x
x
Halictus ligatus
x
x
x
Halictus rubicundus
x
Hylaeus affinis
x
Hylaeus mesillae
x
Hylaeus modestus
x
Lasioglossum (Evylaeus) truncatum
x
Lasioglossum (Dialictus) admirandum
x
x
Lasioglossum (Dialictus) albipenne
x
x
Lasioglossum (Dialictus) bruneri
x
Lasioglossum (Dialictus) cattellae
x
x
x
x
Lasioglossum (Dialictus) coreopsis
x
x
Lasioglossum (Dialictus) cressonii
x
Lasioglossum (Dialictus) illinoense
x
Lasioglossum (Dialictus) imitatum
x
Lasioglossum (Dialictus) laevissimun
x
Lasioglossum (Dialictus) nymphaearum
x
Lasioglossum (Dialictus) oblongum
x
x
x
x
Lasioglossum (Dialictus) obscurum
x
x
Lasioglossum (Dialictus) pectinatum
x
Lasioglossum (Dialictus) pectorale
x
Lasioglossum (Dialictus) pilosum
x
42
x
x
Lasioglossum (Dialictus) platyparium
x
Lasioglossum (Dialictus) rohweri
x
x
Lasioglossum (Dialictus) tegulare
x
x
Lasioglossum (Dialictus) versatum
x
x
Lasioglossum (Dialictus) zephyrum
x
x
Lasioglossum coriaceum
x
Megachile brevis
x
Megachile mendica
x
Megachile rotundata
x
x
x
x
Melissodes bimaculata
x
x
Peponapis pruinosa
x
x
Ptilothrix bombiformis
x
Triepeolus remigatus
x
x
Xylocopa virginica
x
x
713
1
(Subgenus) given when taxonomic debate exists regarding subgenus versus genus status.
714
2
Females of these two species cannot be separated reliably.
43
x
715
Appendix S2: The estimated value insect pollination for selected crops in New Jersey (NJ) and Pennsylvania (PA) in 2005. These
716
estimates include only a subset of crops because records are only kept for the crops with the highest production value, generally
717
those for which a state contributes >1% of the national total. All crops are pollinated primarily by bees (both honey bees Apis
718
mellifera and wild bees); flies contribute a smaller amount of pollination for some crops (Klein et al., 2007).
Crop
State
Value of utilized
Reliance on
Reliance on
Value of insect
Value of insect
production, 2005
insect
insect
pollination, lower
pollination, upper
(millions of U.S.
pollination,
pollination,
bound (millions of
bound (millions of
dollars)1
lower bound2
upper bound
U.S. dollars)3
U. S. dollars)
Apple Malus domestica Borkh.
NJ
13.8
0.4
0.9
20.3
45.8
Apple
PA
50.9
0.4
0.9
5.5
12.4
Bell pepper Capsicum annuum L.
NJ
20.5
0
0.1
0
2.1
Cultivated blueberries4
NJ
55.4
0.4
0.9
22.2
49.9
Cantaloupe Cucumis melo L.
PA
4.1
0.9
1
3.7
4.1
Cranberry Vaccinium macrocarpon L.
NJ
18.1
0.4
0.9
7.3
16.3
Cucumber Cucumis sativus L.
NJ
9.7
0.4
0.9
3.9
8.7
44
Peach Prunus persica (L.) Batsch
PA
19.0
0.4
0.9
7.6
17.1
Peach
NJ
30.9
0.4
0.9
12.3
27.7
Pear Pyrus communis L.
PA
1.2
0.4
0.9
0.5
1.1
Pumpkin Cucurbita pepo L.
PA
16.1
0.9
1
14.5
16.1
Squash Cucurbita pepo L.
NJ
7.9
0.9
1
7.1
7.9
Strawberry Fragaria spp.
PA
12.8
0.1
0.4
1.2
5.1
Sweet cherry Prunus avium L.
PA
1.0
0.4
0.9
0.4
0.9
Tart cherry, Prunus cerasus L.
PA
0.8
0.4
0.9
0.3
0.7
Tomato Solanum lycopersicon L.
NJ
24.9
0
0.1
0
2.5
Tomato
PA
26.1
0
0.1
0
2.6
107.0
221.3
Total
719
1
From.USDA-NASS 2006a, 2006b.
720
2
From Klein et al. 2007, Appendix 2. Upper and lower bounds on reliance on insect pollination are based on experimentally-
721
determined reduction in fruit set when pollinators are excluded.
722
3
Calculated as the product of the dollar value and the reliance on insect pollination (Losey & Vaughan, 2006).
723
4
Predominantly highbush blueberry Vaccinium corymbosum L.
45
724
References, Supplementary Material
725
Losey, J.E. & Vaughan, M. (2006) The economic value of ecological services provided
726
727
by insects. Bioscience, 56(4), 311-23.
Klein, A.M., Vaissière, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A.,
728
Kremen, C. & Tscharntke, T. (2007) Importance of pollinators in changing landscapes
729
for world crops. Proceedings of the Royal Society of London, Series B, 274, 303-13.
730
731
732
733
USDA-NASS. (2006a). Non-citrus fruits and nuts: 2005 summary. United States
Department of Agriculture, National Agricultural Statistics Service, Washington, D.C.
USDA-NASS. (2006b). Vegetables: 2005 Summary. United States Department of
Agriculture, National Agricultural Statistics Service, Washington, D.C.
46