Flower visitation of alien plants is non-linearly related to phylogenetic and floral similarity to native plants

Biological invasions are key to understanding major ecological processes that determine the formation of novel interactions. Flower visitation to alien species may be facilitated by co-flowering natives if they share similar floral traits with the latter. However, if competition for pollinators is important, flower visitation to alien species may be higher if they have traits different from those of native species. We tested whether flower visitation to alien plants depended on phylogenetic relatedness and floral similarity to native plants. In a field experiment, we simulated invaded communities by adding potted alien plants into co-flowering native communities. We recorded flower visitation to pairs of 34 alien and 20 native species totalling 151 species combinations for 3,068 hours. Flower visitation to alien species was highest when they had intermediate floral trait distances to native species, and either low or high phylogenetic distances. The alien plants received more similar flower-visitor groups to natives when they had low phylogenetic and either low or high floral trait distances to native plants. The non-linear patterns between flower visitation and similarity of the alien and native species suggest that an interplay of facilitation and competition simultaneously drives the formation of novel plant-pollinator interactions. The shapes of the relationships of phylogenetic and floral trait distances with flower visitation to alien plants were contrasting, possibly due to different strengths of phylogenetic signal among traits. We provide experimental evidence for the effects of relatedness and functional similarity to native plants on flower visitation of alien plants. We show that such effects might be non-linear, and that effects of trait dissimilarity and phylogenetic distance on pollinator-mediated interactions can reflect different mechanisms.

. Specifically, as many alien organisms have 62 integrated into native resident communities, biological invasions are key to understanding the 63 major ecological processes that determine the formation of novel interactions. 64 As related species are likely to be more similar, they should show strong niche overlap 65 and compete for shared resources. Based on this premise, Darwin (1859)  showed that similarities in flower symmetry and color in alien and native plants increased 107 competition for pollinators (Morales & Traveset, 2009). Therefore, similarity in these floral traits 108 may play critical roles in pollinator-mediated alien-native plant interactions.

109
The patterns of novel plant-pollinator interactions may differ with regard to phylogenetic 110 and floral trait distances. While phylogenetic relatedness is frequently assumed to be a proxy for 111 trait similarity, some floral traits may not be evolutionarily conserved (Sargent & Ackerly, 2008 (Diez et al., 2008). Furthermore, different mechanisms may act simultaneously to drive the 121 success of alien species, as has been shown for direct plant-plant interactions (Malecore et al.,    145 To simulate invaded communities in a field experiment in central Europe, we selected 34 146 herbaceous insect-pollinated neophytes (i.e. alien species introduced after the discovery of the between April and September. The alien species belonged to 14 plant families, 68% were short-150 lived (annual or biennial), and 24% were self-incompatible (Supporting Information Table S1).

151
The species neophyte status was based on information in the Floraweb  (Table S2). 158 Two to three days before adding alien plants to flowering resident native communities, we 159 prospected grasslands around the city of Konstanz, and identified sites that were visibly and for all flowering species ranged from 24 to 4,900 and from 10 to 1,763, respectively (Table   166 S2). To simulate invaded communities, we placed for up to ten alien species (range=4-10,  (Table S2), belonging to ten plant families. In

176
To record flower visitation, we placed a BRINNO TLC200 time-lapse camera 177 (https://www.brinno.com/time-lapse-camera/TLC200) at a vertical distance of 25-30 cm above 178 7 the flowers of the paired flowering alien and native plants (Fig. S1). We set the time-lapse 179 interval at 2 seconds and recorded from 10:30 to 16:30 on a sunny day (except for two set-ups in 180 which we recorded from 9:30 to 14:30 due to logistic constraints). Sampling was done on one day 181 for each setup. By using 40 cameras, we could observe many pairs of plants simultaneously. The 182 alien plants were removed from the experimental plots at the end of the recording day. We 183 collected a total of 3,068 hours of observations, which corresponded to 3 terabytes of video files. 184 We analyzed each video file manually using the Blender software (https://www.blender.org/).  (Table S3). 207 To test whether flower visitation to the alien species was influenced by relatedness and 208 floral similarity between the alien and the native species, we calculated phylogenetic and floral   production (absent=0, present=1). To account for potential variation of flower-visitor activity 241 during the day, we additionally included the time interval during the day (one-hour intervals as a 242 discrete variable) as an explanatory variable. To test for the effects of either the phylogenetic or 243 the floral trait distance between the alien and the native species, we also included them as 244 explanatory variables. All covariates were centered to means of zero and scaled to standard 245 deviations of one. To test for potential nonlinear effects, we included the quadratic terms for the 246 time interval during the day and phylogenetic or floral trait distance. As the latter were centered 247 and scaled, the non-linear effects test for hump-and U-shaped relationships. We also ran the 248 models with linear terms only, and we present the results of the model with the lowest AIC. To 249 account for non-independence of observations within species, we included identities of the added 250 alien species and the host native species as random factors. Models including site as an additional 251 random factor to account for potential variation due to site characteristics, such as floral 252 abundance, did not converge, as native species and site were largely confounded (Table S1).

253
Similarly, models including date of observation as a random factor to account for potential 254 variation due to change in flower visitor communities along the growing season did not converge.

255
To understand whether the resulting patterns were driven by a particular flower visitor group, we   (Tables 1 and 2).

288
Among the floral traits, flower size and symmetry had strong phylogenetic signals (Table 3). We 289 found that the phylogenetic distance between the alien and the native plant had significant 290 nonlinear effects on the total number of flower visits to the alien plant and on the proportion of 291 flower visits to the alien relative to the total number of visits to alien and native plants (Fig. 1a   292 and 1c, Table 1). Flower visitation to aliens was lowest when they had intermediate phylogenetic 293 distances to natives (Fig. 1a and 1c). The floral trait distance between the alien and the native 294 plant also had significant nonlinear effects on the total number of flower visits to the alien plant 295 and on the proportion of flower visits to the alien relative to the total number of visits to alien and 296 native plants ( Fig. 1b and 1d, Table 2). The alien plants with intermediate floral trait distances to 297 native plants received the most flower visits (Fig. 1b and 1d). We found qualitatively similar 298 results in the analysis of flower visitation by Hymenoptera only (Table S4), but partly different 299 results in the analysis of flower visitation by Diptera only (Table S5). We also found qualitatively 300 similar results in the analyses including both phylogenetic and floral trait distances (Table S6).

301
When single absolute floral trait distances were considered separately, we found that alien 302 plants received significantly more flower visits when flower symmetry, the presence of local 303 maxima in the green wavelength interval 471-540 nm of the reflectance spectra and nectar 304 production were dissimilar to the native plants (Table S7). Alien plants received significantly 305 fewer flower visits when flower size distance was larger and when the presence of local maxima 306 in the blue wavelength interval 401-470 nm was dissimilar to the native plants (Table S7). When 307 hierarchical floral trait distances were considered, we found that alien plants received 308 significantly more flower visits when native plants had local maxima in the blue wavelength 309 interval 401-470 nm of the flower reflectance spectra and the alien plants not (Table S8). Alien  (Table S8).  (Table 4).

317
The alien plants received significantly fewer flower visitors that were similar to those on native 318 plants when the number of observed flowers on the native plants increased (Table 4). We found 319 that the phylogenetic distance between the alien and the native plant had a significant negative 320 effect on the similarity between the flower visitor compositions of the alien and the native plants 321 (Fig. 1e, Table 4). The alien plants with high phylogenetic distances to native plants had the least 322 similar flower visitor composition to native plants. The floral trait distance between the alien and 323 the native plant had a significant nonlinear effect on the similarity between the flower visitor 324 compositions of the alien and the native plants (Fig. 1f, Table 4). The alien plants with to native plants (Fig. 1f).

328
In a field experiment simulating invaded co-flowering communities, we found that flower 329 visitation to alien species was highest when they had intermediate floral trait distances to native 330 species, but either low or high phylogenetic distances. This apparent discrepancy may be due to

362
As we did not quantify visitation to aliens in the absence of natives, we could not quantify 363 facilitation and competition directly. However, as the strength of facilitative and competitive 364 interactions is likely to depend on the dissimilarity of the species, the non-linear patterns we 365 found suggest an interplay of facilitation and competition (Gallien & Carboni, 2017).   (Table S7). However, our result on 395 dissimilarity in flower color was partly different from previous studies, as dissimilarity in 396 different components of floral reflectance was associated with either competition or facilitation.

397
For example, we found a negative relationship between dissimilarity in the blue wavelength 398 patterns of petals and flower visitation to alien plants (Table S7). This may be driven by the most 399 abundant flower visitors in our study, the bees (Hymenoptera), which frequently prefer the blue   (Table 3). Furthermore, single floral trait distances had different directions of effects, suggesting 413 facilitative, neutral or competitive effects (Table S7) Table 1 Results of a negative binomial generalized linear mixed model and a linear mixed model testing how the phylogenetic