Predicting the effects of nectar robbing on plant reproduction: Implications of pollen limitation and plant mating system

Abstract

The outcome of species interactions is often difficult to predict, depending on the organisms involved and the ecological context. Nectar robbers remove nectar from flowers, often without providing pollination service, and their effects on plant reproduction vary in strength and direction. In two case studies and a meta-analysis, we tested the importance of pollen limitation and plant mating system in predicting the impacts of nectar robbing on female plant reproduction. We predicted that nectar robbing would have the strongest effects on species requiring pollinators to set seed and pollen limited for seed production. Our predictions were partially supported. In the first study, natural nectar robbing was associated with lower seed production in Delphinium nuttallianum, a self-compatible but non-autogamously selfing, pollen-limited perennial, and experimental nectar robbing reduced seed set relative to unrobbed plants. The second study involved Linaria vulgaris, a self-incompatible perennial that is generally not pollen limited. Natural levels of nectar robbing generally had little effect on estimates of female reproduction in L. vulgaris, while experimental nectar robbing reduced seed set per fruit but not percentage of fruit set. A meta-analysis revealed that nectar robbing had strong negative effects on pollen-limited and self-incompatible plants, as predicted. Our results suggest that pollination biology and plant mating system must be considered to understand and predict the ecological outcome of both mutualistic and antagonistic plant-animal interactions.

Full text

PREDICTING THE EFFECTS OF NECTAR ROBBING ON PLANT
REPRODUCTION
: IMPLICATIONS OF POLLEN LIMITATION AND
PLANT MATING SYSTEM
1
LAURA A. BURKLE,
2,3,5
REBECCA E. IRWIN,
2,3
AND DANIEL A. NEWMAN
3,4
2
Department of Biological Sciences, Dartmouth College, Hanover, New Hampshire 03755 USA;
3
Rocky Mountain Biological
Laboratory, Crested Butte, Colorado 81224 USA; and
4
Department of Zoology, University of Toronto,
Toronto, Ontario M5S 3G5 Canada
The outcome of species interactions is often difficult to predict, depending on the organisms involved and the ecological
context. Nectar robbers remove nectar from flowers, often without providing pollination service, and their effects on plant
reproduction vary in strength and direction. In two case studies and a meta-analysis, we tested the importance of pollen limitation
and plant mating system in predicting the impacts of nectar robbing on female plant reproduction. We predicted that nectar
robbing would have the strongest effects on species requiring pollinators to set seed and pollen limited for seed production. Our
predictions were partially supported. In the first study, natural nectar robbing was associated with lower seed production in
Delphinium nuttallianum, a self-compatible but non-autogamously selfing, pollen-limited perennial, and experimental nectar
robbing reduced seed set relative to unrobbed plants. The second study involved Linaria vulgaris, a self-incompatible perennial
that is generally not pollen limited. Natural levels of nectar robbing generally had little effect on estimates of female reproduction
in L. vulgaris, while experimental nectar robbing reduced seed set per fruit but not percentage of fruit set. A meta-analysis
revealed that nectar robbing had strong negative effects on pollen-limited and self-incompatible plants, as predicted. Our results
suggest that pollination biology and plant mating system must be considered to understand and predict the ecological outcome of
both mutualistic and antagonistic plant–animal interactions.
Key words: Bombus occidentalis; Colorado; Delphinium nuttallianum; Linaria vulgaris; meta-analysis; nectar robbing;
pollen limitation; self-compatibility.
Mutualisms, interspecific interactions in which both mem-
bers undergo a net fitness benefit, are ubiquitous in natural
systems (Bronstein, 2001a) and bestow critical ecosystem
services (e.g., Kearns et al., 1998). However, exploitation of
mutualisms, including those between flowering plants and their
pollinators, is also widespread (reviewed in Bronstein, 2001b).
Exploiters gain access to rewards without providing services or
rewards in return. Cheaters of plant–pollinator mutualisms are
common among both plants and floral visitors. For example,
mate-mimicking or nectarless orchids cheat their pollinators of
nectar rewards, thus ensuring pollination service without
incurring the potential costs of producing nectar (e.g., Gigord
et al., 2002; Thakar et al., 2003). Conversely, nectar robbers
are animals that access floral nectar by biting holes in the sides
of flowers, often without providing pollination service (Inouye,
1980; Maloof and Inouye, 2000; Irwin et al., 2001). This paper
focuses on nectar robbing and its impacts, in conjunction with
pollen limitation and plant mating system, on female plant
reproductive success. Cheaters in mutualistic systems are
universal (Bronstein, 2001b), and investigating the mecha-
nisms by which the impacts of cheaters are reduced or
strengthened will provide a better understanding of their role in
community interactions.
Nectar robbers are widespread both geographically and
taxonomically (Irwin and Maloof, 2002), and robbing may
affect up to 100% of flowers in some plant populations (e.g.,
Stout et al., 2000; Irwin and Maloof, 2002; Newman and
Thomson, 2005). In one review, Maloof and Inouye (2000) list
reports of negative, neutral, and positive fitness effects of
robbing on female plant function. Alternatively, Irwin et al.
(2001) used a meta-analysis approach and found, on average, a
marginal negative effect of nectar robbing. Despite the
abundance of robbing in nature and the wealth of attention
this form of exploitation has received (e.g., Arizmendi et al.,
1996; Morris, 1996; Irwin and Brody, 1999; Maloof and
Inouye, 2000; Navarro, 2000; Lara and Ornelas, 2001; Irwin,
2003; Kjonaas and Rengifo, 2006), it remains difficult to
predict the degree to which robbers will affect plant fitness.
Irwin et al. (2001) found that both the identity of the robber and
the pollinator (i.e., insects vs. birds) influenced the effects of
robbers on plant reproduction. A number of other potential
variables, such as the diversity of other floral resources
available in the community, the effect of robbers on nectar-
standing crop, and the scarcity of nectar in the environment,
have also been suggested as factors determining the plant
fitness impacts of robbers (Maloof and Inouye, 2000). While
these variables are clearly important, some of them (in
particular, nectar availability in the environment) are difficult
to measure (Pleasants and Zimmerman, 1983) and may be too
variable across populations and across systems (Schmida and
Kadmon, 1991) to draw general conclusions. Nectar robbing
may be context dependent to some degree, but some characters
may be robust as predictor variables of the effects of robbing
on plant reproduction. Here, we propose that two alternative
1
Manuscript received 14 May 2007; revision accepted 9 October 2007.
The authors thank P. Flanagan, S. Keller, and A. Toth for help in the
field and A. Brody, S. Elliott, B. DeGasperis, D. Inouye, P. Muller, and L.
Rolfe for valuable comments on the manuscript. The staff of Rocky
Mountain Biological Laboratory (RMBL) provided access to field sites.
Field research was supported by grants from the National Science
Foundation (DEB-9806501), the RMBL (Lee. R. G. Snyder Fund), and the
Colorado Mountain Club. Laboratory work was supported by a grant from
the National Science Foundation (DEB-0089643).
5
Author for correspondence (e-mail: Laura.A.Burkle@Dartmouth.
edu)
1935
American Journal of Botany 94(12): 1935–1943. 2007.

variables may strongly predict the degree to which robbing
affects female plant reproduction: (1) the extent to which plants
are pollen limited and (2) the plant mating system. Although
the notion that pollen limitation and the plant mating system
may be used to predict the effects of robbing on female
reproductive success is not new (e.g., Arizmendi et al., 1996;
Stout et al., 2000; Arizmendi, 2001), the importance of these
two variables has not been stringently tested. We are focusing
here on cases in which robbers do not damage female
reproductive organs (e.g., stigmas, styles, ovaries) while
robbing. Moreover, we are focusing on the effects of robbing
on female plant reproduction because too few studies have
tested the effects of robbing on male plant reproduction to draw
any general conclusions (e.g., Morris, 1996; Irwin and Brody,
2000).
Pollen limitation—If robbed plants receive fewer pollinator
visits than unrobbed plants (e.g., Roubik, 1982; Irwin and
Brody, 1998) and if plant reproduction is limited by pollinator
visits and pollen receipt (e.g., Young and Young, 1992; Burd,
1994; Ashman et al., 2004), then the fitness costs associated
with robbing may be high. Thus, the degree to which plants are
pollen limited should strongly impact the effects of robbers on
plant reproduction. Conversely, the costs of robbing to female
plant fitness may be negligible if pollinator visitation is
sufficient to fertilize all available ovules. Such might be the
case if pollinators cannot identify robbed flowers (Goulson et
al., 1998), if alternative nectar sources are scarce (Irwin et al.,
2001), if legitimate visitation increases as a consequence of
reduced nectar volume in robbed flowers (Maloof and Inouye,
2000), or if robbing animals also occasionally visit legitimately
(Morris, 1996) or pollinate while robbing (e.g., Higashi et al.,
1988). Robbing may also have no effect on female plant
reproduction if seed production is limited by resources rather
than pollinators (Campbell and Halama, 1993). Additionally,
robbing may not affect female plant reproduction if pollinators
discriminate between robbed and unrobbed flowers but deposit
fewer pollen grains of higher quality on stigmas of robbed
flowers because of longer pollinator flights between consec-
utive visits (Maloof, 2001); this argument assumes that
increased pollen quality balances reduced pollen deposition
for female plant reproduction and that plant populations have
strong spatial genetic structuring. Although pollen quality may
be important to the effects of robbing on plant reproductive
success (Aizen and Harder, 2007), we focused here on a
quantitative perspective of pollen limitation in relation to nectar
robbing. Finally, positive indirect effects of nectar robbers,
such as increasing the number of flowers and plants visited by
legitimate pollinators, presumably due to changes in nectar
rewards, has also been observed (e.g., Lara and Ornelas, 2002).
Plant mating system—Life history traits of plants, such as
mating system, may also strongly affect the plant fitness
consequences associated with robbing. For example, for plants
that are predominantly selfing, changes in pollinator behavior
following robbing may have little consequence on plant
reproductive success. The same might be expected for plant
species with mixed mating systems that suffer little or no
fitness consequences as a result of selfing (Travers and Mazer,
2000). Alternatively, plants with mixed mating systems can
incur strong selfing consequences due to early-acting inbreed-
ing depression (Husband and Schemske, 1996). For self-
incompatible plants, pollinator behavior following robbing may
strongly impact plant reproductive success through, for
example, changes in outcrossing distance (Maloof, 2001) or
geitonogamy (Irwin, 2003). Self-incompatibility is typically
defined as a genetic mechanism that reduces or inhibits self-
fertilization (de Nettancourt, 2001). Here, we distinguish
between self-incompatible species that can autogamously
self-pollinate from those that require pollinators to move self-
pollen from anthers to stigmas within flowers or plants. This
distinction highlights all cases in which pollinator visitation is
important for female plant reproduction.
To understand the effects of pollen limitation on the outcome
of plant–nectar robber interactions, we compared the effects of
nectar robbing on the female reproductive success of two
subalpine plants, Delphinium nuttallianum (Ranunculaceae)
and Linaria vulgaris (Scrophulariaceae). The two species share
a common nectar-robbing bumble bee, Bombus occidentalis
(Apidae), but differ in the degree of pollen limitation. Seed set
of D. nuttallianum is generally limited by pollinator visitation
(Waser and Price, 1990), whereas that of L. vulgaris is
generally not limited by pollinator visits (R. E. Irwin,
unpublished data). Using a combination of observational and
experimental studies, we measured the effects of robbing on
female plant reproductive success in both species. We
predicted that robbing would have a stronger negative effect
on D. nuttallianum than on L. vulgaris because female
reproduction in D. nuttallianum is more strongly pollen
limited. We then combined existing published studies with
the results presented in this paper and conducted a meta-
analysis to more broadly understand the degree to which pollen
limitation and plant mating system (self-compatibility) affect
the outcome of robbing on female plant reproduction. A meta-
analytical approach is advantageous in that it allows us to
generalize across a wide variety of systems involving nectar
robbers. Although this work focused on nectar robbers and
their impacts on female plant reproduction, the concepts
presented are applicable to other species interactions that
involve exploitation and the traits that mitigate or compound
the costs of exploiters (for a review, see Bronstein, 2001b).
Investigating the fitness consequences of cheating is a first step
toward understanding the ecological and evolutionary impli-
cations of cheating on populations and communities.
MATERIALS AND METHODS
Study system—Fieldwork was conducted from June through September
from 1997 to 1999 in and around the Rocky Mountain Biological Laboratory,
Colorado, USA (RMBL; 38857
0
29
00
N, 106859
0
06
00
W; altitude, 2900 m a.s.l.).
Delphinium nuttallianum Pritzel (¼ D. nelsonii Greene) blooms in the early
summer around the RMBL (late May to early July; Saavedra et al., 2003).
Delphinium nuttallianum is a perennial (mean life span until first flower ¼3–7
years; Waser and Price, 1991), producing 1–10 purple, zygomorphic flowers on
1–2 racemose inflorescences (Waser, 1978). The flowers of D. nuttallianum
have a nectar spur in which nectar collects (nectar production rate, 0.02 lL/h;
percentage of sugar concentration, 51.2 6 0.7% [mean 6 SE]; Waser, 1978).
The major pollinators of D. nuttallianum around the RMBL are broad-tailed
hummingbirds (Selasphorus platycercus Swainson, Trochilidae), queen bumble
bees (Bombus appositus Cresson, B. flavifrons Cresson, B. californicus F.
Smith, and B. nevadensis Cresson; Apidae), and to a lesser extent, small
solitary bees and other insect visitors (Waser, 1978, 1982; Waser and Price,
1990). Although the flowers of D. nuttallianum are self-compatible to some
degree (Price and Waser, 1979), protandry prevents autogamous self-
pollination, and plants require pollinators to transfer even self-pollen within
flowers and plants (Waser and Price, 1990); D. nuttallianum sets very few
seeds in the absence of pollinator visitation (Waser, 1978). Only 7% of seeds
1936 AMERICAN JOURNAL OF BOTANY [Vol. 94

on average are produced from selfing, with some variation among populations
(Williams et al., 2001). Seed set is often limited by pollinator visitation (Waser
and Price, 1990), with greater pollinator visitation increasing stigma pollen
loads and seed set as a saturating function (Bosch and Waser, 1999).
Linaria vulgaris Mill. blooms in the late summer around the RMBL (late
July to early October). Native to Europe, L. vulgaris was introduced into areas
around the RMBL in the late 1800s. Linaria vulgaris is a perennial (mean root
lifespan, 3.8 yr; Robocker, 1974), with a single stalk producing 26.8 6 2.6
yellow, zygomorphic flowers (R. E. Irwin, unpublished data). Nectar
production rate of L. vulgaris flowers averages 0.09 lL/h, with a sugar
concentration of 37% in eastern North America (Arnold, 1982). Around the
RMBL, the sugar concentration of L. vulgaris can reach as high as 61.0 6
1.6% (R. E. Irwin, unpublished data). Linaria vulgaris reproduces extensively
by clonal growth (Saner et al., 1995) as well as through seeds. An individual
stalk can produce .5000 seeds (Zilke and Coupland, 1954), and a genet can
produce up to 30 000 seeds (Saner et al., 1995). At the RMBL, the self-
incompatible flowers of L. vulgaris are pollinated primarily by bumble bees (B.
appositus, B. californicus, B. fervidus, B. flavifrons, B. frigidus) and to a lesser
extent, small solitary bees and rufous hummingbirds (S. rufus). Pollen
limitation of female reproductive success varies among regions in which L.
vulgaris occurs. In eastern North America, pollinator visitation likely limits
seed production (Arnold, 1982); however, stalks are generally not pollen
limited for seed production around the RMBL (R. E. Irwin, unpublished
manuscript). For example, in 2002, 35 stalks each across two sites were
provided with supplemental pollen or left open to natural levels of pollination.
There was no significant effect of hand pollination on seed production per stalk
relative to stalks without hand pollination (F
1,67
¼1.14, P¼0.29; Appendix S1,
see Supplemental Data with online version of this article). We would have
needed at least 238 stalks per treatment to find a significant effect of hand
pollination at a ¼ 0.05 (R. E. Irwin, unpublished data).
The nectar spurs of both D. nuttallianum and L. vulgaris are robbed by the
bumble bee Bombus occidentalis Greene, Apidae. The bee uses its toothed
mandibles to chew a hole through the sides of the nectar spurs; it then inserts its
proboscis into the hole and removes nectar, bypassing the floral opening used
by legitimate visitors. Data on natural nectar-robbing rates on D. nuttallianum
and L. vulgaris have been published previously (Irwin and Maloof, 2002); here
we expand our understanding of plant–nectar robber interactions in these plant
species with the inclusion of new data to assess the relationship between natural
levels of nectar robbing and plant reproduction and with the inclusion of
experimental data from manipulating nectar robbing and measuring plant
response. Robbing rates on D. nuttallianum range from 0 to 100% of available
flowers with peak robbing of 56.5 6 8.5% (mean 6 1 SE) of available flowers
robbed (Irwin and Maloof, 2002). The effects of robbing on D. nuttallianum
reproductive success are unknown. Robbing rates on L. vulgaris range from 0
to 100% of available flowers robbed with peak robbing of 79.3 6 15.4% (mean
6 1 SE) of available flowers robbed (Irwin and Maloof, 2002). For L. vulgaris
growing in southern England, high levels of robbing (96% of open flowers
robbed) by B. lapidarius, B. lucorum, and B. terrestris had no effect on female
reproductive success (Stout et al., 2000); however, the effects of nectar robbing
by B. occidentalis on L. vulgaris in western North America are unknown.
Field methods
—The effects of nectar robbing on Delphinium nuttallia-
num—We measured the effects of natural and experimental nectar robbing on
female reproductive success of D. nuttallianum in 1999.
Observational study—To understand the relationship between natural levels
of nectar robbing and female plant function, we randomly chose 50 budding D.
nuttallianum at two sites (30 plants at site 1, 38857
0
41.27
00
N, 106859
0
05.05
00
W; 20 plants at site 2, 38857
0
39.60
00
N, 106859
0
01.43
00
W) in early June 1999.
We measured robbing levels by counting the number of open flowers and the
number of nectar robber holes in each flower once per week throughout the
flowering season (Irwin and Maloof, 2002).
When plants senesced, we collected and counted all of the fruits and seeds
produced by each plant. In addition, we counted the number of aborted ovules
in each fruit under a dissecting microscope. For each plant, we calculated
percentage of seed set (number of seeds produced divided by the total number
of ovules per plant) and total seed set as measures of female plant reproduction.
Percentage of seed set and total seed set were the most biologically relevant
measures of female reproductive success for D. nuttallianum (as opposed to
percentage of fruit set) because an individual plant does not produce many
flowers in a given year, and the vast majority of flowers set fruit (almost 90% of
flowers set fruit in this study). Moreover, percentage of seed set takes into
account the finite number of ovules and underlying resource status of the plant,
and total seeds is indicative of whole-plant reproductive success.
To assess the relationship between female plant reproduction and natural
levels of nectar robbing, we performed regressions of percentage of seed set per
fruit and total seeds per plant (log-transformed) on mean nectar robbing per
plant (proportion of robbed flowers per plant, arcsine-square root transformed).
Percentage of seed set and total seed production per plant were not significantly
correlated (r ¼ 0.27, N ¼ 28, P ¼0.17) and thus were analyzed separately. To
avoid the confounding influence of a correlation between nectar robbing and
female plant reproduction among sites, rather than among individuals within
sites (the relevant hypothesis), we tested for homogeneity of the covariance
structure between nectar robbing and mean percentage of seed set and total seed
set across sites (PROC DISCRIM, option POOL ¼TEST, SAS version 8; SAS
Institute, 2001). We combined the data across sites because we found no
evidence to reject the null hypothesis of homogeneity of the covariance
structure for percentage of seed set or total seed set (in both cases, v
2
, 5.55, p
¼ 0.14). Nectar robbing may affect not only plant success but also per flower
success, especially if pollinators discriminate on a per flower, rather than on a
per plant, basis or if unrobbed flowers within plants can compensate for those
that are robbed (Irwin and Brody, 1999). Thus, we also examined percentage of
seed set per fruit and number of seeds per fruit of robbed and unrobbed flowers
using a MANOVA. By using a multivariate approach here and for analysis of
experimental data, we could test the effects of nectar robbing on several
interrelated measures of female plant reproduction (Scheiner, 1993; Rencher,
1995). We removed 12 plants from all statistical analyses because they were
consumed by unknown herbivores before we could collect the fruits.
Experimental manipulation—From the observational study, we cannot tease
apart the correlative effects of robbing on plant reproduction from the causal
effects; thus, we also measured the effects of experimental nectar robbing on
female plant reproduction in D. nuttallianum. In 1999, we randomly chose 60
budding D. nuttallianum at a third site (38857
0
36.57
00
N, 106858
0
58.78
00
W)
with naturally low robbing rates. We randomly assigned 30 plants each to an
artificially superimposed ‘‘unrobbed’’ treatment (0% of available flowers
robbed) or ‘‘complete nectar-robbing’’ treatment (100% of available flowers
robbed). These levels of robbing are common in the field for D. nuttallianum
(Irwin and Maloof, 2002). Because natural nectar robbing was uncommon at
this site, we did not need to collar flowers to deter robbers. Nectar robbing by
B. occidentalis was recorded and made up less than 11% of the available
flowers open on control plants. To our knowledge, D. nuttallianum flowers do
not respond positively to repeated nectar removal. Originally, we had an equal
number of plants per robbing treatment (30 plants per treatment); however, six
plants in the unrobbed treatment (final N ¼ 24 plants) and three plants in the
complete robbing treatment (final N ¼ 27 plants) were consumed by unknown
herbivores and thus were excluded from the study.
To rob flowers artificially, we made a small hole in the side of the nectar spur
with dissecting scissors and removed all available nectar through the hole with a
10-lL microcapillary tube. Robbing treatments were performed every other day
throughout the blooming season of each plant. If flowers had been robbed
previously, the flowers were robbed again through the existing holes. This
method of artificial robbing has been used successfully to mimic the effects of
robbing by B. occidentalis on a co-occurring plant species, Ipomopsis aggregata
(Polemoniaceae) (Irwin and Brody, 1998). To ensure that our experimental
robbing treatment mimicked natural robbing in terms of plant response by D.
nuttallianum, we used an ANOVA to compare mean percentage of seed set from
plants with naturally vs. artificially robbed flowers. We found no difference in
percentage of seed set between natural and artificial robbing (F
1,44
¼0.22, P ¼
0.64), suggesting that our nectar-robbing method mimics natural robbing by B.
occidentalis. Plants in the unrobbed treatment were left unmanipulated, but the
flowers on these plants were physically handled each day to control for the
effects of flower handling associated with the robbing treatments.
Once plants ceased blooming, we collected all of the fruits and measured
percentage of seed set per fruit and total seeds for each plant, as described
earlier. To understand the effects of experimental nectar robbing on female
plant reproduction, we used a MANOVA with robbing treatment as a fixed
effect and percentage of seed set and total seeds (square-root transformed) as
response variables. A significant MANOVA was followed by univariate
ANOVAs for each response variable.
The effects of nectar robbing on Linaria vulgaris
—We also measured the
effects of natural and experimental nectar robbing on female reproductive
success in L. vulgaris. The methods for L. vulgaris differed slightly from those
December 2007] BURKLE ET AL.—EFFECTS OF NECTAR ROBBING ON PLANT REPRODUCTION 1937

of D. nuttallianum because of differences in life history characteristics and
robbing levels between the two species.
Observational study—We quantified the relationship between natural levels
of nectar robbing and female plant reproduction from 1997 to 1999 in one large
site in the RMBL town site (38857
0
32.39
00
N, 106859
0
24.36
00
W). We measured
natural robbing by B. occidentalis on focal stalks once per week throughout the
flowering season, as described for D. nuttallianum. We measured robbing on 24
stalks in 1997, 35 stalks in 1998, and 50 stalks in 1999 (Irwin and Maloof, 2002).
We realize that the best unit of replication would have been to measure the effects
of nectar robbing on a whole-plant basis. However, L. vulgaris is clonal, and it is
difficult to determine which stalks (ramets) are connected without digging up an
entire population. Although working with ramets has disadvantages (i.e., we are
not measuring whole-plant reproduction), ramet dynamics have been successfully
compared to genet dynamics in other systems (e.g., Caswell, 1985; Eriksson and
Jerling, 1990; Silvertown et al., 1993). In addition, there is competition for
resources within L. vulgaris genets, and undamaged ramets may not direct
resources toward damaged ramets (Hellstrom et al., 2006). If, however, resource
translocation occurs from damaged to undamaged ramets, our results could be
biased toward finding effects of nectar robbing.
Once plants ceased blooming, we collected all expanded fruit capsules and
recorded aborted fruits. Because an individual L. vulgaris stalk can produce
.5000 seeds (Zilke and Coupland, 1954), counting the number of seeds
produced per stalk was not practical. Instead, for each stalk we measured the
mass of 10 seeds on a microbalance (UMX2 Mettler Toledo Ultramicro
Balance, Columbus, Ohio, USA) to the nearest 0.0001 g. Then we measured
the mass of all seeds the stalk produced. We estimated the number of seeds
produced per stalk by calculating seed number as: number of seeds ¼ (mass of
total seeds 3 10 seeds)/mass of 10 seeds. Seed masses were only calculated for
black seeds (likely viable seeds; Saner et al., 1995). White seeds are inviable
(L. A. Burkle, personal observation) and were not included in the total seed
number. For each stalk, we calculated three measures of female plant
reproduction: (1) percentage of fruit set (number of expanded fruits divided
by the total number of flowers produced; arcsine square-root transformed), (2)
number of seeds per fruit (number of seeds divided by the number of successful
fruits), and (3) seed production (total seeds produced per stalk; log
transformed). We used this suite of variables to estimate female plant
reproduction to identify the mechanisms by which robbers might affect plant
reproduction. For example, if robbers affect percentage of fruit set and total
seeds but not seeds per fruit, this would suggest that robbers affect total seeds
by changing the probability that a plant will produce a successful fruit.
To assess the relationship between female plant reproduction and natural
levels of nectar robbing, we performed multivariate regressions of percentage
of fruit set, number of seeds per fruit, and seed production on mean nectar
robbing per stalk in each year. Separate analyses were performed in each year
of the study because (1) robbing levels, percentage of fruit set, number of seeds
per fruit, and seed production varied significantly among years (MANOVA:
F
6,184
¼ 52.6, P ¼ 0.0001; Table 1) and (2) the relationship between robbing
level and female reproduction was confounded with year (PROC DISCRIM,
option POOL ¼TEST, SAS version 8; in all cases v
2
. 24.03, p , 0.001). We
also compared percentage of fruit set and number of seeds per fruit between
robbed and unrobbed flowers within stalks in each year using Hotelling’s T
2
;
we only included stalks that had robbed and unrobbed flowers in this analysis.
Experimental manipulation—We experimentally manipulated nectar rob-
bing and measured subsequent effects on female plant reproduction in 1997.
We haphazardly chose 21 stalks in the one large population of L. vulgaris in
which we recorded natural robbing. We randomly assigned 10 stalks to an
experimental ‘‘unrobbed’’ treatment (0% of available flowers robbed) and 11
stalks to an experimental ‘‘complete nectar-robbing’’ treatment (100% of
available flowers robbed) for which flowers on each stalk were either
artificially robbed as described for D. nuttallianum or naturally robbed by B.
occidentalis. We removed all of the available nectar per flower in this robbing
treatment because the bumble bees typically remove all available nectar when
robbing flowers of L. vulgaris. These levels of robbing are common in the field
for L. vulgaris by B. occidentalis (Irwin and Maloof, 2002).
Artificially and naturally robbed flowers were marked with small dots of
different colors of indelible ink on their calyces. For the low robbing treatment,
nectar spurs were either ‘‘collared’’ to deter robbers, or flowers were left
unmanipulated. The collars were made of small pieces of translucent drinking
straws; one end was folded over and stapled shut with a small stapler. The other
end was placed over the spur, and the collar was tied to the pedicel and stalk of
the plant using thread. The unmanipulated flowers were visited each day to
determine whether they had been naturally robbed. Once the flowers fell off, all
flowers that were not robbed by B. occidentalis were marked with small dots of
ink on their calyces. Collared flowers were marked with a different color of ink.
To ensure that our artificial robbing treatments and collaring treatments
mimicked naturally robbed and unrobbed flowers, we compared percentage of
fruit set (arcsine-square root transformed) and number of seeds per fruit for (1)
artificially vs. naturally robbed flowers and (2) collared vs. unrobbed flowers
using paired t tests within stalks. We found no difference between artificial and
natural nectar robbing for percentage of fruit set (t
10
¼ 0.43, P ¼ 0.68) or for
number of seeds per fruit (t
10
¼ 1.31, P ¼ 0.22). Moreover, we found no
difference between collared and naturally unrobbed flowers for percentage of
fruit set (t
8
¼0.20, P ¼0.85) or number of seeds per fruit (t
8
¼1.32, P ¼0.22).
These results suggest that our artificial treatments mimicked natural conditions
in terms of female plant response.
To determine the effects of the experimental robbing treatments on female
reproductive success, we used a MANOVA with robbing treatment (artificially
robbed vs. collared) as a fixed effect and percentage of fruit set (arcsine-square
root transformed) and number of seeds per fruit as response variables. We
calculated percentage of fruit set and number of seeds per fruit as mean values
on treated flowers on a per stalk basis.
Meta-analysis: the influence of pollen limitation and plant mating system
on the effects of nectar robbing on female plant reproduction
—We
performed a meta-analysis (Gurevitch and Hedges, 2001) to examine
quantitatively the degree to which pollen limitation and plant mating system
influence the effects of nectar robbing on female plant reproduction. Using the
keyword nectar rob* (with an asterisk as a wild card), we searched the
published literature on Web of Science from 1945 to 2007 as well as literature-
cited sections from papers on nectar robbing. We included all studies that we
found (27 total) that reported a measure of female reproductive success of
robbed and unrobbed flowers or plants for which we could also locate
information on self-compatibility and pollen limitation (see Table 2 for the list
of studies included in the meta-analysis). Studies were excluded if robbers
directly damaged female reproductive organs while robbing (i.e., Galen, 1983)
in order to focus the meta-analysis on the indirect effects of robbing on female
reproduction via changes in pollination. An effect size (d) was calculated for
each study, expressing the difference in mean female reproduction in plants or
flowers that were robbed or unrobbed, divided by their pooled SD and
corrected for bias due to small sample size. Negative effects of robbing are
specified by a negative effect of size and vice versa. An effect size of 0.2
(absolute value) is considered small, 0.5 medium, 0.8 large, and 1 very large
(Cohen, 1969). The overall impact of nectar robbing on female plant
reproduction was weak but negative (d ¼0.25), with effect sizes ranging
from 3.02 to þ2.74. Our results are comparable to those found in a similar
meta-analysis by Irwin et al. (2001; d ¼0.27), with the current analysis
containing 10 additional studies. We then calculated effect sizes for groups of
studies that had plants with evidence of the presence vs. absence of pollen
limitation of female reproductive success in at least one year of study
(recognizing that pollen limitation can vary annually; Ashman et al., 2004) and
plants with different mating systems (self-compatible vs. self-incompatible).
The between-class homogeneity statistic (Q
b
) was used to statistically analyze
these comparisons. Delphinium nuttallianum was the only species that we
could find in which the literature distinguished between the genetic mechanism
of self-compatibility vs. the requirement that pollinators move both selfed and
outcrossed pollen in a genetically self-compatible species. Thus, we performed
the meta-analysis twice, with D. nuttallianum classified as self-compatible or
functionally self-incompatible (requiring pollinators to move self and outcross
pollen). Because the outcome of the statistical comparison was the same
whether D. nuttallianum was classified as self-compatible or self-incompatible,
we report the results for only the latter test.
TABLE 1. Yearly differences in percentage of robbing, percentage of fruit
set, numbers of seeds per fruit, and number of seeds per stalk in
natural populations of Linaria vulgaris. Values are mean 6 1 SE.
Year Percentage of robbing Percentage of fruit set No. seeds per fruit No. seeds per stalk
1997 39.5 6 5.8 75.1 6 3.8 90.7 6 7.1 2306.0 6 291.1
1998 23.1 6 3.0 65.7 6 2.9 88.4 6 4.6 2991.7 6 285.3
1999 95.4 6 0.6 25.2 6 2.4 100.9 6 8.1 770.8 6 94.1
1938 AMERICAN JOURNAL OF BOTANY [Vol. 94

We also performed these same meta-analyses on a subset of the studies,
including only one entry per plant genus and only one entry per study, to
minimize the possibility of biases due to author or species (Appendix S2, see
Supplemental Data with online version of this article). In addition, we explored
other traits as predictors of the ecological outcomes of nectar robbing for the
meta-analysis, including flower color, flower shape, and plant species
distribution (temperate vs. tropical/subtropical; Appendix S2).
RESULTS
The effects of nectar robbing on Delphinium nuttallia-
num—Observational study—At the plant level, there was no
relationship between percentage of seed set and natural nectar
robbing (r
2
¼ 0.02, N ¼ 28, P ¼ 0.46), but we found a
significant negative relationship between total number of seeds
per plant and robbing (r
2
¼0.15, N ¼28, P ¼0.04, Fig. 1). At
the flower level, there was no statistically significant effect of
robbing on female plant reproduction (MANOVA, Wilks’ k ¼
0.17, F
2,29
¼2.52, P ¼0.10; percentage of seed set, unrobbed,
67.8 6 4.5%; robbed, 52.7 6 4.6%; number of seeds per fruit,
unrobbed, 25.8 6 2.8 seeds per fruit; robbed, 21.0 6 2.4 seeds
per fruit [mean 6 SE]). However, we had low power to detect
a significant difference (power ¼ 0.25 at a ¼ 0.05, using
MACRO MPOWER in SAS version 8; SAS, 2001).
Experimental manipulation—In the experimental population
of D. nuttallianum, nectar robbing had a significant effect on
female plant reproduction (MANOVA, Wilks’ k ¼ 0.25, F
2,48
¼ 6.08, P ¼ 0.004). Complete nectar robbing reduced the
percentage of seed set by 22% and total seed set by 49%
compared to unrobbed plants (percentage of seed set, F
1,49
¼
5.42, P ¼0.02, Fig. 2A; total number of seeds per plant, F
1,49
¼
12.29, P ¼ 0.001, Fig. 2B).
The effects of nectar robbing on Linaria vulgaris—
Observational study—At the plant level in all three years of
the study, we found no significant relationship between
measures of L. vulgaris female plant reproduction per stalk
and natural levels of nectar robbing (Table 3).
At the flower level, however, when we compared robbed vs.
unrobbed flowers, the results were variable among years. We
found no difference in measures of female reproduction
(percentage of fruit set and number of seeds per fruit) between
robbed and unrobbed flowers in 1997 (Hotelling’s T
2
2,32
¼
3.51, P . 0.05) and 1998 (Hotelling’s T
2
2,64
¼1.47, P . 0.05).
However, in 1999, there was a significant difference between
robbed and unrobbed flowers in measures of female reproduc-
tion (Hotelling’s T
2
2,36
¼ 19.86, P , 0.01). This result was
driven by a difference in number of seeds per fruit between
robbed and unrobbed flowers (t
19
¼ 3.77, P ¼ 0.0013).
Surprisingly, robbed flowers produced almost three times more
seeds per fruit than unrobbed flowers (robbed flowers, 111.44
6 13.47 seeds per fruit; unrobbed flowers, 41.79 6 13.96
seeds per fruit). There was no difference in percentage of fruit
set between robbed and unrobbed flowers in 1999 (t
19
¼ 0.11,
P ¼ 0.91).
Experimental manipulation—We found a significant differ-
ence in female plant reproduction between artificially robbed
and unrobbed (collared) flowers (MANOVA, F
2,17
¼3.98, P ¼
0.038), with unrobbed flowers having, on average, more than
twice as many seeds per fruit as artificially robbed flowers
(ANOVA, F
1,18
¼ 8.3, P ¼ 0.01; unrobbed, 150.25 6 22.31
seeds; robbed, 63.62 6 20.18 seeds). There was no significant
difference in percentage of fruit set between artificially robbed
and unrobbed flowers (ANOVA, F
1,18
¼ 0.11, P ¼ 0.74;
artificially robbed, 68.2 6 10.2%; unrobbed, 74.1 6 11.8%).
Meta-analysis: the influence of pollen limitation and plant
mating system on the effects of nectar robbing on female
plant reproduction—Both the presence vs. absence of pollen
limitation (Q
b
¼ 51.23, df ¼1, P , 0.0001, N ¼19) and self-
compatibility (Q
b
¼ 21.5, df ¼ 1, P , 0.0001, N ¼ 25)
significantly affected the outcome of nectar robbing on plant
reproduction. A very large negative effect of robbing was
observed for pollen-limited plants (d ¼1.13, N ¼ 10 studies)
compared to plants with no evidence of pollen limitation (d ¼
0.22, N ¼ 9 studies). The effect of robbing on female
reproduction was medium and negative in self-incompatible
species (d ¼0.53, N ¼ 17 studies) and weak and negative in
self-compatible species (d ¼0.25, N ¼ 13 studies).
DISCUSSION
The effects of nectar robbing on plant reproductive success
have received extensive attention (reviewed in Maloof and
Inouye, 2000; Irwin et al., 2001). Yet the range of effects has
TABLE 2. Studies included in the meta-analysis and plant characters
(pollen limitation and self-compatibility). Dashed lines indicate species
whose pollen limitation is unknown or whose pollen limitation for
seed set varied to such a degree that the plant could not be classified.
We included studies that examined the effects of nectar robbers
(organisms that steal nectar by biting holes through flowers) and nectar
thieves (organisms that steal nectar without pollinating because of a
mismatch between floral and visitor morphology; Inouye, 1980).
Plant species Reference
Pollen
limited
Self-
compatible
Aconitum lycoctonum Utelli and Roy (2001) — yes
Anthyllis vulneraria Navarro (2000) no yes
Aphelandra sinclairiana McDade and Weeks (2004) no no
Asclepias curassavica Wyatt (1980) — no
Asclepias syriaca Fritz and Morse (1981) yes no
Centaurea solstitialis Agrawal et al. (2000) yes yes
Chilopsis linearis Richardson (2004) yes no
Corydalis ambigua Higashi et al. (1988) — no
Corydalis caseana Maloof (2001) no yes
Delphinium nuttallianum This study yes no
Fouquiera splendens Waser (1979) yes yes
Frasera speciosa Norment (1988) — no
Fuchsia magellanica Traveset et al. (1998) — yes
Fuchsia microphylla Arizmendi et al. (1996) no yes
Hamelia patens Lasso and Naranjo (2003) yes no
Impatiens capensis Rust (1979) yes yes
Impatiens capensis Temeles and Pan (2002) yes yes
Ipomoea heredifolia Lara and Ornelas (2001) — no
Ipomopsis aggregata Irwin and Brody (1999) yes no
Linaria vulgaris This study no no
Linaria vulgaris Stout et al. (2000) — no
Mertensia paniculata Morris (1996) — no
Pavonia dasypetala Roubik (1982) — yes
Petrocoptis grandiflora Guitian et al. (1994) no no
Quassia amara Roubik et al. (1985) — no
Salvia iodantha Arizmendi (2001) — yes
Salvia mexicana Arizmendi (2001) no yes
Salvia mexicana Arizmendi et al. (1996) no yes
Symphytum officinale Goulson et al. (1998) no no
Vaccinium ashei Dedej and Delaplane (2004) yes no
December 2007] BURKLE ET AL.—EFFECTS OF NECTAR ROBBING ON PLANT REPRODUCTION 1939

precluded broad generalizations about the impacts of robbing
on female plant reproduction. Here we hypothesized that the
degree of pollen limitation and plant mating system (self-
compatibility) would predict the effects of nectar robbing on
female plant reproduction. Our predictions were partially
upheld in the case studies involving montane plants (D.
nuttallianum and L. vulgaris) and fully upheld in the meta-
analysis. Natural nectar robbing of D. nuttallianum, a pollen-
limited perennial, reduced total seed set at the plant level, and
experimental nectar robbing reduced percentage of seed set and
total seed set at the plant level. Conversely, in L. vulgaris,a
species that is generally not pollen limited at our study site,
natural nectar robbing did not affect most estimates of ramet
(1997–1999) or per-flower (1997–1998) reproduction, but
experimental nectar robbing reduced seed set per fruit and not
percentage of fruit set. When we broadened our results to
include other studies in a meta-analysis, we found a stronger
negative effect of robbing on plants that were pollen limited for
seed set and on plants that were self-incompatible.
Because D. nuttallianum is pollen limited, we predicted that
high levels of nectar robbing would reduce estimates of female
reproductive success. This prediction assumes that pollinators
of D. nuttallianum detect and avoid nectar-robbed plants and
flowers. The behavior of D. nuttallianum pollinators in
response to nectar robbing in other plant systems partially
supports this assumption. Delphinium nuttallianum is pollinat-
ed both by broad-tailed hummingbirds and by queen bumble
bees (Waser and Price, 1990). Previous studies with broad-
tailed hummingbirds have shown that these birds can avoid
nectar-robbed plants and flowers in the sympatric Ipomopsis
aggregata (Irwin and Brody, 1998), likely by detecting and
avoiding flowers with reduced nectar volume (Irwin, 2000).
Conversely, bumble bees around the RMBL do not avoid
nectar-robbed plants and flowers of Corydalis caseana
(Maloof, 2000), and bumble bees do not discriminate between
robbed and unrobbed flowers in other plant species (Rust,
1979; Richardson, 2004). If hummingbirds can discriminate
between robbed and unrobbed flowers and bumble bees
cannot, then the negative effects of robbing observed here
may be driven primarily by hummingbird pollinators. Thus, the
effects of robbing on plant reproduction may be strongly
affected by the relative abundance of hummingbird vs. bumble
bee pollinators, which can vary both spatially and temporally
(e.g., Pyke, 1982; Inouye et al., 1991). Given the strong effects
of nectar robbing on D. nuttallianum reproduction observed
here, further studies quantifying hummingbird and bumble bee
pollinator foraging behavior in response to robbing are
warranted.
One caveat about the effects of robbing on D. nuttallianum is
that we found negative effects of natural nectar robbing on total
seed set at the plant level, but we found no relationship
between natural robbing and percentage of seed set at the plant
level or on measures of female plant reproduction at the flower
level. We know that this result is not due to increased rates of
fruit abortion because D. nuttallianum rarely aborted fruits in
our study. Instead, this result may be driven in part by the
behavior of robbers. If percentage of seed set remains
consistent across robbing levels while total seed set per plant
decreases with robbing, then nectar robbers may be selecting
plants with fewer total ovaries. Although we did not measure
floral traits of D. nuttallianum, there may be trade-offs in
resource allocation between the production of ovaries vs. floral
rewards. For example, plants with low ovary production may
have relatively more resources to allocate to characters
associated with pollinator attraction, such as higher nectar
production, which could attract more nectar robbing. In
contrast, we assigned robbing treatments randomly to plants
when we experimentally manipulated robbing to D. nuttallia-
num. Thus, robbing was superimposed on random levels of
ovule production and floral attractive characters. Studies
assessing the manner in which nectar-robbing bees select D.
Fig. 1. Relationship between total seeds per plant and natural nectar
robbing in Delphinium nuttallianum. The relationship [Log(total seeds) ¼
4.2  0.74 3 robbing; arcsine square-root transformed] was negative and
significant.
Fig. 2. Effects of experimental nectar robbing on Delphinium
nuttallianum. (A) Percentage of seed set per plant and (B) total seeds
per plant. Both variables were significantly higher in unrobbed plants (0%
of available flowers robbed; N ¼24 plants) than in experimentally robbed
plants (100% of available flowers robbed; N ¼ 27 plants). Bars indicate
mean 6 1 SE.
1940 AMERICAN JOURNAL OF BOTANY [Vol. 94

nuttallianum to forage on would provide additional insight into
the mechanisms driving the patterns we observed.
In the primarily bumble bee-pollinated Linaria vulgaris,we
found that natural nectar robbing had no effect on ramet
measures of reproduction in any year of study and had no effect
on per-flower reproductive success in 1997 and 1998. These
results are consistent with those reported for the effects of
nectar robbing on L. vulgaris reproduction in southern England
(Stout et al., 2000). Stout et al. (2000) reported that bumble bee
pollinators of L. vulgaris continued to visit plants even with
96% of available flowers robbed. These results also match
studies described previously in this paper suggesting that
bumble bee pollinators do not discriminate between robbed and
unrobbed plants and flowers (also reviewed in Maloof and
Inouye, 2000). Pollinators may not discriminate between
robbed and unrobbed flowers if robbing has little effect on
nectar standing crop or if pollinators cannot identify potentially
unrewarding robbed plants and flowers. Distinguishing be-
tween these two hypotheses would require manipulating robber
holes and nectar availability and measuring the foraging
behavior of bumble bee pollinators (as in Irwin, 2000 with
hummingbirds). We did, however, find an effect of experi-
mental robbing on L. vulgaris reproduction. Experimental
nectar robbing reduced the number of seeds produced in each
successful fruit but not the probability that a flower would
produce a successful fruit. Mechanistically, this result suggests
that robbed and unrobbed flowers both received enough pollen
to produce similar numbers of seeds; however, the mechanism
driving reduced seeds per fruit is not clear, especially because
L. vulgaris generally are not pollen limited for seed production
around the RMBL (R. E. Irwin, unpublished data). Moreover,
that we found no consistent effect of natural nectar robbing on
plant performance but a reduction in seeds per fruit when we
manipulated nectar robbing suggests again that robbers may
not be choosing plants at random (see earlier in Discussion for
D. nuttallianum and Irwin, 2006). In this case, one relevant
hypothesis is that nectar robber selection of plants may
positively covary with plant resource status. Mechanistic work
to assess how robbing affects pollinator visitation, the quantity
and quality of pollen received, and seeds produced under
different resource environments as well as the factors that
impact robber selection of plants may provide some insight.
We found one anomalous result with respect to nectar
robbing in L. vulgaris. In 1999, naturally robbed flowers
produced significantly more seeds per fruit than unrobbed
flowers on the same stalk. The mechanisms explaining this
result are not completely clear. At the very least, we can rule
out one factor. Anatomically, nectar robbers cannot pollinate
while they rob flowers of L. vulgaris; thus, pollinating while
robbing does not explain this result. One factor that may be
involved in this unexpected result is that the percentage of
flowers robbed per stalk in 1999 was significantly higher than
in 1997 and 1998 (Table 3), suggesting that primary (and likely
secondary) nectar robbers were dominant floral visitors in
1999. Bumble bees require both nectar and pollen to provision
their larvae. If B. occidentalis were robbing flowers for nectar
and then visiting those same flowers legitimately to collect
pollen, then robbed flowers might have received higher
quantity and/or quality of pollen loads, although we did not
estimate pollen limitation of seed production for L. vulgaris in
1999. The same logic may also apply to bumble bees that act as
secondary nectar robbers and pollinators in different floral
visits (i.e., B. flavifrons is a dominant pollinator and secondary
robber of L. vulgaris). Because we did not observe rates of
robber or pollinator visits, we do not know how often B.
occidentalis and secondary nectar-robbing bees switch between
robbing and pollinating behaviors nor the relative abundances
of pollinators vs. robbers in the year of study. Nectar robbers,
however, are known to switch between secondary robbing and
legitimate pollination in this system (Newman and Thomson,
2005), and behavioral switches are common for other bee
species (e.g., Rust, 1979; Morris, 1996).
When we incorporated our results from D. nuttallianum and
L. vulgaris into a meta-analysis with other published studies,
we found that nectar robbing had the strongest negative effects
on species that were pollen limited for female reproduction and
that were self-incompatible. The degree of pollen limitation of
a plant is driven, at least in part, by the rate of pollinator
visitation, as well as pollen availability within the community
(Ashman et al., 2004). Moreover, self-incompatible plants
require pollinators for successful reproduction. These results
suggest that understanding the effects of nectar robbing on
female plant reproduction requires not only an understanding
of plant mating system but also a multispecies perspective,
including pollinators as well as nectar robbers and plants. There
is growing appreciation that the ecological and evolutionary
outcomes of multispecies interactions are difficult to predict
from the outcome of pairwise relationships (e.g., Strauss and
Irwin, 2004; Morris et al., 2007). Future studies that
incorporate the activity of pollinators and plant life history
traits into plant–robber interactions will likely provide strong
ecological and evolutionary insights (see Results and online
Appendix S2). Finally, pollen limitation and plant mating
system are certainly not the only traits that may be strong
predictors of the effects of nectar robbing on female plant
reproduction. For example, little attention has been paid to the
importance of flower longevity; nonetheless, long-lived
flowers may be at higher risk to robbing relative to short-
lived flowers and may suffer the strongest negative effects,
assuming robbing has not selected for resistance or tolerance to
robbing. In addition, the sequence of nectar robbing and
pollination is likely important for long-lived flowers; if
pollination occurs before nectar robbing, robbing may have
little effect on plant reproduction. The importance of alternative
traits in predicting the effects of robbing on female (and male)
plant reproduction warrants further attention.
The work presented here is focused solely on the effects of
nectar robbing on female plant reproduction. Only one study,
to our knowledge, has examined the effects of robbing on
realized male reproduction (seeds sired; Irwin and Brody,
2000), although a number of studies have examined the effects
of robbing on estimates of male reproduction (pollen removal,
donation, and/or dispersal distance; Wyatt, 1980; Morris, 1996;
Irwin and Brody, 1999; Maloof, 2001; Temeles and Pan, 2002;
Richardson, 2004). Given that a large number of angiosperms
TABLE 3. Relationship between measures of Linaria vulgaris female
plant reproduction and natural levels of nectar robbing. The
relationship was insignificant for all three years.
Year Wilks’ k df FP
1997 0.077 3, 15 0.38 0.77
1998 0.017 3, 31 0.17 0.92
1999 0.019 3, 46 1.75 0.83
December 2007] BURKLE ET AL.—EFFECTS OF NECTAR ROBBING ON PLANT REPRODUCTION 1941

are hermaphroditic (Weiblen et al., 2000), studies are
desperately needed that test the effects of nectar robbing, in
combination with pollen limitation and plant life history traits,
on male plant function. Male reproduction is often more
strongly limited by pollinator visits than female reproduction
(i.e., Young and Stanton, 1990); thus, we may expect to find
stronger effects of pollinator activity on plant–robber interac-
tions through male than female plant reproduction (but see
Irwin and Brody, 2000).
In conclusion, we found that nectar robbing can have
variable effects on female plant reproduction. This variation
can be explained, at least in part, by the mating system and
level of pollen limitation of the plant, with self-incompatible
and pollen limited plants suffering from strong negative effects
of robbing. In combination with a wealth of previous research,
this study has highlighted the effects of nectar robbing on
plants (Maloof and Inouye, 2000; Irwin et al., 2001). The time
is ripe to move from the plant perspective to the perspective of
the pollinators in order to understand more fully how robbing
affects pollinator-foraging behavior and fitness. Only studies
that combine such a dual perspective will provide insight into
how nectar robbers affect coevolutionary relationships between
plants and pollinators.
LITERATURE CITED
AGRAWAL, A. A., J. A. RUDGERS,L.W.BOTSFORD,D.CUTLER,J.B.GORIN,
C. J. L
UNDQUIST,B.W.SPITZER, AND A. L. SWANN. 2000. Benefits and
constraints on plant defense against herbivores: spines influence the
legitimate and illegitimate flower visitors of yellow star thistle,
Centaurea solstitialis L. (Asteraceae). Southwestern Naturalist 45:
1–5.
A
IZEN, M. A., AND L. D. HARDER. 2007. Expanding the limits of the pollen-
limitation concept: effects of pollen quantity and quality. Ecology 88:
271–281.
A
RIZMENDI, M. C. 2001. Multiple ecological interactions: nectar robbers
and hummingbirds in a highland forest in Mexico. Canadian Journal
of Zoology 79: 997–1006.
A
RIZMENDI, M. C., C. A. DOMINGUEZ, AND R. DIRZO. 1996. The role of an
avian nectar robber and of hummingbird pollinators in the
reproduction of two plant species. Functional Ecology 10: 119–127.
A
RNOLD, R. 1982. Pollination, predation and seed set in Linaria vulgaris
(Scrophulariaceae). American Midland Naturalist 107: 360–369.
A
SHMAN, T., T. KNIGHT,J.STEETS,P.AMARASEKARE,M.BURD,D.
C
AMPBELL,M.DUDASH,M.JOHNSTON,S.MAZER,R.MITCHELL,M.
M
ORGAN, AND W. WILSON. 2004. Pollen limitation of plant
reproduction: ecological and evolutionary causes and consequences.
Ecology 85: 2408–2421.
B
OSCH, M., AND N. M. WASER. 1999. Effects of local density on pollination
and reproduction in Delphinium nuttallianum and Aconitum colum-
bianum (Ranunculaceae). American Journal of Botany 86: 871–879.
B
RONSTEIN, J. L. 2001a. The costs of mutualism. American Zoologist 41:
825–839.
B
RONSTEIN, J. L. 2001b. The exploitation of mutualisms. Ecology Letters 4:
277–287.
B
URD, M. 1994. Bateman’s principle and plant reproduction: the role of
pollen limitation in fruit and seed set. Botanical Review 60: 83–139.
C
AMPBELL, D. R., AND K. J. HALAMA. 1993. Resource and pollen limitations
to lifetime seed production in a natural plant population. Ecology 74:
1043–1051.
C
ASWELL, H. 1985. The evolutionary demography of clonal reproduction.
In J. Jackson, L. Buss, and R. Cook [eds.], Population biology and
evolution of clonal organisms, 187–224. Yale University Press, New
Haven, Connecticut, USA.
C
OHEN, J. 1969. Statistical power analysis for the behavioral sciences.
Academic Press, New York, New York, USA.
DE NETTANCOURT, D. 2001. Incompatibility and incongruity in wild and
cultivated plants. Springer-Verlag, New York, New York, USA.
D
EDEJ, S., AND K. DELAPLANE. 2004. Nectar-robbing carpenter bees reduce
seed-setting capability of honey bees (Hymenoptera: Apidae) in
rabbiteye blueberry, Vaccinium ashei,‘Climax.’Environmental
Entomology 33: 100–106.
E
RIKSSON, O., AND L. JERLING. 1990. Heirarchical selection and risk
spreading in clonal plants. In J. van Groenendael and H. de Kroon
[eds.], Clonal growth in plants, 79–94. SPB Academic Publishing,
The Hague, Netherlands.
F
RITZ, R. S., AND D. H. MORSE. 1981. Nectar parasitism of Asclepias
syriaca by ants—effect on nectar levels, pollinia insertion, pollinaria
removal and pod production. Oecologia 50: 316–319.
G
ALEN, C. 1983. The effect of nectar-thieving ants on seedset in floral
scent morphs of Polemonium viscosum. Oikos 41: 245–249.
G
IGORD, L., M. MCNAIR,M.STRITESKY, AND A. SMITHSON. 2002. The
potential for floral mimicry in rewardless orchids: an experimental
study. Proceedings of the Royal Society of London, B, Biological
Sciences 269: 1389–1395.
G
OULSON, D., J. C. STOUT,S.A.HAWSON, AND J. A. ALLEN. 1998. Floral
display size in comfrey, Symphytum officinale L. (Boraginaceae):
relationships with visitation by three bumble bee species and
subsequent seed set. Oecologia 113: 502–508.
G
UITIAN, J., J. M. SANCHEZ, AND P. GUITIAN. 1994. Pollination ecology of
Petrocoptis grandiflora Rothm (Caryophyllaceae)—a species endem-
ic to the north-west part of the Iberian Peninsula. Botanical Journal of
the Linnean Society 115: 19–27.
G
UREVITCH, J., AND L. HEDGES. 2001. Meta-analysis: combining the results
of independent experiments. In S. Scheiner and J. Gurevitch [eds.],
Design and analysis of ecological experiments, 347–369. Oxford
University Press, New York, New York, USA.
H
ELLSTROM, K., M. M. KYTOVIITA,J.TUOMI, AND P. RAUTIO. 2006. Plasticity
of clonal integration in the perennial herb Linaria vulgaris after
damage. Functional Ecology 20: 413–420.
H
IGASHI, S., M. OHARA,H.ARAI, AND K. MATSUO. 1988. Robber-like
pollinators—overwintered queen bumble bees foraging on Corydalis
ambigua. Ecological Entomology 13: 411–418.
H
USBAND, B. C., AND D. W. SCHEMSKE. 1996. Evolution of the magnitude
and timing of inbreeding depression in plants. Evolution 50: 54–70.
I
NOUYE, D. 1980. The terminology of floral larceny. Ecology 61: 1251–
1253.
I
NOUYE, D. W., W. A. CALDER, AND N. M. WASER. 1991. The effect of floral
abundance on feeder censuses of hummingbird populations. Condor
93: 279–285.
I
RWIN, R. E. 2000. Hummingbird avoidance of nectar-robbed plants:
spatial location or visual cues. Oikos 91: 499–506.
I
RWIN, R. E. 2003. Impact of nectar robbing on estimates of pollen flow:
conceptual predictions and empirical outcomes. Ecology 84: 485–
495.
I
RWIN, R. E. 2006. The consequences of direct versus indirect species
interactions to selection on traits: pollination and nectar robbing in
Ipomopsis aggregata. American Naturalist 167: 315–328.
I
RWIN, R. E., AND A. K. BRODY. 1998. Nectar robbing in Ipomopsis
aggregata: effects on pollinator behavior and plant fitness. Oecologia
116: 519–527.
I
RWIN, R. E., AND A. K. BRODY. 1999. Nectar-robbing bumble bees reduce
the fitness of Ipomopsis aggregata (Polemoniaceae). Ecology 80:
1703.
I
RWIN, R. E., AND A. K. BRODY. 2000. Consequences of nectar robbing for
realized male function in a hummingbird-pollinated plant. Ecology
81: 2637–2643.
I
RWIN, R. E., A. K. BRODY, AND N. M. WASER. 2001. The impact of floral
larceny on individuals, populations, and communities. Oecologia
129: 161–168.
I
RWIN, R. E., AND J. E. MALOOF. 2002. Variation in nectar robbing over
time, space, and species. Oecologia 133: 525–533.
K
EARNS, C., D. INOUYE, AND N. M. WASER. 1998. Endangered mutualisms:
the conservation of plant–pollinator interactions. Annual Review of
Ecology and Systematics 29: 83–112.
1942 AMERICAN JOURNAL OF BOTANY [Vol. 94

KJONAAS, C., AND C. RENGIFO. 2006. Differential effects of avian nectar-
robbing on fruit set of two Venezuelan Andean cloud forest plants.
Biotropica 38: 276–279.
L
ARA, C., AND J. F. ORNELAS. 2001. Preferential nectar robbing of flowers
with long corollas: experimental studies of two hummingbird species
visiting three plant species. Oecologia 128: 263–273.
L
ARA, C., AND J. F. ORNELAS. 2002. Effects of nectar theft by flower mites
on hummingbird behavior and the reproductive success of their host
plant, Moussonia deppeana (Gesneriaceae). Oikos 96: 470–480.
L
ASSO, E., AND M. NARANJO. 2003. Effect of pollinators and nectar robbers
on nectar production and pollen deposition in Hamelia patens
(Rubiaceae). Biotropica 35: 57–66.
M
ALOOF, J. E. 2000. The ecological effects of nectar robbers, with an
emphasis on the reproductive biology of Corydalis caseana.
University of Maryland, College Park, Maryland, USA.
M
ALOOF, J. E. 2001. The effects of a bumble bee nectar robber on plant
reproductive success and pollinator behavior. American Journal of
Botany 88: 1960–1965.
M
ALOOF, J. E., AND D. W. INOUYE. 2000. Are nectar robbers cheaters or
mutualists? Ecology 81: 2651–2661.
M
CDADE, L. A., AND J. A. WEEKS. 2004. Nectar in hummingbird-pollinated
neotropical plants. II. Interactions with flower visitors. Biotropica 36:
216–230.
M
ORRIS, W. F. 1996. Mutualism denied? Nectar-robbing bumble bees do
not reduce female or male success of bluebells. Ecology 77: 1451–
1462.
M
ORRIS, W. F., R. A. HUFBAUER,A.A.AGRAWAL,J.D.BEVER,V.A.
B
OROWICZ,G.S.GILBERT,J.L.MARON,C.E.MITCHELL,I.M.PARKER,
A. G. P
OWER,M.E.TORCHIN, AND D. P. VAZQUEZ. 2007. Direct and
interactive effects of enemies and mutualists on plant performance: a
meta-analysis. Ecology 88: 1021–1029.
N
AVARRO, L. 2000. Pollination ecology of Anthyllis vulneraria subsp.
vulgaris (Fabaceae): nectar robbers as pollinators. American Journal
of Botany 87: 980–985.
N
EWMAN, D. A., AND J. D. THOMSON. 2005. Effects of nectar robbing on
nectar dynamics and bumble bee foraging strategies in Linaria
vulgaris (Scrophulariaceae). Oikos 110: 309–320.
N
ORMENT, C. J. 1988. The effect of nectar-thieving ants on the
reproductive success of Frasera speciosa (Gentianaceae). American
Midland Naturalist 120: 331–336.
P
LEASANTS, J., AND M. ZIMMERMAN. 1983. The distribution of standing crop
of nectar: what does it really tell us? Oecolgia 57: 412–414.
P
RICE, M. V., AND N. M. WASER. 1979. Pollen dispersal and optimal
outcrossing in Delphinium nelsonii. Nature 277: 294–297.
P
YKE, G. 1982. Local geographic distributions of bumble bees near
Crested Butte, Colorado: competition and community structure.
Ecology 63: 555–573.
R
ENCHER, A. C. 1995. Methods of multivariate analysis. John Wiley &
Sons, New York, New York, USA.
R
ICHARDSON, S. 2004. Are nectar-robbers mutualists or antagonists?
Oecolgia 139: 246–254.
R
OBOCKER, W. 1974. Life history, ecology, and control of dalmatian
toadflax, Technical Bulletin no. 79. Washington Agricultural
Experiment Station, Pullman, Washington, USA.
R
OUBIK, D. W. 1982. The ecological impact of nectar-robbing bees and
pollinating hummingbirds on a tropical shrub. Ecology 63: 354–360.
R
OUBIK, D. W., N. M. HOLBROOK, AND G. PARRA. 1985. Roles of nectar
robbers in reproduction of the tropical treelet Quassia amara
(Simaroubaceae). Oecologia 66: 161–167.
R
UST, R. 1979. Pollination of Impatiens capensis: pollinators or nectar
robbers. Journal of Kansas Entomological Society 52: 297–308.
S
AAVEDRA, F., D. W. INOUYE,M.V.PRICE, AND J. HARTE. 2003. Changes in
flowering and abundance of Delphinium nuttallianum (Ranuncula-
ceae) in response to a subalpine climate warming experiment. Global
Change Biology 9: 885–894.
S
ANER, M., D. CLEMENTS,M.HALL,D.DOOHAN, AND C. CROMPTON. 1995.
The biology of Canadian weeds. 105. Linaria vulgaris Mill.
Canadian Journal of Plant Science 75: 525–537.
S
CHEINER, S. M. 1993. MANOVA: multiple response variables and
multispecies interactions. In S. M. Scheiner and J. Gurevitch [eds.],
Design and analysis of ecological experiments. Chapman and Hall,
New York, New York, USA.
S
CHMIDA, A., AND R. KADMON. 1991. Within-plant patchiness in nectar
standing crop in Anchusa strigosa. Vegetatio 94: 95–99.
S
ILVERTOWN,J.,M.FRANCO,I.PISANTY, AND A. MENDOZA. 1993.
Comparative plant demography—relative importance of life-cycle
components to the finite rate of increase in woody and herbaceous
perennials. Journal of Ecology 81: 465–476.
S
TOUT, J., J. ALLEN, AND D. GOULSON. 2000. Nectar robbing, forager
efficiency and seed set: bumble bees foraging on the self-
incompatible plant Linaria vulgaris (Scrophulariaceae). Acta Oeco-
logica 21: 277–283.
S
TRAUSS, S., AND R. IRWIN. 2004. Ecological and evolutionary conse-
quences of multi-species plant–animal interactions. Annual Review of
Ecology and Systematics 35: 435–466.
T
EMELES, E., AND I. PAN. 2002. Effect of nectar robbery on phase duration,
nectar volume, and pollination in a protandrous plant. International
Journal of Plant Sciences 163: 803–808.
T
HAKAR, J., K. KUNTE,A.CHAUHAN,A.WAVTE, AND M. WATVE. 2003.
Nectarless flowers: ecological correlates and evolutionary stability.
Oecologia 136: 565–570.
T
RAVERS,S.,AND S. MAZER. 2000. The absence of cryptic self-
incompatibility in Clarkia unguiculata (Onagraceae). American
Journal of Botany 87: 191–196.
T
RAVESET, A., M. F. WILLSON, AND C. SABAG. 1998. Effect of nectar-
robbing birds on fruit set of Fuchsia magellanica in Tierra del Fuego:
a disrupted mutualism. Functional Ecology 12: 459–464.
U
TELLI, A. B., AND B. A. ROY. 2001. Causes and consequences of floral
damage in Aconitum lycoctonum at high and low elevations in
Switzerland. Oecologia 127: 266–273.
W
ASER, N. M. 1978. Competition for hummingbird pollination and
sequential flowering in two Colorado wildflowers. Ecology 59: 934–
944.
W
ASER, N. M. 1979. Pollinator availability as a determinant of flowering
time in Ocotillo (Fouquieria splendens). Oecologia 39: 107–121.
W
ASER, N. M. 1982. A comparison of distances flown by different visitors
to flowers of the same species. Oecologia 55: 251–257.
W
ASER, N. M., AND M. V. PRICE. 1990. Pollination efficiency and
effectiveness of bumble bees and hummingbirds visiting Delphinium
nelsonii. Collectanea Botanica 19: 9–20.
W
ASER, N. M., AND M. V. PRICE. 1991. Outcrossing distance effects in
Delphinium nelsonii: pollen loads, pollen tubes, and seed set. Ecology
72: 171–179.
W
EIBLEN, G., R. OYAMA, AND M. DONOGHUE. 2000. Phylogenetic analysis
of dioecy in monocotyledons. American Naturalist 155: 46–58.
W
ILLIAMS, C. F., J. RUVINSKY,P.E.SCOTT, AND D. K. HEWS. 2001.
Pollination, breeding system, and genetic structure in two sympatric
Delphinium (Ranunculaceae) species. American Journal of Botany
88: 1623–1633.
W
YATT, R. 1980. The impact of nectar-robbing ants on the pollination
system of Asclepias curassavica. Bulletin of the Torrey Botanical
Club 107: 24–28.
Y
OUNG, H., AND M. STANTON. 1990. Influences of floral variation on pollen
removal and seed production in wild radish. Ecology 71: 536–547.
Y
OUNG, H., AND T. YOUNG. 1992. Alternative outcomes of natural and
experimental high pollen loads. Ecology 73: 639–647.
Z
ILKE, S., AND R. COUPLAND. 1954. The reproductive capacity of toadflax
(Linaria vulgaris Hill) by seed. Research report. National Weed
Committee, Western Section, Canada Department of Agriculture,
Ottawa, Ontario, Canada.
December 2007] BURKLE ET AL.—EFFECTS OF NECTAR ROBBING ON PLANT REPRODUCTION 1943