Impact of introduced honey bees on native pollination interactions of the endemic Echium wildpretii (Boraginaceae) on Tenerife, Canary Islands

Abstract

The aim of this study was to investigate effects of introduced honey bees (Apis mellifera) on native pollination interactions of Echium wildpretii ssp. wildpretii in the sub-alpine desert of Tenerife. We selected two study populations, one dominated by honey bees, while the other was visited by many native insects. During peak activity period of insects, nectar was nearly completely depleted in flowers of the first, but not the latter population. Thus, a high abundance of honey bees may have suppressed visitation by native animals due to exploitative competition. Honey bees stayed longer and visited more flowers on the same inflorescence than native bees, thus potentially promoting self-pollination of the plants. Level of seed set and viability was similar in the two study populations. However, we cannot rule out long-term changes in genetic population structure due to changes in gene-flow patterns caused by foraging behaviour of honey bees vs. native flower-visitors.

Full text

Impact of introduced honey bees on native pollination interactions
of the endemic Echium wildpretii (Boraginaceae) on Tenerife,
Canary Islands
Yoko L. Dupont
*
, Dennis M. Hansen, Alfredo Valido, Jens M. Olesen
Department of Ecology and Genetics, University of Aarhus, Ny Munkegade Building 540, Aarhus C 8000, Denmark
Received 11 June 2003; received in revised form 6 September 2003; accepted 20 September 2003
Abstract
The aim of this study was to investigate effects of introduced honey bees (Apis mellifera) on native pollination interactions of
Echium wildpretii ssp. wildpretii in the sub-alpine desert of Tenerife. We selected two study populations, one dominated by honey
bees, while the other was visited by many native insects. During peak activity period of insects, nectar was nearly completely de-
pleted in flowers of the first, but not the latter population. Thus, a high abundance of honey bees may have suppressed visitation by
native animals due to exploitative competition. Honey bees stayed longer and visited more flowers on the same inflorescence than
native bees, thus potentially promoting self-pollination of the plants. Level of seed set and viability was similar in the two study
populations. However, we cannot rule out long-term changes in genetic population structure due to changes in gene-flow patterns
caused by foraging behaviour of honey bees vs. native flower-visitors.
Ó2003 Elsevier Ltd. All rights reserved.
Keywords: Disruption of native mutualisms; Interspecific competition; Apis mellifera; Conservation
1. Introduction
In recent years the impact of introduced honey bees
(Apis mellifera L.) on native flora and fauna has been
much debated. Results of some studies indicate that
foraging patterns and abundance of native pollinators
are altered in the presence of honey bees (Roubik, 1978;
Schaffer et al., 1983; Sugden and Pyke, 1991; Paton,
1993; Wenner and Thorp, 1994; Vaughton, 1996; Gross
and Mackay, 1998; Gross, 2001; Hansen et al., 2002).
Although stressed as important by most researchers, it
has been difficult to investigate potential detrimental
effects of introduced honey bees on food storage or on
reproduction of native bee species (Roubik, 1983; Sug-
den et al., 1996; Butz Huryn, 1997; Steffan-Dewenter
and Tscharntke, 2000; Thorp et al., 2000). The impact of
honey bees on the pollination of native flora includes
effects on pollen dispersal and thus patterns of seed set
and genetic structure of plant populations. Honey bees
are often found to be less efficient pollinators compared
to native flower-visiting animals (Schaffer et al., 1983;
Taylor and Whelan, 1988; Westerkamp, 1991; Paton,
1993; Vaughton, 1996; Gross and Mackay, 1998; Han-
sen et al., 2002). However, other studies have found that
A. mellifera does not adversely affect plant reproductive
success, perhaps due to the numerical abundance of
honey bees compared to native bees (Vaughton, 1992;
Gross, 2001). Furthermore, effects of introduced honey
bees on native flora or fauna are often difficult to assess
due to a lack of suitable control sites (i.e., absence of A.
mellifera). Lastly, patterns induced by honey bees may
be swamped by demographic, stochastic, and environ-
mental variation.
Two features of island pollination networks leave
them susceptible to invasion by introduced generalist
species, such as honey bees: Low species diversity
(Kennedy et al., 2002) and the generalised nature of
interactions (Olesen et al., 2002). Several studies of is-
land ecosystems have reported a decline in both native
bee and bird species visiting flowers in the presence of A.
mellifera (Roubik, 1978; Kato, 1992; Wenner and
Biological Conservation 118 (2004) 301–311
www.elsevier.com/locate/biocon
BIOLOGICAL
CONSERVATION
*
Corresponding author. Fax: +45-86-127-191.
E-mail address: yoko.dupont@biology.au.dk (Y.L. Dupont).
0006-3207/$ - see front matter Ó2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biocon.2003.09.010

Thorp, 1994; Kato et al., 1999; Hansen et al., 2002).
Honey bees are widespread in the Canary Islands, and
bee-keeping has been practiced for centuries (M
endez,
2000). The colonisation of the islands by A. mellifera is
ancient, and honey bees are considered native based on
mitochondrial DNA data (De la R
ua et al., 1998).
However, A. mellifera is absent from the two eastern
arid islands, Fuerteventura and Lanzarote, where the
climate is dry and the flowering season too short to
support perennial colonies of bees (De la R
ua et al.,
2001). Similar climatic conditions prevail in the semi-
arid, sub-alpine desert zone, found in the mountain re-
gions above 2000 m a.s.l. on Tenerife. Combined with
the geographical isolation provided by both crater rim
and the surrounding pine forest, this suggests a natural
absence of A. mellifera in this habitat. Few, but very
distinct plant species inhabit these altitudes, exposed to
high irradiation, drought, strong winds and extreme
temperature ranges. Most of the native sub-alpine biota
is confined within Teide National Park (18,900 ha),
which is legally protected. However, apiculture is per-
mitted. Every year bee-keepers bring thousands of bee-
hives to the mountain areas of Tenerife during the short
sub-alpine summer (Anon., 2000).
Observations in 2000 and 2001 showed a marked
seasonal shift in the flower-visiting fauna of Echium
wildpretii ssp. wildpretii Pearson ex Hook. f. (Boragin-
aceae) in Teide National Park. In early season, native
passerine birds (Phylloscopus collybita (Vieillot) and
Serinus canarius L.) and native insects visited the red,
nectar-rich flowers. However, coinciding with a sudden
increase in honey bee activity, the birds stopped forag-
ing for nectar (Valido et al., 2002; unpubl. data from
2000). These observations indicate that the introduced
honey bees may alter native pollination interactions.
While the cessation of bird visits was easily observed,
effects on native insects were less obvious. The main
objective of this study is to investigate the relative im-
portance of native insects versus introduced honey bees
as flower-visitors and pollinators of E. wildpretii. Na-
tional Park staff controls the placement and numbers of
beehives, and we were thus able to obtain information
about temporal and geographical distribution of hives
throughout the flowering season.
The controlled framework of bee-keeping in Teide
National Park provides a man-induced ecological ex-
perimental setup, allowing us to assess the impact of
introduced honey bees on reproductive success, and
hence long-term persistence of E. wildpretii. We chose
two study populations, one population close to man-
aged beehives within the caldera and an adjacent pop-
ulation on the outside of the crater rim. We hypothesise
that honey bees, when present in large numbers, deplete
flowers of nectar, thus leading to exploitative competi-
tion with native flower-visitors. Moreover, differences in
foraging patterns of honey bees and natives could alter
pollen flow, and thus affect seed set or seed viability of
E. wildpretii. Therefore, we studied the following key
aspects of the pollination interactions in the two study
populations: (1) Diurnal and seasonal patterns of visi-
tation by native and introduced flower-visitors. (2)
Nectar secretion in animal-excluded flowers and flowers
open to visitation. (3) Patterns of seed set.
2. Materials and methods
2.1. Study system
The study was carried out in Teide National Park,
which is part of the region Las Ca~
nadas, a volcanic al-
luvial plain delimited by a crater rim. The National Park
(NP) covers an area of 18,900 ha and encompasses the
high-altitude sub-alpine zone of Tenerife, Canary Is-
lands. Vegetation is a low and sparse shrub dominated
by a few species.
The Teide bugloss, Echium wildpretii ssp. wildpretii
(Boraginaceae) (E. wildpretii hereafter) was the focal
species of our study. This endemic plant is almost ex-
clusively confined to the sub-alpine mountain zone of
Tenerife (Bramwell and Bramwell, 1990). Although this
subspecies is categorised as rare (G
omez Campo, 1996),
it is locally abundant and the large red-flowered inflo-
rescences are conspicuous in the landscape during the
flowering season. The plant is monocarpic and grows as
a rosette for 5–10 years before producing a single
flowering shoot. The columnar inflorescence is ca. 0.5–
2.5 m in height, a basal diameter of 10–70 cm, tapering
towards the top. Flowers are borne in cymes, which are
arranged spirally on the inflorescence. Cymes have a
total of four (in apical cymes) to 30 flowers (in basal
cymes) with 1–3 open flowers at a time. The flower is
protandrous and is open for 2.5–3 days (Olesen, 1988).
Each flower contains four ovules, developing into a
maximum of four nutlets. Individual plants flower for 3–
5 weeks, but since the time of flowering varies among
individuals, the total flowering period of a population is
ca. 1–1.5 months. In addition to the passerine birds and
honey bees, 16 native insect species and juveniles of the
endemic lacertid lizard Gallotia galloti Dum
eril &
Bibron have been observed visiting the flowers (Valido
et al., 2002).
2.2. Experimental design
Beehives are brought to the park during the flowering
season of the nectar-rich plant species, mainly Descu-
rainia bourgeauana Webb ex O.E. Schulz (Brassicaceae),
Spartocytisus supranubius (L.) Webb & Berth (Fabaceae)
and E. wildpretii. The timing of flowering varies between
years and is tracked by the bee-keepers. Generally, a
maximum of 2500 hives at 25 sites are allowed in the NP
302 Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311

each year. Each hive houses 10,000–70,000 bees. How-
ever, due to drought and hence scarcity of floral re-
sources, the actual number of beehives placed in the NP
was lower than the allowed maximum in 2001, the year
of our experiment. A total of 1393 beehives were placed
at 17 sites in Teide NP from April 27 to June 10 2001.
Beehives were removed again from May 25 and on-
wards. By July 12th 2001 all hives had been removed
(Fig. 1). Information about numbers of beehives placed
and removed at each site throughout the flowering sea-
son was obtained from Teide NP office. Due to the
limited area of the national park, all populations of E.
wildpretii inside the caldera are located few kilometres
from beehives, and thus visited heavily by honey bees.
Only E. wildpretii populations isolated from bee hives by
a geographical barrier were visited by fewer honey bees.
For this reason, we selected two study populations ca.
2 km apart, separated by the crater rim surrounding Las
Ca~
nadas.
Population 1 (hereafter Pop1) was located at ÔCe-
menterio de los TajinastesÕ(28°130N, 16°380W, 2050 m
a.s.l.), on the inner slope of the crater rim. Within 5 km
of Pop1, there were a total of 356 beehives in five lo-
cations. Pop1 was expected to be heavily exploited by
honey bees, as 90% of foraging trips by honey bees are
within a 5-km radius of the colony (Visscher and Seeley,
1982). Vegetation at the study site was dominated by S.
supranubius. Other abundant species were Scrophularia
glabrata Aiton (Scrophulariaceae), Pimpinella cumbrae
Buch ex DC (Apiaceae), Erysimum scoparium (Brouss.)
Wettst. (Brassicaceae), Tolpis webbii Sch. Bip. (Astera-
ceae), and E. wildpretii. Approximately 120 flowering
individuals of E. wildpretii were found at this site.
Population 2 (hereafter Pop2) was located at ÔValle de
los TajinastesÕ(28°120N; 16°370W, 2150 m a.s.l.), less
than 2 km from Pop1, but on the outer side of the crater
rim. This study site was isolated from nearby beehives in
the NP by a 400-m high mountain ridge. Moreover,
Pop2 is isolated from beehives at lower altitudes by pine
forest, a uniform vegetation type consisting of few spe-
cies. We have no information about bee-keeping activity
outside the NP in the nearby pine forest. However, feral
colonies are unlikely to persist in this zone due to the
arid conditions. Thus, Pop2 was predicted to have a low
visitation by A. mellifera. Vegetation was dominated by
S. supranubius and Carlina xeranthemoides L. fil (As-
teraceae). Due to severe drought only very few plant
individuals set flowers in 2001, and most species were
only observed in vegetative condition. E. wildpretii,
represented by 60–70 flowering individuals, was the
most conspicuously flowering species and the only
species offering large amounts of floral resources to
animals.
To compare plant characteristics of Pop1 and Pop2,
we measured height of flowering plants, size of inflo-
rescences (length, basal diameter and surface approxi-
mated to a cone), flower density, number of flowers per
cyme and total number of flowers per inflorescence.
Since number of flowers per cyme varied along the
length of the inflorescence, 20 cymes from each of the
lower, mid and upper part of an inflorescence were used
to calculate the average number of flowers per cyme.
Furthermore, we measured the distance to the three
nearest neighbouring conspecific flowering plants and
the size of these. Data from the two study populations
were compared by t-tests when assumptions of para-
metric tests were met. Otherwise the non-parametric
Mann–Whitney U-test was used.
2.3. Flower visitation
To investigate diurnal and seasonal patterns of visi-
tation, we measured visitation rate of native insects and
honey bees from sunrise (ca. 8:00 h) to sunset (ca. 21:00
h) from early to late in the flowering season. We ob-
served bird visits with a binocular from a hideout >10 m
away, this is treated in further detail in another paper,
Valido et al. (2002). In Pop1 insect visitation rates were
recorded on nine observation days (1–27 May) and in
Pop2 on five observation days (15–30 May) interspersed
regularly through the flowering season. To capture
variation due to differences in floral display, spatial lo-
cation and stage of flowering between individual plants,
visitation rate was recorded alternately in 20 study
plants in Pop1 and in six plants in Pop2. Each plant was
observed for 10-min periods consisting of five 2-min
intervals. Depending on the level of visitation, one side
of the whole inflorescence, a proportion (1/2, 1/3 or 1/4)
or a square of 10 10 cm was observed. At the begin-
ning of each 2-min census period, all insect individuals
present on the observed part of the inflorescence were
Fig. 1. Columns indicate total number of beehives within the area of
Teide National Park during 10-day periods (days 1–10, 11–20, etc.).
Number of hives within 5 km of Pop1 is shown in black. Day 1 cor-
responds to April 27, 2001. The last beehives were removed on day 77 –
July 12, 2001.
Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311 303

counted, and during the 2 min all newcomers were ad-
ded to this number. One visit was defined as an indi-
vidual landing on the inflorescence foraging for nectar
and/or pollen until its departure from the inflorescence.
Since insect individuals were not marked, repeated visits
by the same individual were counted as new visits. Thus,
this census method could lead to an overestimate of
visitation rate especially in the 10 10 cm squares, since
an insect individual may arrive and depart from the
square several times during a single visit to the whole
inflorescence. For this reason we made simultaneous
measurements by two observers, recording visitation
rate in a whole (or half) inflorescence and in 10 10 cm
squares in 32 2-min periods. Based on these data we
constructed regression equations, which were subse-
quently used to convert visitation rate measured in
10 10 squares to visitation rate of the whole inflores-
cence. Visitation rates recorded in 1/2, 1/3 or 1/4 of an
inflorescence were multiplied by 2, 3 and 4, respectively,
to estimate visitation rate per plant. When insects were
counted only in a proportion of the inflorescence, we
alternately observed lower, mid and upper parts of the
inflorescences. Spatial differences in visitation rate were
investigated using seven different plants in Pop2, for
which visits to the upper and lower halves of the inflo-
rescence were recorded. During all visitation observa-
tion periods, we monitored air temperature at the
surface of the inflorescence at mid-inflorescence height.
To investigate differences in foraging behaviour be-
tween honey bees and native insects, we observed the
behaviour of individual honey bees, Anthophora alluaudi
Perez and Eucera gracilipes Perez (Apidae). These two
endemic anthophorid type bees were the most abundant
native visitors. For each bee individual we recorded the
time spent and the number of flowers probed per visit to
an inflorescence.
2.4. Nectar
To investigate diurnal patterns of nectar secretion
and the influence of flower-visitors on nectar availability
in E. wildpretii, we measured nectar volumes using mi-
cro-capillary tubes and sugar concentration using a
hand-held Bellingham–Stanley refractometer. Sampling
was done during peak flowering season, when the ac-
tivity of flower-visiting animals was high (May 13–14 in
Pop1, 15 and 17 May in Pop2). Two adjacent plants of
approximately equal size and stage of flowering were
selected in each study population. The inflorescence of
one plant was covered with fine nylon mesh prior to the
activity period of flower-visitors to exclude animals
(‘‘excluded’’). Flowers of the other plant were left non-
manipulated to be visited by flower-visiting animals
(‘‘open’’). Nectar volume per flower and sugar concen-
tration were measured from sunrise (8:00 h) to sunset
(21:00 h). Sugar weight in sucrose-equivalents was cal-
culated as volume concentration. In each sampling
period, nectar characteristics were measured in 10
flowers of each treatment, five in male and five in female
phase. The light regime (sun or shade) of the sampled
flowers was also recorded. When possible, equal num-
bers of flowers in sun and shade were sampled in each
period. Flowers were removed from the inflorescence
after sampling to avoid repeated sampling. An estimate
of the total lifetime nectar amount offered by an E.
wildpretii individual was calculated as total nectar pro-
duction per plant, based on number of flowers per plant
(using plants of average size and maximum size, re-
spectively), average nectar volume per flower per day
and an average nectar production period of 2 days per
flower.
2.5. Seed set and viability
After flowering had ended and fruit development
started, the level of seed set was recorded in 22 plants in
Pop1 and 12 plants in Pop2. In each plant, the infruct-
escence was divided into three parts (lower, mid and
upper). In each part, seed set per flower was counted in
flowers of 20 randomly selected cymes. Seed set was
calculated as a percentage of maximum seed set, which is
four seeds per flower. To assess the importance of ani-
mal pollinators for seed set, we excluded animal visitors
by bagging whole inflorescences of six plants in Pop1
and four in Pop2 before flowering. Using metal wire and
fine nylon mesh, we constructed cages around each in-
florescence, to avoid the mesh from touching (and thus
pollinating) the flowers. At the end of the flowering
season, cages and bags were removed and seed set re-
corded in 60 randomly selected cymes per plant.
Mature seeds from 19 plants of Pop1 and 14 plants of
Pop2 were collected and tested for viability using a 2,3,5
triphenyl tetrazolium chloride (TTC) enzyme activity
test (Heydecker, 1965, 1968). For each plant, 30 seeds
were collected from the lower, mid and upper parts of the
infructescence. Seeds were cut into halves, placed in petri
dishes with the 2,3,5 TTC solution and left in darkness
for 24 h. Seeds were considered viable if the embryo
turned red, a reaction indicating enzyme activity.
2.6. Data analysis
Average visitation rate per 2-min period was calcu-
lated for each 10-min census period for both honey bees
and native insects. Diurnal patterns were analysed by
dividing visitation rate observations into 2-h periods
from 7:00 to 21:00 h, pooling data from all observation
dates.
To investigate diurnal changes in nectar availability
of open and excluded flowers, the mean volume and
concentration in ten flowers sampled at each time
were used to represent nectar characteristics (volume,
304 Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311

concentration and sugar weight) at a given time of the
day. All analyses were run separately for Pop1 and
Pop2.
To analyse patterns of variation in final seed set (log
transformed) at different levels, we carried out an AN-
OVA (GLM procedure) with the levels: populations,
plants within populations, infructescence parts (lower,
mid and upper) within plants and cymes within parts.
All these variables were included in the model as ran-
dom effects and the Type III Sum of Squares were cal-
culated (Shaw and Mitchell-Olds, 1993).
3. Results
3.1. Plant characteristics
Although plants of Pop1 and Pop2 did not differ in
total height, inflorescences of plants in Pop2 were sig-
nificantly shorter, and thus had a smaller surface area.
Moreover, flower density was lower in Pop2, and plants
of this population generally had fewer flowers per in-
florescence per day compared to Pop1, although the
difference was not significant. Furthermore, number of
flowers per cyme was significantly lower in Pop2 (Table
1). Hence, in conclusion, plants in Pop2 had a lower
number of flowers, both on a per day and on a seasonal
basis. Spatial patterns of plants (nearest neighbours) did
not differ between Pop1 and Pop2, although neigh-
bouring plants in Pop2 were smaller, reflecting the
general difference in plant size between populations
(Table 1).
3.2. Flower visitation
A total of 16 native insect species and the introduced
A. mellifera were observed visiting E. wildpretii (see
Table 1 in Valido et al., 2002). Of these, the most
dominant species were A. mellifera,A. alluaudi and E.
gracilipes (Apidae). Diurnal visitation patterns of both
honey bees and native insects were strongly influenced
by temperature in both study populations: Visitation
started at sunrise (when temperatures were 10–11 °C)
and increased steadily until ca. 11:00 h, when the air
temperature levelled off at ca. 23–24 °C (Figs. 2 and 3).
Insect activity decreased again around 18:00 h, when the
temperature decreased. A marked difference in compo-
sition of the flower-visitor pool was found between Pop1
and Pop2. In Pop1, no significant difference was found
in levels of visitation by honey bees and native insects in
the early season (before the 8th of May). However, in
mid and late season (after the 8th of May), visitation by
honey bees increased suddenly, significantly exceeding
visitation by native insects (Fig. 4). This increase was
coincident with an increase in numbers of beehives
placed within the National Park, including the sites close
to Pop1 (Fig. 1). In contrast, in Pop2, level of visitation
by native insects was higher than that of honey bees
both diurnally (Fig. 3) and throughout the season (al-
though only significantly so in mid season) (Fig. 4).
Honey bees, A. alluaudi and E. gracilipes, differed in
foraging patterns, both in time spent per inflorescence
(ANOVA: N¼95, F¼31:31;P<0:0001, R2¼0:40)
and number of flowers probed per visit (ANOVA:
N¼95, F¼21:59, P<0:0001, R2¼0:32) (Table 2).
Table 1
Plant characteristics for E. wildpretii at the two study sites
Characteristic Pop1 Pop2 Statisticsc
Total plant height (cm) 188.5 34.9 (55) 182.5 34.6 (22) t¼0:61n:s:
Height of inflorescence (cm) 139.6 33.8 (55) 111.4 26.0 (25) t¼3:84
Basal diameter of inflorescence (cm) 28.8 11.3 (55) 27.6 6.7 (25) t¼1:10n:s:
Surface of inflorescence (cm2) 6795.3 4236.6 (55) 5087.9 2424.9 (25) t¼3:05
Density of flowers (no. of flowers/cm2) 0.40 0.09 (54) 0.32 0.08 (10) t¼2:89
Open flowers per plant per day 3301 2241 (20) 1836 762 (10) t¼1:97n:s:
Total number of flowers per cymea17.3 4.7 (1320) 15.5 5.5 (660) U¼7:21
Open flowers per cyme per day – 1.76 0.52 (250) –
Distance to nearest neighbour (m) 6.0 8.1 (22) 13.8 13.6 (12) U¼1:27n:s:
Mean distance to three nearest neighbours (m) 9.1 7.6 (22) 16.5 13.9 (12) U¼1:32n:s:
Surface of nearest neighbour (cm2) 4750.9 3877.9 (22) 5942.6 2370.1 (12) t¼2:43
Nectar volume per flower per day (ll)b9.79 4.61 (111) 4.99 2.13 (80) U¼7:95
Male phase flowers 8.73 3.38 (55) 4.88 1.40 (40) t¼7:35
Female phase flowers 10.82 4.77 (56) 5.09 2.69 (40) U¼5:91
Nectar sugar concentration (%) 15.5 4.1 (111) 15.3 3.3 (76) t¼0:09n:s:
Male phase flowers 13.4 1.9 (55) 14.0 3.2 (36) t¼0:89n:s:
Female phase flowers 16.7 4.6 (56) 16.6 2.9 (40) U¼1:22n:s:
All values are given as means SD (N).
a
Mean of 20 cymes from each of the upper, middle and lower parts of the inflorescence.
b
Excluded flowers, based on data from May 13 to 14 (Pop1) and May 15 and 17 (Pop2).
c
Pop1 and Pop2 were compared using t-tests when data conformed to assumptions of parametric tests, and the non-parametric Mann–Whitney U-
test otherwise. Significance level: ,P<0:005,  ,P<0:0005, n.s., P>0:05.
Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311 305

Pair-wise comparisons revealed that honey bees spent
significantly more time and visited significantly more
flowers per inflorescence than the native bees. However,
A. alluaudi and E. gracilipes did not differ in foraging
pattern (Tukey–Kramer test). Furthermore, native in-
sects visited the upper part of the inflorescence much
more frequently than the lower part (Wilcoxon signed
ranks test, visitation rate of upper versus lower:
U¼141:5;P<0:001, N¼28). In contrast, visitation
rate of honey bees was higher in the lower part than in
the upper part of an inflorescence (t-test: t¼2:31,
P¼0:029, N¼28).
Two native species of passerine birds, the Common
chiff-chaff (P. collybita) and the Canary (S. canarius),
were commonly observed visiting the flowers for nectar
in early season in both populations. On one occasion we
observed a pair of another passerine, the Blue tit (Parus
caerulius), visiting six inflorescences for nectar in Pop2.
Bird visits continued occasionally in Pop2 throughout
the flowering season. However, after May 8 no bird
visitors were observed in Pop1. The disappearance of
birds was coincident with the increase in honey bee ac-
tivity (Fig. 4).
3.3. Nectar
In Pop1, nectar volume of excluded flowers remained
at a constant level throughout the day (regression:
6 10141822
VISITORS / PLANT / PERIOD
0
1
2
3
4
5
6 10 14 18 22
0
1
2
3
4
5
6 10141822
0
20
40
60
80
100
6 10141822
0
20
40
60
80
100
EARLY SEASON
LATE SEASON
Fig. 2. Diurnal variation in visitation rates (individuals/2 min/inflorescence) of native insects and honey bees in Pop1 in early (1–8 May) and late
season (8–27 May). Notice differences in levels of visitation between early and late season (different scales of the Y-axis).
6 10 14 18 22
VISITORS / PLANT / PERIOD
0
20
40
60
80
100
6 10141822
0
20
40
60
80
100
Fig. 3. Diurnal variation in visitation rates (individuals/2 min/inflorescence) of native insects and honey bees in Pop2.
306 Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311

lnðvolÞ¼2:00–0.25 h, F¼0:03, P¼0:87, R2¼0:002).
In contrast, nectar volume of open flowers decreased
significantly during the day (regression: lnðvolÞ¼3:99–
0.26 h, F¼7:67, P¼0:01, R2¼0:34). Rapid decrease
to near-zero level occurred from sunrise to ca. 12:00 h,
after which nectar volume remained at a constant low
level until ca. 20:00 h, where a slight increase was ob-
served (Fig. 5). Hence, sugar weight of open flowers was
significantly lower than the level found in excluded
flowers from 09:50 h onwards (Mann–Whitney U-tests,
all P<0:05). In Pop2, nectar volume of excluded
flowers tended to increase slightly during the day (re-
gression: vol ¼)0.06 + 0.46 h, F¼5:61, P¼0:06,
R2¼0:48), albeit only to about half the level of that
found in Pop1 (Table 1). In open flowers, nectar volume
was constant throughout the day (regression: vol ¼3.65
– 0.07 h, F¼0:15, P¼0:71, R2¼0:024) (Fig. 5).
Overall, average values in the four groups (excluded and
open in Pop1 and Pop2, respectively) illustrate the ef-
fects of drought and nectar exploitation by visitors:
Differences in nectar level between excluded flowers of
Pop1 and Pop2 can be explained by extreme drought in
Pop2, while differences between open flowers reflect ex-
ploitation by a flower-visitor fauna dominated by honey
bees (Pop1) versus native insects (Pop2) (Fig. 5, dashed
lines).
Nectar secretion in excluded flowers in Pop1 was
significantly influenced by their sexual phase: nectar of
flowers in female phase had a higher sugar concentra-
TIME (h)
6 10141822
NECTAR VOLUME ( l)
0
4
8
12
16
OPEN FLOWERS
EXCLUDED FLOWERS
TIME (h)
8101214
NECTAR VOLUME ( l)
0
4
8
12
16
POP1 - EXCL
POP2 - EXCL
POP2 - OPEN
POP1 - OPEN
TIME (h)
6 10141822
NECTAR VOLUME (µl)
0
4
8
12
16
OPEN FLOWERS
EXCLUDED FLOWERS
TIME (h)
8101214
NECTAR VOLUME (µl)
0
4
8
12
16
POP1 - EXCL
POP2 - EXCL
POP2 - OPEN
POP1 - OPEN
Fig. 5. Diurnal variation in nectar volume of animal-excluded and open (non-excluded) flowers of E. wildpretii. Each point represents the mean of 10
flowers. Dashed lines indicate the mean daily levels of nectar in the four groups, plants with excluded and open flowers in Pop1 and Pop2, re-
spectively. Mean of open flowers in Pop1 was calculated excluding data before 10:00 h, when visitation was low.
*
*n.s. *** ***
1 6 11 16 21 26 31
0
10
20
30
40
50
** *
n.s. n.s.
(a)
(b)
1 6 11 16 21 26 31
VISITATION RATE
(VISITORS / PLANT / PERIOD)
0
10
20
30
40
Apis mellifera
Native insects
Fig. 4. Seasonal visitation pattern. Daily averages + SD of visitation
rates of honey bees and native insects in (a) Pop1 and (b) Pop2. Wil-
coxon signed ranks tests were used for comparisons. Significant dif-
ferences are indicated (* for P<0:05, n.s. for non-significance).
Numbers on the X-axis indicate dates in May 2001.
Table 2
Foraging patterns of two native bee species and the introduced honey
bee visiting Echium wildpretii
Species NSeconds/visit Flowers/visit
Eucera gracilipes 62 16.5 23.8 6.7 8.2
Anthophora alluaudi 13 18.7 10.8 9.9 9.5
Apis mellifera 20 137.1131.7 34.5 29.6
Values are given as means SD.
Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311 307

tion, larger volume, and hence higher sugar weight than
flowers in male phase (Mann–Whitney U-tests, all
P<0:0001) (Table 1). Sugar weight decreased during
the course of the day in male phase flowers and in-
creased steadily in female phase flowers (Fig. 6). In
Pop2, female phase flowers also had a larger sugar
concentration (U¼3:55, P<0:001, N¼76) and sugar
weight (U¼1:99, P<0:05, N¼80) than male phase
flowers. However, differences in nectar volumes were not
significant (U¼1:12, P¼0:26, N¼80). Light regime
(sun or shade) did not affect any nectar characteristics of
flowers in Pop1 or Pop2 (Mann–Whitney U-tests, all
P>0:05).
Total lifetime nectar production differed greatly be-
tween populations and among plants within each pop-
ulation. In Pop1, an average-sized plant produced an
estimated amount of 0.54 L during its lifetime, while in
Pop2 this amount was 0.16 L. However, inter-plant
variation was considerable due to variation in size. The
largest plant in Pop1 was estimated to produce 1.75 L,
and in Pop2 the largest lifetime production of a plant
was 0.43 L.
3.4. Seed set and viability
Seed set per flower per plant was slightly higher in
Pop2 than Pop1 (Table 3), although the difference was
not significant (Table 4). Considerable variation in seed
set was found between plant individuals within the
populations, and seed set was positively correlated with
plant height (Pearson r¼0:42, P¼0:01, N¼34).
However, seed set did not differ between populations
when using plant height as a covariate (F¼3:12,
P¼0:11). Furthermore, all the interactions terms cal-
culated in the GLM analysis were highly significant, i.e.,
seed set varied among plants within populations, among
parts of the infructescence within single plants and
among cymes within parts of an infructescence (Table
4). Seed set of infructescences, which had been excluded
from flower-visiting animals, revealed that plants set
some seeds even in the absence of animal pollen vectors
(Table 3). However, seed set was much lower than in the
Table 4
Results of the ANOVA analysis (GLM procedure using Type III Sum of Squares) of seed set (log transformed) at different levels (population, plant
individual, part of the inflorescence and cyme)
Source of variation SS df MS FP
Population 0.49 1 0.49 0.15 0.709
Plant 93.57 20 4.68 1.47 0.263
Part 1.4 2 0.70 6.94 0.002
Cyme 0.72 19 0.04 0.96 0.527
Population Plant 34.14 10 3.41 129.81 <0.001
Plant Part 4.21 40 0.10 4.00 <0.001
Part Cyme 1.53 38 0.04 1.53 0.020
Error 882.47 33,556 0.03
All variables were treated as random effects. The obtained model was statistically significant (F¼2912:8; P¼0:008).
6 10141822
0
1
2
3
6 10141822
SUGAR WEIGHT PER FLOWER
(µg SUCROSE EQUIVALENTS)
0
1
2
3
MALE PHASE FEMALE PHASE
Fig. 6. Diurnal variation in sugar weight of nectar of excluded flowers in male and female phase. Each point represents the mean of five flowers. Data
were only obtained for Pop1.
Table 3
Seed set and viability in open pollinated flowers, and flowers excluded
from flower-visitors
Parameter Pop1 Pop2
Seed setb, open 48.0 27.1 (22,732) 54.2 23.7 (10,955)
Seed setb, excluded 23.3 27.2 (6020) 29.6 30.7 (3744)
Seed viabilityc, open 84.4 9.1 (57) 83.9 10.2 (48)
Means SD (N)a.
a
Seed viability N, number of viability tests. Seed set N, number of
flowers.
b
Percentage of maximum seed set (four seeds per flower).
c
Percentage viable seeds in viability tests of 30 seeds.
308 Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311

controls for both Pop1 (U¼58:71, P<0:00005,
N¼28;752) and Pop2 (U¼44:26, P<0:00005,
N¼14;695).
There were no significant differences in seed viability
between populations (Table 3). Viability differed signif-
icantly among plants in Pop2 (ANOVA: df ¼15,
F¼3:84, P¼0:0007, R2¼0:64), but not in Pop1
(df ¼18, F¼1:68, P¼0:09, R2¼0:44).
4. Discussion
4.1. Impact of honey bees on native insects and birds
The potential impact of introduced A. mellifera in
native systems is a major concern, but few studies pro-
vide quantitative data and clear-cut evidence of the ef-
fects of honey bees (Sugden et al., 1996; Butz Huryn,
1997; Kearns et al., 1998). Results of this study show
that the site dominated by honey bees is characterised by
low visitation rate by native insects throughout the
season. In contrast, a high level of visitation by native
insects is maintained during the flowering season in
Pop2, which had a lower level of honey bee visitation.
The high abundance of honey bees may have resulted in
exploitative competition between honey bees and native
insects, as nectar standing crops were reduced to near-
zero levels in Pop1, but not in Pop2, despite the more
arid conditions prevailing there (Fig. 5). On the other
hand, the level of visitation by native insects remained
constant throughout the flowering season in Pop1, and
thus appeared to be unaffected by the emergence of
honey bees (Fig. 4). Two alternative scenarios may ex-
plain this pattern: (1) abundance of native insects was
limited by low temperatures in early season, and later
suppressed by honey bee dominance or (2) because ex-
tensive honey bee keeping has been practiced for cen-
turies in Las Ca~
nadas (M
endez, 2000), we may be
observing the Ôghost of competition pastÕ, i.e., numbers
of native flower-visitors may already have been reduced
by honey bee dominance in Pop1. It would be interest-
ing to address this question in a future study.
Exploitative competition may also be acting between
honey bees and passerine birds. In two consecutive years
we observed that nectar-feeding birds ceased flower-
visitation in Pop1 when honey bees became abundant.
At this stage it is not possible to determine if this pattern
is connected to depletion of nectar or to a seasonal
change in the diet of birds related to, e.g., availability of
insects or breeding activity. E. wildpretii is one of the
largest nectar resources in the otherwise very dry and
sparsely vegetated environment of Las Ca~
nadas. Thus,
overexploitation of E. wildpretii could force native
flower-visitors to switch to other, less profitable floral
resources. Furthermore, preliminary observations of
flower-visiting insects in a population of E. wildpretii
ssp. trichosiphon on La Palma Island revealed a lower
abundance of A. mellifera, higher levels of standing
nectar crop and a higher diversity of native insects
compared to the honey bee dominated population on
Tenerife (unpubl. data).
As mentioned in Section 1, other studies have shown
that introduced honey bees negatively affect visitation
rate and species diversity of native flower-visitors (Kato,
1992; Wenner, 1993; Kato et al., 1999; Wenner et al.,
2000; Gross, 2001; Hansen et al., 2002). An interesting
question is whether this pattern translates into reduced
reproductive output of native insects, and hence
threatens their long-term persistence in Teide National
Park. In a long-term study, Roubik and Wolda (2001)
found no decreases in population size of native insects in
the presence of africanised honey bees in a Central
American rain forest. On the other hand, an experi-
mental study in an Australian tree-grass plain showed
population declines of the native bee Exoneura asimil-
lima in the presence of managed bee hives, possibly due
to resource competition with honey bees (Sugden and
Pyke, 1991). Resource level has also been shown to af-
fect reproduction in the leaf-cutter bee Megachile api-
calis (Kim, 1999). Obviously, the long-term impact of A.
mellifera on a native pollination system varies between
regions and habitat types, and thus extrapolation of
above results to desert areas like Las Ca~
nadas should be
made with extreme caution (Paton, 1993). More long-
term studies are clearly necessary to assess the impact of
A. mellifera at the population level of native flower-
visitors.
4.2. Impact of honey bees on E. wildpretii
Introduced honey bees are known to reduce fitness of
some native plant species (Gross and Mackay, 1998).
However, in other cases, seed set has been shown to be
unaffected (Vaughton, 1992) or pollination even aug-
mented by the presence of honey bees (Paton, 1993;
Gross, 2001). In our study, seed set was not significantly
different in the study population dominated by honey
bees and the population visited predominantly by native
insects, the level of seed set being only slightly lower in
the former. Hence, the effect of A. mellifera on seed
production is minor, if any. Some studies have shown
that honey bees are poorer pollinators than native spe-
cies (Westerkamp, 1991; Freitas and Paxton, 1998;
Gross and Mackay, 1998; Hansen et al., 2002), but a
high abundance may compensate for lowered pollina-
tion efficiency (Butz Huryn, 1995; Kraemer and Schmitt,
1997; England et al., 2001). Moreover, breeding system
of the plant influences level of seed set. E. wildpretii was
capable of producing a considerable number of seed
without being visited by animals, and thus may to some
extent be pollinated by wind or even set seeds by au-
togamy. On the other hand, open pollination increases
Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311 309

seed set, and most aspects of anthesis seem to be related
to animal pollination: Most flowers open in the morn-
ing, live for 2.5 days, with approximately 1 day in male
phase and 1 day in female phase (Olesen, 1988). Our
study showed that flowers had two peaks in sugar
weight of nectar, one in male phase and one in female
phase, separated by a minimum during the sexual
transition phase (Fig. 6). Thus, flowers are most at-
tractive to flower-visitors during peak pollen presenta-
tion and again during stigma receptivity.
It is difficult to assess the influence of animal visita-
tion for variation in seed set, superimposed upon the
background reproductive output by spontaneous au-
togamy and wind pollination. Our analyses indicate a
positive correlation between plant size and seed set,
which is concordant with a preference of animals to visit
larger plants (Valido et al., 2002). However, the same
pattern could be explained by resource allocation in
semelparous organisms, larger plants having accumu-
lated more resources before reproduction.
4.3. Future perspectives
Although level of seed set can be slightly affected by
foraging patterns of the visitors, the primary impact of
introduced honey bees may be changes in pollen flow,
and thus genetic structure of the plant population. In
contrast to native insects and birds, honey bees visited
many flowers on each inflorescence and rarely moved
between plants, which is likely to promote geitonogamy
and hence inbreeding. However, how and if this affects
long-term persistence of the plant population is un-
known. We found no differences in seed viability be-
tween plants pollinated mainly by honey bees or native
insects. One explanation could be that E. wildpretii is
relatively unaffected by inbreeding, since it is capable of
producing a large number of seeds by selfing alone (up
to 50% of the seed set level of open pollinated plants).
Plants in Las Ca~
nadas have been pollinated by honey
bees since these were introduced in the 16th century
(M
endez, 2000). Thus, honey bees have influenced the
pattern of pollen transfer in 80–100 plant generations,
which may have contributed to purging of deleterious
alleles expressed through inbreeding. On the other hand,
inbreeding effects in life stages other than seed viability
cannot be ruled out. In a close relative, Echium vulgare
L., no effect of selfing was found at the stage of seed
production (Mensler et al., 1997). Yet, late-acting in-
breeding depression in male and female function of
offspring derived from selfing has been reported
(Mensler et al., 1999). Future studies should address
patterns of genetic variation in populations of E.
wildpretii pollinated predominantly by introduced honey
bees versus those pollinated by native animals. For in-
stance, in Grevillea macleayana (Proteaceae), which is
visited mainly by native birds and introduced honey
bees, outcrossing rate was reduced significantly when
birds were excluded from the inflorescences (England
et al., 2001). Furthermore, an interesting question is the
role of birds as pollinators and potential long-distance
pollen vectors of E. wildpretii in early flowering season,
before the onset of bee-keeping activities.
Many studies call for further investigation of the ef-
fects of honey bees on the reproductive output of plants
and long-term persistence of native flower-visiting animal
populations. However, impact of honey bees are difficult
to disentangle from confounding biotic and abiotic fac-
tors. Las Ca~
nadas offers a unique and simple study sys-
tem. The total flowering season of the sub-alpine desert
system is short (ca. 2 months) and the plant–flower-visi-
tor network is simple, consisting of a few species isolated
by the crater rim and surrounding pine forest (Dupont
et al., 2003). Furthermore, placement of beehives can be
controlled both spatially (as in Paton, 1993) and tem-
porally, creating gradients of honey bee visitation pres-
sure over the season and between plant populations.
Acknowledgements
We are grateful to M. Durb
an, A. Ba~
nares and J.
Rever
on (Teide National Park, Tenerife), and A. Palo-
mares and A. Revol
e (La Caldera de Taburiente Na-
tional Park, La Palma) for their collaboration, for
information about the bee keeping activities in the Na-
tional Parks, for research permissions and for letting us
stay at the field stations. We also thank C. Skov, E. P
ıo
and E. Portellano for help in the field and for fruitful
discussions. A. Sølling carried out the viability tests. The
manuscript was improved through comments and sug-
gestions by C. Rasmussen. This project received finan-
cial support from Mr. and Mrs. FiedlerÕs grant (to Y.L.
Dupont) the Augustinus Foundation (to Y.L. Dupont)
and the Danish National Science Research Council (to
J.M. Olesen). During the writing of this paper A. Valido
was supported by a Marie Curie individual fellowship
(MCFI-2000-1995).
References
Anon., 2000. Normas que regulan la actividad ap

ıcola en el Parque
Nacional del Teide. In: Ministerio de Medio Ambiente. Parque
Nacional del Teide.
Bramwell, D., Bramwell, Z., 1990. Flores silvestres de las Islas
Canarias. Rueda, Madrid.
Butz Huryn, V.M., 1995. Use of native New Zealand plants by honey
bees (Apis mellifera L.): a review. New Zealand Journal of Botany
33, 497–512.
Butz Huryn, V.M., 1997. Ecological impacts of introduced honey bees.
Quarterly Review of Biology 72, 275–297.
De la R
ua, P., Gali
an, J., Serrano, J., Moritz, R.F.A., 2001. Genetic
structure and distinctness of Apis mellifera L. populations from the
Canary Islands. Molecular Ecology 10, 1733–1742.
310 Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311

De la R
ua, P., Serrano, J., Gali
an, J., 1998. Mitochondrial DNA
variability in the Canary Islands honeybees (Apis mellifera L.).
Molecular Ecology 7, 1543–1547.
Dupont, Y.L., Hansen, D.M., Olesen, J.M., 2003. Structure of a plant–
flower-visitor network in the high-altitude sub-alpine desert of
Tenerife, Canary Islands. Ecography 26, 301–310.
England, P.R., Beynon, F., Ayre, D.J., Whelan, R.J., 2001. A
molecular generic assessment of mating-system variation in a
naturally bird-pollinated shrub: contributions from birds and
introduced honeybees. Conservation Biology 15, 1645–1655.
Freitas, B.M., Paxton, R.J., 1998. A comparison of two pollinators:
the introduced honey bee Apis mellifera and an indigenous bee
Centris tarsata on cashew Anacardium occidentale in its native
range of NE Brazil. Journal of Applied Ecology 35, 109–121.
G
omez Campo, C., 1996. Libro rojo de especies vegetales amenazados
de las Islas Canan_
as, Viceconsejer
ıa de Med

ıo Ambiente del
Gobierno de Canan_
as.
Gross, C.L., 2001. The effect of introduced honeybees on native bee
visitation and fruit-set in Dillwynia juniperina (Fabaceae) in a
fragmented ecosystem. Biological Conservation 102, 89–95.
Gross, C.L., Mackay, D., 1998. Honeybees reduce fitness in the
pioneer shrub Melostoma affine (Melostomataceae). Biological
Conservation 86, 169–178.
Hansen, D.M., Olesen, J.M., Jones, C.G., 2002. Trees, birds and bees
in Mauritius: exploitative competition between introduced honey
bees and endemic nectarivorous birds. Journal of Biogeography 29,
721–734.
Heydecker, W., 1965. Report of the rigour test committee (1962–1963).
Proceedings of the International Seed Testing Assay 30, 369–379.
Heydecker, W., 1968. Report of the rigour test committee (1965–1968).
Proceedings of the International Seed Testing Assay 34, 751–773.
Kato, M., 1992. Endangered bee fauna and its floral hosts in the
Ogasawara Islands. Japanese Journal of Entomology 60, 487–494.
Kato, M., Shibata, A., Yasui, T., Nagamasu, H., 1999. Impact of
introduced honeybees, Apis mellifera, upon native bee communities
in the Bonin (Ogasawara) Islands. Researches on Population
Ecology 41, 217–228.
Kearns, C.A., Inouye, D.W., Waser, N.M., 1998. Endangered mutu-
alisms: the conservation of plant–pollinator interactions. Annual
Review of Ecology and Systematics 29, 83–112.
Kennedy, T.A., Naeem, S., Howe, K.M., Knops, J.M.H., Tilman, D.,
Reich, P., 2002. Biodiversity as a barrier to ecological invasion.
Nature 417, 636–638.
Kim, J.-Y., 1999. Influence of resource level on maternal investment in
a leaf-cutter bee (Hymenoptera: Megachilidae). Behavioral Ecology
10, 552–556.
Kraemer, M., Schmitt, U., 1997. Nectar production patterns and
pollination of the Canarian endemic Echium wildpretii Pearson ex
Hook. fil. Flora (Jena) 192, 217–221.
M
endez, T., 2000. Antecedentes historicos del Teide y Las Canadas,
Canarina, La Orotava, Tenerife.
Mensler, C., Bijleveld, A., Klinkhamer, P.G.L., 1999. Late-acting
inbreeding depression in both male and female function of Echium
vulgare (Boraginaceae). Heredity 83, 162–170.
Mensler, C., Rademaker, M.C.J., Klinkhamer, P.G.L., 1997. Selection
on pollen donors by Echium vulgare (Boraginaceae). Sexual Plant
Reproduction 10, 305–312.
Olesen, J.M., 1988. Floral biology of the Canarian Echium wildpretii:
bird–flower or a water resource to desert bees? Acta Botanica
Neerlandica 37, 509–513.
Olesen, J.M., Eskildsen, L.I., Venkatasamy, S., 2002. Invasion of
oceanic island-pollination networks: importance of invader com-
plexes and endemic super generalists. Diversity and Distributions 8,
181–192.
Paton, D.C., 1993. Honeybees in the Australian environment. BioSci-
ence 43, 95–103.
Roubik, D.W., 1978. Competitive interactions between neotropical
pollinators and africanized honey bees. Science 201, 1030–1032.
Roubik, D.W., 1983. Experimental community studies: time-series
tests of competition between African and neotropical bees. Ecology
64, 971–978.
Roubik, D.W., Wolda, H., 2001. Do competing honey bees matter?
Dynamics and abundance of native bees before and after honey bee
invasion. Population Ecology 43, 53–62.
Schaffer, W.M., Zeh, D.W., Buchmann, S.L., Kleinhans, S., Valentine
Schaffer, M., Antrim, J., 1983. Competition for nectar between
introduced honeybees (Apis mellifera) and native North American
bees and ants. Ecology 64, 564–577.
Shaw, R.G., Mitchell-Olds, T., 1993. ANOVA for unbalanced data: an
overview. Ecology 74, 1638–1645.
Steffan-Dewenter, I., Tscharntke, T., 2000. Resource overlap and
possible competition between honey bees and wild bees in central
Europe. Oecologia 122, 288–296.
Sugden, E.A., Pyke, G.H., 1991. Effects of honey bees on colonies of
Exoneura asimillima, an Australian native bee. Australian Journal
of Ecology 16, 171–181.
Sugden, E.A., Thorp, R.W., Buchmann, S.L., 1996. Honey bee-native
bee competition: focal point for environmental change and
apicultural response in Australia. Bee World 77, 26–44.
Taylor, G., Whelan, R.J., 1988. Can honeybees pollinate Grevillea?
Australian Zoologist 24, 193–196.
Thorp, R.W., Wenner, A.M., Barthell, J.F., 2000. Pollen and nectar
resource overlap among bees on Santa Cruz Island. In: Proceedings
of the 5th Californian Islands Symposium, pp. 261–268.
Valido, A., Dupont, Y.L., Hansen, D.M., 2002. Native birds and
insects, and introduced honey bees visiting Echium wildpretii
(Boraginaceae) in the Canary Islands. Acta Oecologica 23, 413–
419.
Vaughton, G., 1992. Effectiveness of nectarivorous birds and honey-
bees as pollinators of Banksia spinulosa (Proteaceae). Australian
Journal of Ecology 17, 43–50.
Vaughton, G., 1996. Pollination disruption by European honeybees in
the Australian bird-pollinated shrub Grevillea barklyana (Protea-
ceae). Plant Systematics and Evolution 200, 89–100.
Visscher, P.K., Seeley, T.D., 1982. Foraging strategy of honeybee
colonies in a temperate deciduous forest. Ecology 63, 1790–
1801.
Wenner, A., 1993. The honey bees of Santa Cruz. Bee Culture, 272–
275.
Wenner, A.M., Thorp, R.W., 1994. Removal of feral honey bee (Apis
mellifera) colonies from Santa Cruz Island. In: Halvorson, W.L.,
Maender, G.J. (Eds.), The Fourth California Islands Symposium:
Update on the Status of Resources. Santa Barbara Museum of
Natural History, Santa Barbara, pp. 513–522.
Wenner, A.M., Thorp, R.W., Barthell, J.F., 2000. Removal of
European honey bees from the Santa Cruz Island ecosystem. In:
Proceedings of the 5th Californian Islands Symposium, pp. 256–
260.
Westerkamp, C., 1991. Honeybees are poor pollinators – why? Plant
Systematics and Evolution 177, 71–75.
Y.L. Dupont et al. / Biological Conservation 118 (2004) 301–311 311