Interspecific pollen transport between non-native fennel and an island endemic buckwheat: assessment of the magnet effect

Abstract

Non-native plant species can disrupt plant–pollinator interactions by altering pollinator foraging behavior, which can in turn affect levels of interspecific pollen transfer between native and non-native plant species. These processes may be amplified in cases where introduced plant species act as magnet taxa that enhance pollinator visitation to other plant species. We investigated these interactions on Santa Cruz Island (Santa Barbara Co., California) between non-native fennel (Foeniculum vulgare), a widespread and abundant invader, and the endemic Santa Cruz Island buckwheat (Eriogonum arborescens), which broadly overlaps fennel in its local distribution and blooming phenology. A fennel flower removal experiment revealed that this invader acts as a magnet species by increasing insect visitation to adjacent buckwheat flowers. Analysis of the amount of pollen carried on the bodies of insect pollinators (i.e., pollen transport) revealed that 96% of visitors to buckwheat flowers carried fennel pollen and 72% of visitors to fennel flowers carried buckwheat pollen. Pollen transport analyses and visitation rate data further suggest that members of three bee genera (primarily Augochlorella) may be responsible for the majority of fennel pollen deposited on the stigmas of buckwheat flowers (i.e., pollen transfer) and vice versa. Lastly, fennel pollen transport appeared to occur at a larger spatial scale than the magnet effect that fennel plants exert on floral visitors to neighboring buckwheat plants. The ability of fennel to act as a magnet species, coupled with the fact that it is widespread invader with known allelopathic capacities, suggests that future studies could evaluate if the transfer of fennel pollen adversely affects native plant reproduction in areas where fennel is introduced.

Appendices

Appendix 1

GPS coordinates for paired control and removal plots on Santa Cruz Island.

Plot	LAT	LONG
2 Control	34.01118	− 119.68977
2 Removal	34.01122	− 119.69038
4 Control	34.01139	− 119.69183
4 Removal	34.01152	− 119.69225
5 Control	34.01099	− 119.69656
5 Removal	34.01115	− 119.69634
6 Control	34.01068	− 119.69737
6 Removal	34.01047	− 119.69750
7 Control	34.00974	− 119.69774
7 Removal	34.01019	− 119.69727
8 Control	33.99738	− 119.72720
8 Removal	33.99789	− 119.72679
9 Control	33.99704	− 119.72903
9 Removal	33.99762	− 119.72884
10 Control	33.99949	− 119.73544
10 Removal	33.99976	− 119.73565
11 Control	33.99918	− 119.73835
11 Removal	34.00029	− 119.73942
12 Control	34.01192	− 119.69320
12 Removal	34.01170	− 119.69447

Appendix 2

GPS coordinates for the locations where insects were collected for pollen transport analyses. (P = Prisoner’s, FS = Field Station, and C = Cabins).

Location	LAT	LONG
P-1	34.01054	− 119.69741
P-2	34.00830	− 119.69832
P-3	34.00817	− 119.69798
P-4	34.00691	− 119.70004
P-5	34.00567	− 119.70230
P-6	34.00556	− 119.70303
FS-1	33.99697	− 119.72897
FS-2	33.99755	− 119.72709
FS-3	33.99797	− 119.72761
FS-4	33.99765	− 119.72859
FS-5	33.99769	− 119.72916
FS-6	33.99770	− 119.72990
C-1	33.99961	− 119.73545
C-2	33.99966	− 119.73590
C-3	33.99910	− 119.73695
C-4	33.99893	− 119.73786
C-5	33.99950	− 119.73860
C-6	33.99958	− 119.73880

Appendix 3

Maps of fennel removal experimental plots and sites used to estimate pollen transfer potential in three locations: (a) Prisoner’s, (b) Field Station and (c) Cabins along the La Cañada wash on Santa Cruz Island. Removal experimental plots are denoted with blue squares (C = control plot, R = removal plot) and pollen transfer potential sites are denoted with red circles. (P = Prisoner’s, FS = Field Station, and C = Cabins).

Appendix 4

Statistical output of linear mixed-effects models examining impacts of fennel-removal treatment on insect visitation to buckwheat (dependent variable). Independent variables for both pre-removal and post-removal models included the treatment status (control versus removal) of the plot, the spatial extent of floral coverage on the observed buckwheat individual, and the total plot-level spatial extent of floral coverage (i.e., combining all focal and non-focal buckwheat individuals, as well as fennel when applicable). Plot identity was included as a random-intercept term to account for multiple sampling in the same plot. The model for post-removal surveys additionally included survey round (first, second, third) as an independent variable.

Model	Fixed Effects	t	P	Slope ± SE
All visitors, pre-removal:	Treatment	− 0.06	0.96	− 0.318 ± 5.809
	Plot floral area	− 0.48	0.64	− 0.404 ± 0.840
	Bush area	2.60	0.013	6.343 ± 2.438
All visitors, post-removal:	Treatment	− 3.43	0.0034	− 14.032 ± 4.091
	Plot floral area	− 2.66	0.012	− 1.466 ± 0.551
	Bush area	− 2.57	0.011	4.614 ± 1.794
	2nd survey	1.39	0.17	5.269 ± 3.780
	3rd survey	6.63	< 0.0001	29.568 ± 4.460
Bees, pre-removal:	Treatment	− 0.03	0.98	− 0.157 ± 5.657
	Plot floral area	− 0.06	0.95	− 0.049 ± 0.821
	Bush area	1.83	0.07	4.532 ± 2.484
Bees, post-removal:	Treatment	− 3.36	0.0038	− 13.656 ± 4.062
	Plot floral area	− 2.71	0.010	− 1.475 ± 0.544
	Bush area	2.16	0.032	3.786 ± 1.753
	2nd survey	1.34	0.18	4.926 ± 3.688
	3rd survey	6.75	< 0.0001	29.447 ± 4.360
LGH, pre-removal:	Treatment	− 0.22	0.83	− 0.941 ± 4.337
	Plot floral area	0.003	0.99	0.002 ± 0.626
	Bush area	1.21	0.23	2.152 ± 1.782
LGH, post-removal:	Treatment	− 1.55	0.14	− 7.040 ± 4.530
	Plot floral area	− 1.40	0.17	− 0.723 ± 0.517
	Bush area	1.02	0.31	1.350 ± 1.329
	2nd survey	0.89	0.37	2.444 ± 2.733
	3rd survey	3.29	0.0013	11.136 ± 3.389
Non-bees, pre-removal:	Treatment	− 0.17	0.87	− 0.165 ± 0.980
	Plot floral area	− 2.43	0.025	− 0.347 ± 0.143
	Bush area	3.69	0.00060	1.720 ± 0.467
Non-bees, post-removal:	Treatment	− 0.41	0.68	− 0.407 ± 0.996
	Plot floral area	− 0.03	0.98	− 0.004 ± 0.131
	Bush area	2.02	0.05	0.824 ± 0.408
	2nd survey	0.40	0.69	0.338 ± 0.856
	3rd survey	0.08	0.93	0.084 ± 1.018