Consequences of secondary nectar robbing for male components of plant reproduction
Item Type Article
Authors Richman, Sarah K.; Irwin, Rebecca E.; Bosak, John T.; Bronstein, Judith L.
Citation Richman, S. K., R. E. Irwin, J. T. Bosak, J. L. Bronstein. 2018. Consequences of secondary nectar robbing for male components of plant reproduction. American Journal of Botany 105(5): 943– 949.
DOI 10.1002/ajb2.1082
Publisher BOTANICAL SOC AMER INC
Journal AMERICAN JOURNAL OF BOTANY
Rights © 2018 Botanical Society of America.
Download date 01/10/2021 12:57:00
Item License http://rightsstatements.org/vocab/InC/1.0/
Version Final accepted manuscript
Link to Item http://hdl.handle.net/10150/628275 Manuscript Click here to download Manuscript 2017_IpoRobbing_MaleFitnessNectarRemoval_ToSubmit_NoFig
1
2
3 Consequences of secondary nectar robbing for male components of plant reproduction
4
5
6 Sarah K. Richman*1,2, Rebecca E. Irwin2,3, John T. Bosak1,2, Judith L. Bronstein1,2
7
8 1Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721
9 USA
10 2Rocky Mountain Biological Laboratory, Crested Butte, CO 81224 USA
11 3Department of Applied Ecology, North Carolina State University, Raleigh, NC 27695 USA
12
13 * Corresponding author: Sarah Richman
15
16 Running headline: Secondary nectar robbing and male plant function
17
18
19
20
21
22
23 24 Abstract
25 Premise of the study: Organisms engage in multiple species interactions simultaneously.
26 While pollination studies generally focus on plants and pollinators exclusively, secondary
27 robbing, a behavior that requires other species (primary robbers) to first create access holes in
28 corollas, is common. While secondary robbing can reduce plants’ female fitness, we lack
29 knowledge about its impact on male plant fitness.
30 Methods: We experimentally simulated primary and secondary robbing in the
31 monocarpic perennial Ipomopsis aggregata (Polemoniaceae), then quantified direct effects on
32 pollen loss and indirect effects on pollinator-mediated pollen (dye) donation. We also assessed
33 whether continual nectar removal via the floral opening has similar effects on pollinator behavior
34 as continual secondary robbing through robber holes.
35 Key Results: We found no significant direct or indirect effects of secondary robbing on
36 Ipomopsis male fitness. Secondary robbers dislodged some pollen grains, but there was no
37 statistically significant difference between robbing treatments or between secondary robbing and
38 the control (unrobbed) treatment. Although robbing did reduce pollen (dye) donation due to
39 hummingbird-pollinator avoidance of robbed plants, pollen donation did not differ between the
40 two robbing treatments. The effects of secondary robbing on hummingbird behavior resembled
41 chronic nectar removal by pollinators, providing further evidence that pollinators may cue in on
42 nectar rewards to make foraging decisions.
43 Conclusions: This study combined with prior research suggests that secondary robbing is
44 not as costly to male fitness as it is to female fitness in Ipomopsis, broadening our knowledge of
45 the overall costs of mutualism exploitation to total plant fitness.
46
Richman et al. 2 47 Key words: Ipomopsis aggregata; hummingbird pollination; male plant fitness; nectar
48 robbing; pollen loss; dye donation
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
Richman et al. 3 70 INTRODUCTION
71 Mutualisms are prone to exploitation, in which organisms obtain benefits or rewards
72 while providing no benefits to the partner in return (Bronstein, 2001). Just as organisms may
73 interact with multiple mutualists simultaneously or sequentially, they may also interact with
74 multiple exploiters. For example, Pseudocabima caterpillars usurp territory on Cecropia saplings
75 from mutualistic ant defenders and simultaneously promote colonization by a fungal pathogen
76 that persists inside of Cecropia domatia in the absence of ants (Roux et al., 2011). Similarly,
77 pollination mutualisms can experience exploitation from nectar robbers that obtain nectar from
78 flowers using behaviors that generally do not lead to pollination (Irwin et al., 2010). Primary
79 nectar robbers feed on nectar through holes they make in flowers, in turn opening opportunities
80 for facilitated exploitation (Richman et al., 2017) by secondary robbers, species that can remove
81 nectar through those holes (Inouye, 1980). Secondary robbing is common, with at least 315
82 reports from 117 plant species in 34 families reported thus far (Irwin et al., 2010 and unpubl.
83 data). In general, multispecies interactions can result in additive and in some cases non-additive
84 effects (Morris et al., 2007), and sometimes differential effects through components of male and
85 female plant reproduction (Schaeffer et al., 2013). However, in contrast to our growing
86 understanding of multispecies interactions (Strauss and Irwin, 2004; Nunn et al., 2014), we know
87 comparatively little about the frequency or importance of multispecies exploitation of mutualism.
88 Multispecies exploitation may lead to unexpected consequences for whole-plant fitness
89 via differential effects on female and male functions. The majority of flowering species are
90 hermaphroditic. Nonetheless, most studies quantify the effects of species interactions on whole-
91 plant reproduction using female components of reproduction as a surrogate (Stanton et al., 1986).
92 While male and female components of reproduction often respond similarly to abiotic and biotic
Richman et al. 4 93 interactions (Schaeffer et al., 2013), there are also cases of sexual conflict (Barrett, 2002) in
94 which environmental contexts that make plants better male parents do not make them better
95 female parents, and vice versa (Contreras and Ornelas, 1999; Madjidian, 2009). In particular,
96 theory predicts that female components of plant reproduction should be more limited by
97 resources whereas male components should be more limited by mating opportunities (Bateman,
98 1948), although this prediction has been debated (Wilson et al., 1994). Given that nectar robbing
99 can indirectly affect plant reproduction via changes in pollinator behavior and subsequent mating
100 opportunities, and that several studies show that robbing results in reductions in male
101 components of plant reproduction (Irwin et al., 2010, and references therein), there is reason to
102 suspect that secondary robbing may have additional effects on male components of plant
103 reproduction beyond that of primary robbing alone, especially in cases where pollinator behavior
104 is strongly affected by nectar availability. There may also be direct effects of robbing on male
105 components of plant reproduction. For example, robbers may jar flowers hard enough while
106 landing as to dislodge pollen from anthers. The more times flowers are landed upon by these
107 visitors, the less pollen might become available for seed siring. The degree to which secondary
108 robbing affects male components of plant reproduction beyond that of primary robbing is
109 unknown.
110 The goal of this work was to assess if and how facilitated exploitation by primary
111 followed by secondary nectar robbing affects male components of plant reproduction. We
112 studied the hummingbird-pollinated plant Ipomopsis aggregata (Polemoniaceae) which
113 experiences primary robbing by bumble bees and secondary robbing by bumble bees, flies, and
114 wasps. Prior research has shown that secondary robbing inflicts additional costs on female plant
115 reproduction beyond that of primary robbing alone (Richman et al., 2017). Here we examine
Richman et al. 5 116 three ways in which secondary nectar robbing might affect male fitness and/or pollinator
117 behavior. First, we asked whether primary nectar robbers can directly affect pollen loss by
118 dislodging pollen from anthers in the process of landing on flowers, and whether additional
119 landings by secondary robbers lead to more pollen being dislodged. Ipomopsis anthers range
120 from being inserted in the corolla tube to exerted outside the corolla tube (Waser and Price,
121 1984). Thus, pollen could be dislodged from the anthers either onto the corolla tube bottom
122 and/or lost from the front of the flower in the air and eventually onto the ground. We predicted
123 that pollen loss would be minimal due to the relatively weak force of bumble bees when they
124 land on flowers to rob; however, if anthers do lose pollen, we predicted that additional secondary
125 robbing would result in more pollen being dislodged. Second, we asked whether secondary
126 nectar robbing affected pollinator visitation and subsequent pollen (dye) donation beyond that of
127 primary robbing. Because robbing of Ipomopsis indirectly decreases pollinator visitation via
128 decreasing nectar rewards (Irwin and Brody, 1998) and secondary robbed flowers generally
129 receive fewer visits than primary robbed flowers (Richman et al., 2017), we predicted that plants
130 with secondary-robbed flowers would donate less pollen. Third, finding effects of robbing in
131 general on pollinator behavior, we then explored whether the effects of robbing on pollinator
132 visitation are mechanistically equivalent to chronic nectar removal by any other visitor, or if
133 robbing inflicts unique additional effects on pollinator visitation that cannot be predicted simply
134 from the removal of nectar. Secondary robbing can result in chronic nectar removal from
135 flowers, as can high visitation rates by pollinators, although the former does not result in pollen
136 deposition in this system whereas the later does (Irwin et al., 2015). By addressing the direct and
137 indirect effects of primary as well as secondary robbing on male fitness, which is underexplored
Richman et al. 6 138 relative to female fitness, this work combined with prior research provides empirical insight into
139 the total plant fitness costs associated with multispecies exploitation of pollination.
140
141 MATERIALS AND METHODS
142 Study System—
143 We studied Ipomopsis aggregata (Polemoniaceae; hereafter Ipomopsis) at the Rocky
144 Mountain Biological Laboratory (RMBL, elevation 2895 m), Gothic, Colorado, USA. Ipomopsis
145 is a monocarpic, perennial, self-incompatible herb that produces approximately 50 red, tubular
146 flowers on (usually) a single stalk from mid-June to mid-August (Waser, 1978). The
147 hermaphroditic flowers are protandrous, with male phase lasting 1-2 days and female phase
148 lasting 2-3 days (Waser, 1978). Pollen is dispersed an average of 1.27-2.63 m from the parent
149 plant (Campbell and Waser, 1989); because seed dispersal is limited, pollen movement is
150 thought to be a major component of gene flow (Levin and Kerster, 1974; Campbell and Waser,
151 1989). Flowers continually produce nectar at a rate of up to 5 μL/day and nectar removal does
152 not affect subsequent nectar production rate (Pleasants, 1983). Broad-tailed (Selasphorus
153 platycercus) and rufous (Selasphorus rufus) hummingbirds visit Ipomopsis flowers for nectar
154 rewards and are the primary pollinators (Mayfield et al., 2001).
155 The bumble bee Bombus occidentalis (Apidae) primary-robs Ipomopsis flowers by
156 piercing the corolla tissue using toothed mandibles, generally removing all available nectar.
157 Nectar production continues following a primary robbing event (Irwin et al., 2015), which
158 encourages secondary robbing by B. occidentalis as well as by other Bombus species, wasps, and
159 flies that lack the mouthparts necessary to primary-rob. Nectar robbing has been shown to reduce
160 Ipomopsis female fitness: it discourages pollinator visitation, which leads to reduced fruit and
Richman et al. 7 161 seed set (Irwin and Brody, 1998, 1999). Moreover, secondary robbing of Ipomopsis results in a
162 greater female fitness reduction than primary robbing alone (Richman et al. 2017). Nectar
163 robbing also reduces estimates of male plant reproduction, including pollen donation and the
164 number of seeds sired, due to hummingbird-pollinator avoidance of robbed plants and flowers
165 (Irwin and Brody, 1999, 2000). However, Irwin and Brody (1999, 2000) did not separate the
166 effects of primary and secondary robbing on male components of plant reproduction, which we
167 do here.
168
169 Field methods—
170 1) Does secondary robbing directly result in pollen being knocked off anthers beyond that
171 caused by primary robbing?
172 We haphazardly selected 36 Ipomopsis plants for manipulations in a single population
173 (GPS: 38.9405 N, -107.0230 W) in July 2016. We focused on plants that had three unrobbed
174 flowers in early male phase with fresh, dehiscing anthers. In rare instances, if we could not find
175 three unrobbed flowers in male phase on the same plant, we selected a flower on an adjacent
176 plant. The three flowers were each assigned to one of three robbing treatments: (1) Primary
177 robbing, (2) Primary and secondary robbing, and (3) Control (unrobbed). We simulated a
178 bumble bee landing on a flower to rob by gently tapping the flower. While using a living robber
179 bee to exert the flower movement would have been the most realistic, it was not possible to do so
180 while also collecting the plume of pollen that might be knocked off flowers and into the air;
181 doing so would have required working in very close proximity to the robber, and would have
182 interfered with its foraging behavior. It is likely that our tapping was more vigorous than the
183 effects exerted by bumble bees landing on flowers to rob (R. E. Irwin, pers obs); thus, our results
184 should be viewed as an upper bound in terms of the direct effects of robbing on pollen loss. To
Richman et al. 8 185 simulate primary robbing, we tapped the flower once. To simulate primary plus secondary
186 robbing (or two robbing events in general), we tapped the flower twice. The control flower was
187 not tapped.
188 To collect pollen lost to the air after flowers were tapped, we held a piece of basic fuchsin
189 gelatin (Kearns and Inouye, 1993) directly at the edge of the flower entrance as the flower was
190 being tapped. To collect pollen that was potentially lost on the bottom of the flower, we removed
191 each flower (being careful not to vigorously move it), inverted it, and removed the dehiscing
192 anthers. We then cut the flower open lengthwise and used a separate piece of basic fuchsin
193 gelatin to collect pollen from the interior of the corolla. A similar procedure was used for the
194 Control flowers. Fine-point forceps and dissecting scissors used to hold fuchsin gelatin, cut
195 flowers, or remove anthers were washed with 70% ethanol after each flower. The fuchsin gelatin
196 pieces were fixed onto microscope slides, and we counted the number of Ipomopsis pollen grains
197 on each slide under a compound microscope.
198 Statistical analysis— We tested whether treatment caused loss of pollen grains (i) to the
199 inside of the corolla and (ii) in a plume of pollen outside of the corolla, using GLM (Negative
200 Binomial family, log link, followed by a Likelihood Ratio Test). Because flowers in the Control
201 treatment were not tapped, we did not collect fuchsin gelatin samples for the plume of pollen
202 outside of the corolla; thus, the only treatment levels for the plume analysis were Primary
203 robbing and Primary and secondary robbing. All analyses were performed in R version 3.3.1 (R
204 Core Team, 2016).
205
206 2) How does secondary robbing indirectly affect hummingbird pollinator visitation and pollen
207 (dye) donation beyond that of primary robbing?
Richman et al. 9 208 We potted 60 budding Ipomopsis from a population south of the RMBL (GPS: 38.7806
209 N, -106.8703 W) on 7 June 2016 and maintained plants in an enclosure. We measured the height
210 of each plant to the nearest cm to use as a covariate in statistical analyses, as prior research has
211 shown that taller Ipomopsis are more likely to be visited by pollinators (Brody and Mitchell,
212 1997). We randomly assigned 20 plants each to three treatments applied at the whole-plant level:
213 (1) Primary robbing (all flowers on each plant were primary robbed one time), (2) Primary and
214 secondary robbing (all flowers on each plant were primary robbed and then secondary robbed
215 once daily), and (3) Control (no robbing). We applied treatments for six consecutive days and
216 started pollinator observations on the second day of treatments. All flowers in all treatments were
217 physically handled to control for effects of touching flowers.
218 To simulate primary robbing, we cut an ~1 mm hole in the side of the corolla with
219 dissecting scissors and removed all available nectar with a 10 μL microcapillary tube
220 (Drummond Scientific, Broomall, Pennsylvania, USA) inserted into the hole. These experimental
221 robbing techniques do not damage nectar-producing structures in flowers (Irwin and Brody,
222 1998; Irwin et al., 2015) and simulate natural robbing in terms of effects on hummingbird
223 pollinator visitation and plant reproduction (Irwin and Brody, 1998). To simulate secondary
224 nectar robbing, we inserted a 10 ul microcapillary tube into the primary robbing holes to remove
225 any additional nectar that was produced once daily (as in Richman et al., 2017). Every day that
226 robbing treatments were performed, we also recorded the number of open flowers on each plant.
227 Following robbing treatments, we placed plants into the field in a 6 m x 10 m array with
228 1-m spacing between plants, matching spacing of Ipomopsis individuals in natural populations.
229 Treatments and plants were assigned randomly to array positions at the start of the experiment
230 and were kept in those same positions daily. After placing plants in the field each day, we
Richman et al. 10 231 observed pollinator behavior for at least 3 hr. Observations began on the second day of treatment
232 applications, to allow for a difference in nectar volume between primary and secondary robbed
233 flowers. For each hummingbird that entered the array, we recorded species and sex, plants
234 visited, and the number of flowers probed per plant. Afterwards, we returned plants to the
235 enclosure until the next day of treatments and observations.
236 To estimate pollen donation, we used powdered fluorescent dyes as pollen analogues
237 (Series JST-300, Radiant Color, Richmond, California, USA). In Ipomopsis, mean dye donation
238 provides a reliable estimate of mean pollen donation (Waser and Price, 1982). We used three dye
239 colors, each assigned at random to one of the treatments. On 14 June 2016, one half of the plants
240 in each treatment were randomly chosen to act as dye donors and the other half of the plants in
241 each treatment as recipients. Dye was applied to the anthers of flowers in male phase with
242 dehiscing pollen using a flathead toothpick. We recorded the number of flowers dyed per donor
243 plant as well as the number of flowers open. Dye was applied in the morning just after placing
244 plants into the field. At the end of the approx. 3-hr pollinator observation period, we collected
245 stigmas from 20% of the female-phase flowers from recipient plants. We counted the number of
246 dye particles of each color on each stigma using a dissecting microscope (as in Irwin and Brody,
247 1999). We repeated this procedure on 17 June 2016, switching the donor and recipient plants,
248 and re-assigning treatments at random to dye colors. Ipomopsis flowers on the experimental
249 plants were lasting approx. 3 d (R. E. Irwin, per obs); thus, with at least 3 d between dye
250 applications, we ensured that any dye from the previous application was no longer in the array on
251 dehiscing flowers or open stigmas. For each recipient plant, we calculated the mean number of
252 dye particles donated to recipient plants per treatment per flower dyed (similar to Dudash et al.,
Richman et al. 11 253 2011). Calculating dye donation on a per-flower dyed basis controlled for any differences in the
254 number of flowers dyed in the three treatments (Campbell, 1989).
255 Statistical analyses— To test whether robbing treatments affected hummingbird foraging
256 behavior, we calculated visitation rate as the number of times plants were visited multiplied by
257 the mean percentage of flowers probed. We used ANCOVA to test if robbing treatment affected
258 hummingbird visitation rate with plant height and mean floral display size as covariates. Neither
259 covariate had a significant effect on hummingbird visitation rate (F1,55 = 1.17, P = 0.28) and were
260 removed from the final analysis.
261 To test whether robbing treatments affected pollen (dye) donation per flower dyed, we
262 used a linear mixed model with robbing treatment, round of dye application, and their interaction
263 as fixed factors. The interaction between robbing treatment and round of dye application was not
264 statistically significant (F2,104 = 1.86, P = 0.16) and so was removed from the final model.
265 Because recipient plants could receive dye from all three donor colors, we included plant ID as a
266 random effect in the analysis to account for observations of multiple dye colors donated to
267 recipient stigmas. Analyses were performed using JMP Pro version 13.0.0.
268
269 3) Are the effects of secondary robbing on pollinator visitation equivalent to chronic nectar
270 removal by pollinators?
271 We conducted whole-plant manipulations to simulate nectar robbing and chronic nectar
272 removal in late June and early July, 2016. We transplanted 60 single-stalked, budding Ipomopsis
273 from a single population at the RMBL (GPS: 38.9585 N, -106.9875 W) into individual pots.
274 Plants were subsequently maintained in an enclosure. We measured the height of each plant to
275 the nearest cm to use as a covariate in statistical analyses.
Richman et al. 12 276 We randomly assigned 20 plants each to one of three treatments: (1) Primary and
277 secondary robbing, (2) Chronic nectar removal, (3) Control. Nectar robbing was performed as
278 described above. In the chronic nectar removal treatment, we removed nectar from all open
279 flowers daily through the floral opening as a pollinator would using 10 μL microcapillary tubes.
280 Treatments were performed daily in the morning before placing plants into the field, and we
281 counted the number of open flowers.
282 Plants were placed into a field array as described above. Pollinators were observed for 9
283 d, starting at 0830 until 1500 or until 10 foraging bouts had been observed. We limited daily
284 observations to 10 foraging bouts to ensure that hummingbird pollinators experienced assigned
285 treatments and not nectar-empty flowers even in the control treatment. We used a digital voice
286 recorder to record the species and sex of the floral visitors, which plants were visited, and the
287 number of flowers probed per plant. At the end of daily observations, the plants were returned to
288 the enclosure.
289 Statistical analyses— We used ANCOVA to test whether treatments affected
290 hummingbird pollinator visitation rate (number of times plants were visited per day multiplied
291 by the mean percentage of flowers probed) with plant height (cm) used as a covariate. Finding
292 significant effects of nectar treatment on hummingbird visitation rate (see Results), we then used
293 similar ANCOVAs to assess the degree to which the number of times plants were visited or the
294 percentage of flowers probed was driving the behavioral results. All significant ANCOVAs were
295 followed by Tukey’s HSD test to assess pairwise treatment comparisons. Analyses were
296 performed in R version 3.3.1 (R Core Team, 2016).
297
298
Richman et al. 13 299 RESULTS
300 1) Does secondary robbing directly result in pollen being knocked off anthers beyond that
301 caused by primary robbing?
302 Simulated secondary robbing resulted in negligible dislodgement of pollen. Flowers in
303 the Primary and secondary robbing treatment lost an average of 15% more pollen grains onto
304 the corolla interior compared to the Primary robbing treatment, and 17% more than the Control
305 (Fig. 1a). However, the overall effect of treatment on pollen grains lost was not statistically
2 306 significant (χ = 0.97, df = 2, P = 0.62). Overall, simulated robbing resulted in very few pollen
307 grains being dislodged in a plume outside of the corolla opening in either robbing treatment
308 (Fig1b), and the difference in grains lost to the outside of the corolla opening was not statistically
2 309 significant (χ = 0.0001, df = 2, P = 0.99).
310
311 2) How does secondary robbing indirectly affect hummingbird pollinator visitation and pollen
312 (dye) donation beyond that of primary robbing?
313 We recorded 53 hummingbird foraging bouts over 5 observation days. All but one was
314 made by male broad-tailed hummingbirds (at least two individuals) with the remaining bout by a
315 female broad-tailed hummingbird. Robbing treatment had a significant effect on hummingbird
316 visitation rate (F2,57 = 4.86, P = 0.01), with plants in the Control treatment experiencing at least
317 20% higher pollinator visitation rate than either of the robbing treatments (Fig. 2a). A post-hoc
318 analysis revealed no significant difference in pollinator visitation rate between plants in the
319 Primary vs. Primary and secondary treatments (P greater than 0.05). The difference in pollinator
320 visitation rate between Control and robbing treatments was driven by a reduction in the number
Richman et al. 14 321 of times that robbed plants were visited (F2,57 = 37.02, P = 0.009); there was no difference in the
322 mean proportion of flowers probed per visit (F2,57 = 0.17, P = 0.84).
323 Changes in pollinator visitation to robbed plants produced changes in pollen (dye)
324 donation. We found a significant effect of robbing treatment on pollen (dye) donation, with
325 plants in the Control treatment donating at least 78% more dye per flower compared to plants in
326 either robbing treatment (F2,106 = 3.14, P = 0.047). Plants in the Primary robbing treatment
327 donated twice the dye as plants in the Primary and secondary robbing treatment, but a post-hoc
328 analysis revealed that this difference was not statistically significant (P greater than 0.05; Fig.
329 2b). Finally, plants in the second round of dye application donated 89% more dye per dyed
330 flower than plants in the first round of dye application (F1,52 = 5.81, P = 0.02).
331
332 3) Are the effects of secondary robbing on pollinator visitation equivalent to chronic nectar
333 removal by pollinators?
334 We observed 214 foraging bouts by broad-tailed hummingbirds over 7 observation days.
335 Three bouts were made by female hummingbirds with the remaining bouts by males.
336 Hummingbird visitation rate was 44% higher in the Control treatment than either the Primary
337 and secondary robbing and Chronic nectar removal treatments (F2,55 = 5.98, P = 0.004; Fig. 3a).
338 A post-hoc analysis revealed both pairwise comparisons with the Control treatment to be
339 statistically significant (Control vs. Primary and secondary robbing, P = 0.01; Control vs.
340 Chronic nectar removal, P = 0.01). However, there was no difference in visitation between the
341 Primary and secondary robbing treatment and the Chronic nectar removal treatment (P greater
342 than 0.05), suggesting that nectar removal in both treatments yielded similar effects on pollinator
Richman et al. 15 343 visitation rate. For the covariate, taller plants experienced a higher visitation rate (F1,55 = 8.19, P
344 = 0.006).
345 The effect of the treatment on visitation rate was driven by a difference in the mean
346 proportion of flowers probed per plant (F2,55 = 3.94, P = 0.02). Plants in the Control treatment
347 experienced a 27% and 35% higher proportion of flowers probed than the Primary and
348 secondary robbing and Chronic nectar removal treatments, respectively (Fig. 3b). A post-hoc
349 analysis revealed the difference between the Control and Chronic nectar removal to be the only
350 significant pairwise comparison (P = 0.02). There was no effect of treatment on the number of
351 visits per plant per day (F2,55 = 2.58, P = 0.09). The covariate plant height revealed that taller
352 plants experienced more visits overall (F1,55 = 17.00, P less than 0.001).
353
354 DISCUSSION
355 Exploitation of mutualism via primary nectar robbing can result in opportunities for
356 additional facilitated exploitation via secondary nectar robbing; however, studies documenting
357 the combined effects of multispecies exploitation are lacking. Furthermore, despite the high
358 frequency of secondary nectar robbing in nature (Irwin et al., 2010), its fitness consequences for
359 male components of plant reproduction and the mechanisms that might underlie such effects
360 have been minimally explored. Here, we report that the effects additional facilitated exploitation
361 may be minimal for some components of plant fitness (i.e., male function), and stronger for
362 others (i.e., female function; Richman et al., 2017). We show that the amount of pollen dislodged
363 from anthers and unavailable for export did not differ significantly between robbing treatments
364 or from the control (unrobbed) treatment. Robbing did reduce per-flower pollen donation relative
365 to control plants, but we did not detect a significant difference between primary and secondary
Richman et al. 16 366 robbing. Finally, hummingbird pollinators were no more likely to avoid secondary-robbed
367 flowers than they were to avoid intact flowers that had experienced chronic nectar removal,
368 suggesting that robbing per se does not affect pollinator visitation rate. Rather, it is simply the
369 availability of nectar that affects pollinator behavior. Robbing is one of many factors that reduce
370 nectar availability, and hummingbird pollinators may use similar cues associated with lack of
371 nectar to make foraging decisions and avoid unrewarding flowers.
372
373 Effect of nectar robbing on estimates of male plant reproduction
374 We found no evidence that secondary nectar robbing affected male components of plant
375 reproduction beyond that of primary robbing. First, secondary robbing did not directly result in
376 significantly more pollen being dislodged from anthers and lost from the system. We suspect this
377 is the case because the force that nectar robbers exert when landing on a flower to rob is
378 relatively weak and likely not strong enough to dislodge large amounts of pollen from the
379 anthers. We did not use live bees for this experiment, but instead simulated robbing events by
380 tapping flowers; we suspect that our tapping likely represents the upper bound of force exerted
381 by a bee landing, although follow-up studies using bees are needed to verify this assumption.
382 Nonetheless, there are several ways in general in which pollen could be passively dislodged from
383 anthers, including wind, rain, or passing animals (Inouye et al., 1994). Therefore, it is reasonable
384 to suspect that if the mechanisms plants possess for holding pollen onto anthers are strong
385 enough to prevent dislodgement by these causes, they are strong enough to prevent dislodgement
386 by nectar robbers landing on flowers. Electrostatic forces have been hypothesized as a
387 mechanism of reducing pollen wastage (Buchmann, 1985; Vaknin et al., 1999), and evidence for
Richman et al. 17 388 this phenomenon is building (Clarke et al., 2017). However, the majority of this work describes
389 efficient transfer of pollen to pollinators, rather than reducing pollen wastage pre-dispersal.
390 Second, robbed flowers donated significantly less pollen on a per-flower basis relative to
391 the control; however, we did not find a statistically significant difference between primary vs.
392 primary plus secondary robbing on pollen (dye) donation per flower dyed. These results match
393 expectations based on our measurements of hummingbird-pollinator behavior, in which
394 pollinators reduced rates of visitation to both robbing treatments relative to the control, but there
395 was no difference in visitation rate between the two robbing treatments. Thus, this work suggests
396 that within the context of our experiment, hummingbird pollinators were most sensitive to initial
397 nectar removal (via primary robbing). Consistent with these results, in a flight cage experiment
398 observing the behavior of Phaethornia longirostris hummingbirds at artificial feeders, Garrison
399 and Gass (1999) found that birds avoided feeders after a sudden removal of nectar, returning 1-2
400 hours later. Ipomopsis flowers slowly refill with nectar over several days via continuous
401 secretion (Pleasants, 1983), which may lead the sudden removal of available nectar to
402 dramatically affect pollinator behavior. One caveat is that we only conducted this experiment
403 over the course of 5 days, which may not have provided enough time for differences in nectar
404 availability to build between the primary and secondary robbing treatments. Future experiments
405 that assess how sensitive pollinators are to small differences in nectar standing crops would yield
406 important predictive insight.
407 Primary and secondary robbing resulted in similar effects on pollen (dye) donation (an
408 estimate of male plant reproduction; Schaeffer et al. 2013) whereas prior work found that
409 secondary robbing resulted in additional reductions in female plant reproduction relative to
410 primary robbing alone (Richman et al., 2017). In both studies, the fitness reductions between
Richman et al. 18 411 robbing treatments or relative to control (unrobbed) plants can be attributed to fewer mating
412 events via changes in pollinator behavior. These differences in results between studies are
413 intriguing in the context of Bateman’s principle, which predicts male fitness should be more
414 sensitive to missed mating opportunities than female fitness (Bateman, 1948). However,
415 alternative hypotheses posit that this need not always be the case (Wilson et al., 1994). In
416 particular, it has been argued that missed mating opportunities should negatively affect female
417 fitness as well, if the species experiences pollen limitation (Burd, 1994). There is evidence that
418 Ipomopsis is pollen limited, although this may be spatiotemporally variable (Hainsworth et al.,
419 1985). In other systems, effects on male versus female fitness can be highly variable, with no
420 clear pattern of one sexual function performing better than the other, although floral
421 attractiveness to pollinators seems to be a common factor. For example, male and female success
422 were correlated with unique floral morphological characteristics (which presumably contributed
423 to pollinator attraction) in Polemonium viscosum (Polemoniaceae), and therefore did not
424 experience differential male-female performance (Galen and Stanton, 1989). Third-party
425 interactions with herbivores (which can affect floral displays) have been shown to either increase
426 (Carper et al., 2016) or decrease (Mutikainen and Delph, 1996) male performance relative to
427 female performance, due to indirect effects via pollinator behavior and pollen tube growth
428 limitation, respectively. Future work that measures the effects of secondary robbing on male and
429 female function on the same plants will yield additional insight, especially if male and female
430 performance as a function of robbing is spatiotemporally variable, as will the measurement of the
431 key hummingbird behaviors that drive each component of plant reproduction.
432
433
Richman et al. 19 434 Nectar robbing versus other forms of nectar removal
435 Hummingbird pollinators foraging on Ipomopsis had similar visitation rates to plants with
436 flowers that had been secondary robbed relative to plants with flowers where nectar had been
437 chronically drained from the corolla opening. This lends further support to the hypothesis that
438 hummingbird pollinators are cuing in on lack of nectar to avoid robbed plants and flowers and
439 not other proximal cues, such as the robber hole (Irwin, 2000). Many factors can affect levels of
440 nectar standing crops, for instance, continual visitation not only by pollinators but also by nectar
441 thieves (organisms that collect nectar in a manner that precludes pollination but that do not
442 damage floral tissue; Inouye, 1980). Nectar reduction by nectar thieves reduced hummingbird
443 visitation and seed production in Bouvardia turnifolia (Rubiaeceae) (Torres et al., 2008).
444 Conversely, nectar-thieving mites stimulated nectar production in Moussonia deppeana
445 (Gesneriaceae), resulting in increased visitation by hummingbirds and increased seed production
446 (Lara and Ornelas, 2002). In both of these cases, changes in nectar availability seem to be the cue
447 that affects bird foraging decisions and behaviors. Given the challenges hummingbirds face in
448 meeting their daily energetic requirements (Wolf et al., 1972), there may be strong selection for
449 them to cue in on nectar to make foraging decisions, particularly if the presence vs. absence of
450 robber holes is not a reliable indicator of nectar rewards. Indeed, several studies demonstrate
451 hummingbirds’ ability to make foraging decisions based on nectar availability via visual cues,
452 spatial cues, or a combination of the two (Hurly and Healy, 2002; Gonzalez-Gomez and
453 Vasqeuz, 2011). However, the exact cue that nectar provides to hummingbirds in the absence of
454 spatial and visual cues remains a mystery (Irwin, 2000).
455
456
Richman et al. 20 457 Footnotes
458 1Manuscript received ______; revision accepted ______
459
460 Acknowledgements
461 The authors thank K. Brennan, S. Spalding, and K. Wang for help with field work, the RMBL
462 for providing access to field sites. Funding for this research was provided by the National
463 Science Foundation (DGE-1143953 to SKR, DEB-1354061 to JLB and DEB-1641243 including
464 an REU supplement to REI).
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
Richman et al. 21 480 Literature cited
481 BARRETT, S.C.H. 2002. Sexual interference of the floral kind. Heredity 88: 154–159.
482 BATEMAN, A. 1948. Intra-sexual selection in Drosophila. Heredity 2: 349–368.
483 BRODY, A.K., and R.J. MITCHELL. 1997. Effects of experimental manipulation of inflorescence
484 size on pollination and pre-dispersal seed predation in the hummingbird-pollinated plant
485 Ipomopsis aggregata. Oecologia 110: 86–93.
486 BRONSTEIN, J.L. 2001. The exploitation of mutualisms. Ecology Letters 4: 277–287.
487 BUCHMANN, S.L. 1985. Bees use vibration to aid pollen collection from non-poricidal flowers.
488 Journal of the Kansas Entomological Society 58: 517–525.
489 BURD, M. 1994. Bateman’s principle and plant reproduction: the role of pollen limiation and fruit
490 and seed set. The Botanical Review 60: 83–139.
491 CAMPBELL, D.R. 1989. Inflorescence size: test of the male function hypothesis. American
492 Journal of Botany 76: 730–738.
493 CAMPBELL, D.R., and N.M. WASER. 1989. Variation in pollen flow within and among
494 populations of Ipomopsis aggregata. Evolution 43(7): 1444–1455.
495 CARPER, A.L., L.S. ADLER, and R.E. IRWIN. 2016. Effects of florivory on plant-pollinator
496 interactions: implications for male and female components of plant reproduction. American
497 Journal of Botany 103: 1061–1070.
498 CLARKE, D., E. MORLEY, and D. ROBERT. 2017. The bee, the flower, and the electric field:
499 electric ecology and aerial electroreception. Journal of Comparative Physiology A: in press.
500 CONTRERAS, P.S., and J.F. ORNELAS. 1999. Reproductive conflicts of Palicourea padifolia
501 (Rubiaceae), a distylous shrub of a tropical cloud forest in Mexico. Plant Systematics and
502 Evolution 219: 225–241.
Richman et al. 22 503 DUDASH, M.R., C. HASSLER, P.M. STEVENS, and C.B. FENSTER. 2011. Experimental floral and
504 inforescence trait manipulations affect hummingbird preference and function in a
505 hummingbird-pollinated plant. American Journal of Botany 92: 275–282.
506 GALEN, C., and M.L. STANTON. 1989. Bumble bee pollination and floral morphology: factors
507 influencing pollen dispersal in the Alpine Sky Pilot, Polemonium viscosum
508 (Polemoniaceae). American Journal of Botany 76: 419–426.
509 GARRISON, J.S.E., and C.L. GASS. 1999. Response of a traplining hummingbird to changes in
510 nectar availability. Behavioral Ecology 10: 714–725.
511 GONZALEZ-GOMEZ, P.L., and R.A. VASQEUZ. 2011. Flexibility of foraging behavior in
512 hummingbirds: the role of energy constraints and cognitive abilities. The Auk 128: 36–42.
513 HAINSWORTH, F.R., L.L. WOLF, and T. MERCIER. 1985. Pollen limitation in a monocarpic species
514 Ipomopsis aggregata. Journal of Ecology 73: 263–270.
515 HURLY, T.A., and S.D. HEALY. 2002. Cue learning by Rufous hummingbirds (Selasphorus
516 rufus). Journal of Experimental Psychology 28: 209–223.
517 INOUYE, D.W. 1980. The terminology of floral larceny. Ecology 61: 1251–1253.
518 INOUYE, D.W., D.E. GILL, M.R. DUDASH, and C.B. FENSTER. 1994. A model and lexicon for
519 pollen fate. American Journal of Botany 81: 1517–1530.
520 IRWIN, R.E. 2000. Hummingbird avoidance of nectar-robbed plants: spatial location or visual
521 cues. Oikos 91: 499–506.
522 IRWIN, R.E., and A.K. BRODY. 2000. Consequence of nectar robbing for realized male function
523 in a hummingbird-pollinated plant. Ecology 81: 2637–2643.
524 IRWIN, R.E., and A.K. BRODY. 1999. Nectar-robbing bumble bees reduce the fitness of Ipomopsis
525 aggregata (Polemoniaceae). Ecology 80: 1703–1712.
Richman et al. 23 526 IRWIN, R.E., and A.K. BRODY. 1998. Nectar robbing in Ipomopsis aggregata: effects on
527 pollinator behavior and plant fitness. Oecologia 116: 519–527.
528 IRWIN, R.E., J.L. BRONSTEIN, J.S. MANSON, and L.L. RICHARDSON. 2010. Nectar Robbing:
529 Ecological and Evolutionary Perspectives. Annual Review of Ecology, Evolution, and
530 Systematics 41:271–292.
531 IRWIN, R.E., P. HOWELL, and C. GALEN. 2015. Quantifying direct vs. indirect effects of nectar
532 robbers on male and female components of plant fitness. Journal of Ecology 103: 1487–
533 1497.
534 KEARNS, C.A., and D.W. INOUYE. 1993. Techniques for pollination biologists. University Press
535 of Colorado, Boulder, Colorado, USA.
536 LARA, C., and J.F. ORNELAS. 2002. Effects of nectar theft by flower mites on hummingbird
537 behavior and the reproductive success of their host plant, Moussonia deppeana
538 (Gesneriaceae). Oikos 96: 470–480.
539 LEVIN, D., and H. KERSTER. 1974. Gene flow in seed plants. Evolutionary Biology 7: 139–220.
540 MADJIDIAN, J.A. 2009. Sexual conflict and sexually antagonistic coevolution in an annual plant.
541 PLoS ONE 4: e5447.
542 MAYFIELD, M.M., N.M. WASER, and M. V PRICE. 2001. Exploring the “most effective pollinator
543 principle” with complex flowers: bumblebees and Ipomopsis aggregata. Annals of Botany
544 88: 591–596.
545 MORRIS, W.F., R.A. HUFBAUER, A.A. AGRAWAL, J.D. BEVER, V.A. BOROWICZ, G.S. GILBERT,
546 J.L. MARON, ET AL. 2007. Direct and interactive effects of enemies and mutualists on plant
547 performance: a meta-analysis. Ecology 88: 1021–1029.
548 MUTIKAINEN, P., and L.F. DELPH. 1996. Effects of herbivory on male reproductive success in
Richman et al. 24 549 plants. Oikos 75: 353–358.
550 NUNN C.L., C. BREZINE, A.E. JOLLES, and V.O. EZENWA 2014. Interactions between micro- and
551 macroparasites predict microparasite species richness across primates. American Naturalist
552 183: 494-505.
553 PLEASANTS, J.M. 1983. Nectar production patterns in Ipomopsis aggregata (Polemoniaceae).
554 American Journal of Botany 70: 1468.
555 R CORE TEAM. 2016. R: A languange and environment for statisical computing. Vienna, Austria.
556 RICHMAN, S.K., R.E. IRWIN, C.J. NELSON, and J.L. BRONSTEIN. 2017. Facilitated exploitation of
557 pollination mutualisms: fitness consequences for plants. Journal of Ecology 105: 188–196.
558 ROUX, O., R. CEREGHINO, P.J. SOLANO, and A. DEJEAN. 2011. Caterpillars and fungal pathogens:
559 two co-occurring parasites of an ant-plant mutualism. PLoS ONE 6(5): e20538.
560 SCHAEFFER, R.N., J.S. MANSON, and R.E. IRWIN. 2013. Effects of abiotic factors and species
561 interactions on estimates of male plant function: a meta-analysis. Ecology Letters 16: 399–
562 408.
563 STANTON, M.L., A.A. SNOW, and S.N. HANDEL. 1986. Floral evolution: attractiveness to
564 pollinators increases male fitness. Science 232: 1625–1627.
565 STRAUSS, S.Y., and R.E. IRWIN. 2004. Ecological and evolutionary consequences of multispecies
566 plant-animal interactions. Annual Review of Ecology, Evolution and Systematics 35: 435–
567 466.
568 TORRES, I., L. SALINAS, C. LARA, and C. CASTILLO-GUEVARA. 2008. Antagonists and their
569 effects in a hummingbird-plant interaction: field experiments. Ecoscience 15: 65–72.
570 VAKNIN, Y., A. BECHAR, B. RONEN, and D. EISIKOWITCH. 1999. The role of electrostatic forces
571 in pollination. Plant Systematics and Evolution 222: 133–142.
Richman et al. 25 572 WASER, N.M. 1978. Interspecific pollen transfer and competition between co-occurring plant
573 sepcies. Oecologia 36: 223–236.
574 WASER, N.M., and M. V PRICE. 1982. A comparison of pollen and fluorescent dye carry-over by
575 natural pollinators of Ipomopsis aggregata (Polemoniaceae). Ecology 63: 1168–1172.
576 WASER, N.M., and M. V PRICE. 1984. Experimental studies of pollen carryover: effects of floral
577 variability in Ipomopsis aggregata. Oecologia 62: 262–268.
578 WILSON, P.W., J.D. THOMSON, M.L. STANTON, and L.P. RIGNEY. 1994. Beyond floral
579 Batemania: gender biases in selection for pollination success. The American Naturalist 143:
580 283–296.
581 WOLF, L.L., F.R. HAINSWORTH, and F.G. STILES. 1972. Energetics of foraging: rate and
582 efficiency of nectar extraction by hummingbirds. Science 176: 1351–1352.
583
584
585
586
587
588
589
590
591
592
593
594
Richman et al. 26 595 Figure Legends
596 Figure 1. The effect of robbing treatments on pollen loss due to simulated secondary robber
597 landings on Ipomopsis flowers. Bars represent mean (±SE) number of pollen grains knocked off
598 of anthers, resulting in (A) grains landing on the inside of the corolla and (B) leaving the flower
599 through the floral opening. Grains leaving the flower through the floral opening where not
600 sampled from the Control treatment, because Control flowers were not manipulated to mimic
601 nectar-robber landing. N.S. represents no significantly different (P greater than 0.05) mean values
602 based on a Likelihood Ratio Test.
603
604 Figure 2. The effect of nectar-robbing treatments on (A) mean (±SE) hummingbird visitation rate
605 and (B) mean (±SE) number of pollen grains (dye) donated per flower dyed. Hummingbird
606 visitation rate is calculated as the number of times plants were visited multiplied by the mean
607 percentage of flowers probed. Different letters represent significantly different (P less than 0.05)
608 mean values based on a Tukey HSD test.
609
610 Figure 3. The effect of simulated secondary nectar robbing and chronic nectar removal on (A)
611 hummingbird visitation rate and (B) the proportion of flowers per plant probed by hummingbirds
612 per day. Bars and error bars represent means ± SEs. Hummingbird visitation rate is calculated as
613 in Figure 2. Different letters represent significantly different (P less than 0.05) mean values
614 based on a Tukey HSD test.
Richman et al. 27 Figure Click here to download Figure A) Richman_AJB_Ipomopsis_fig1.pdf
600 N.S.
400
200 Pollen Grains Lost
0 Control Primary Primary + Secondary Robbing Robbing
Treatment
B)
15 N.S.
10
5 Pollen Grains Lost
0 Primary Primary + Secondary Robbing Robbing
Treatment Figure Click here to download Figure A) Richman_AJB_Ipomopsis_fig2.pdf
a 1.0 b b
0.5
0.0 Hummingbird Visitation Rate Control Primary Primary + Secondary Robbing Robbing
Treatment
B)
0.4 a 0.3
0.2 b b
0.1
Dye Donation/Flower Dyed 0.0 Control Primary Primary + Secondary Robbing Robbing
Treatment Figure Click here to download Figure A) Richman_AJB_Ipomopsis_fig3.pdf
0.5 a 0.4
0.3 b b
0.2
0.1
0.0 Hummingbird Visitation Rate Control Primary + Secondary Nectar Robbing Removal
Treatment
B)
a 0.6
a,b b 0.4
0.2
Proportion Flowers Probed 0.0 Control Primary + Secondary Nectar Robbing Removal
Treatment