Flower colour segregation and flower discrimination under the bee vision model in the polymorphic Lysimachia arvensis

Flower colour segregation and flower discrimination under the bee vision model in the 1 polymorphic Lysimachia arvensis 2 3 Jiménez-López F.J.*, Matas L.*, Arista M. & Ortiz P.L. 4 Departamento de Biología Vegetal y Ecología. Universidad de Sevilla. Apdo. 41080. Sevilla, 5 Spain 6 7 *, these two authors contributed equally to this manuscript 8 Corresponding autor: F.J. Jiménez-López. Departamento de Biología Vegetal y Ecología. 9 Universidad de Sevilla. Apdo. 41080. Sevilla, Spain 10 e-mail: fjimenez[email protected] 11 12 13 Running head: Flower colour heritability in L. arvensis 14 15 16 Abstract 17 Floral colour determines pollinator behaviour, strongly affecting plant-mating systems. 18 Lysimachia arvensis has blue- and red-flowered plants and colour inheritance remains largely 19 unknown. A control of floral colour based on one locus, with the red allele as dominant, has 20 been proposed. This proposal cannot explain the sporadic appearance of other floral colours in 21 wild populations. We studied floral colour segregation in L. arvensis and assessed the 22 possibility that pollinators can visually distinguish colour morphs by using Chittka’s hexagon 23 model, sigmoidal model of bee discrimination and experimental studies on pollinator 24 attendance for two years. Hand crossing between morphs originated a homogeneous F1 with 25 salmon-coloured flowers. In the F2, blue, red, salmon morphs and other plants with 26 intermediate colours appeared, suggesting that more than one single locus are involved in 27 colour segregation. Results from the sigmoidal discrimination model suggest that blue, red 28 and salmon flowers can be discriminated by pollinators. In fact, pollinators showed strong 29 colour constancy and discriminated against the salmon morph. Our study shows that "Flower 30 colour" is a natural marker to assess the rate of crossing between morphs. The extreme rarity 31 of salmon flowers in wild populations and flower constancy of L. arvensis pollinators 32 indicates assortative mating. 33 34 Key words: Anagallis, Chittka hexagon, floral evolution, flower colour discrimination, 35 pollinator preference 36 37 INTRODUCTION 38 Angiosperms exhibit a markedly high diversity of flower colours, with sister species usually 39 differing in intensity, hue or colour pattern of the corolla (e.g. Rausher 2008; Smith and 40 Rausher 2011; Lagomarsino et al. 2017). This diversity implies that there have been 41 numerous evolutionary transitions in the colour of flowers (Weis 1995; Rausher 2008). 42 Flower colour is often correlated with other floral traits, resulting in the common recognition 43 of "pollination syndromes" (Fenster et al. 2004). Flower colour has an enormous importance 44 as a claim in the attraction of pollinators that may have preferences for some colours over 45 others (Chittka and Menzel 1992), so that transitions to different colours may represent 46 adaptation to different sets of pollinators (Faegri and van der Pijl 1966; Grant 1993; Fenster et 47 al. 2004; Rausher 2008). In general, the colour of the flowers is due to the presence of 48 pigments (Kay et al. 1981; Van der koi et al. 2016). There are four large groups of pigments: 49 chlorophylls, carotenoids, betalains and flavonoids. Among these, anthocyanins, a group of 50 flavonoids, are the most important floral pigments and are produced in a well-known and 51 conserved biosynthetic pathway in angiosperms (Rausher et al. 1999). 52 Flower colour polymorphism is the presence of more than one colour morph, genetically 53 determined, within the populations of a species (Huxley 1955). This phenomenon appears by 54 a spontaneous mutation in the biosynthetic route of the pigments that give colour to the 55 flowers. Once a coloured mutant appears in a population, this recent polymorphism can be 56 lost or maintained depending on biotic or abiotic selective factors and gene drift (Narbona et 57 al. 2018). Pollinator preferences play a fundamental role in the maintenance or loss of the 58 flower colour polymorphism. Pollinators can show innate preferences for some colours over 59 others (Shrestha et al. 2013, 2016; Van der Kooi et al. 2018) causing directional selection on a 60 determinate colour morph and leading to the loss of polymorphism (Waser and Price 1981). 61 However, balancing selection imposed by pollinators can result in the maintenance of 62 polymorphism. Thus, in the rewardless Dactylorhiza sambucina (L.) Soó, pollinators visit 63 different colour morphs in alternation as they switch to a different morph when they visit an 64 empty flower, thus maintaining colour polymorphism (Gigord et al. 2001). Similarly, if 65 species are visited by a wide variety of pollinators, they can show preferences for different 66 colour morphs thereby maintaining colour polymorphism (Schemske and Bradshaw 1999). 67 The behaviour of pollinators can induce changes in plant fertility, cross-pollination ratios, and 68 pollen flow among colour morphs (Malerba and Nattero 2012). These changes can lead to the 69 genetic differentiation of individuals with different flower colour, promoting ultimately the 70 speciation processes (Servedio et al. 2011). However, in order for pollinators to discriminate 71 between floral colours and act as selection agents, it is imperative they can differentiate them 72 visually. Therefore, a subjective evaluation of the floral colours according to the human 73 vision can lead to misleading interpretations in relation to the behaviour of pollinators, being 74 necessary an objective measurement of colours and their evaluation according to the visual 75 system of pollinators. 76 Lysimachia arvensis (L.) U. Manns & Anderb. is a tetraploid annual species, native to the 77 Mediterranean Basin and Europe that presents flower colour polymorphism. In natural 78 populations, there are plants with blue and red flowers, and these colours are due to the 79 presence of different types of anthocyanins. Malvidin is mainly responsible for the blue 80 colour and pelargonidin for the red colour (Wiering and de Vlaming in Harborne 1968; 81 Ishikura 1981). Selective abiotic factors influence a geographic distribution pattern of colour 82 morphs, with blue being much better represented in more xeric environments (Arista et al. 83 2013). In addition, in Mediterranean environments, pollinators show a higher preference for 84 the blue morph and the red morph has lower fitness; despite this, it remains in the populations 85 although in a low proportion (Ortiz et al. 2015). 86 The inheritance of flower colour in L. arvensis is unknown, and unravelling it could help to 87 understand the maintenance of the red morph in Mediterranean populations, despite being 88 subject to negative selection (Arista et al. 2013). In a simple scenario, if a recessive allele 89 were responsible for the red colour, it would be protected in the heterozygotes that would 90 show the blue dominant phenotype. However, in an oral communication in a Congress in 91 1910, Weiss explained that in experimental crosses between plants with flowers of different 92 colour, the F1 obtained was all homogeneously red. Therefore, he concluded that the flower 93 colour in L. arvensis depended on a single gene with two alleles, being the red allele 94 dominant over the blue. The fact that flowers of intermediate colour do not usually appear in 95 natural populations would support this dominance-recessive relationship between the two 96 alleles. Later, Marsden-Jones & Weiss (1938) confirmed that result, although in some 97 populations they found some plants of L. arvensis with flowers of salmon colour and others of 98 pale blue colour. Salmon-flowered plants, although rare, had also been described previously 99 by other authors who had suggested a hybrid origin between blue and red morphs since they 100 only appeared when the two morphs, blue and red, coexist (Hoffmann 1879; Pax 1905). 101 However, Marsden-Jones & Weiss (1938) found these plants in monomorphic red 102 populations, and thus they attributed salmon plants to spontaneous mutations. In a recent 103 sampling, over 19 mixed populations of Lysimachia arvensis in Western Europe, salmon- 104 flowered plants appeared in two of them (Jiménez-López et al., unpub results). Their scarce 105 representation in populations makes it difficult to know if they result from spontaneous 106 mutations or by crossing between the red and the blue morphs. In the latter case, only a low 107 frequency of crossing between morphs or a low success of the progeny of that crossing would 108 explain the almost absence of salmon-flowered individuals in mixed natural populations. 109 110 The objectives of the present work are: (1) to establish if the salmon morph results from the 111 crossing between the blue and the red morph in Lysimachia arvensis, (2) to know how the 112 flower colour is inherited, (3) to characterize quantitatively the flower colours that can appear 113 in this species by using the model of colour vision of Chittka (1992) and (4) to determine if 114 they can be differentiated by bees in two ways, by calculating their discrimination 115 probabilities from the sigmoidal-shaped model by Garcia et al. (2017), and by studying 116 pollinator attendance in experimental stands during two reproductive cycles. 117 118 MATERIAL AND METHODS 119 120 Heritability of flower colour 121 To study the inheritance of colour in Lysimachia arvensis, hand pollinations were carried out 122 in the greenhouse. Flower colour segregation was first quantified in offspring based on human 123 vision. The plants used originally came from seeds obtained in natural populations of Hinojos 124 (Spain), Tanger (Morocco), Tabarka (Tunisia) and Corsica (France). These plants were grown 125 in a greenhouse, and by manual self-pollinations two successive generations were obtained to 126 select pure colour lines. These pure lines, blue (B) and red (R), were used as parental (P) in 127 this study. Crossings were carried out between parents of the same colour and different colour 128 in order to obtain the F1. Crosses between parents of different colours were carried out in 129 both directions, that is, the blue plants as pollen donors and the red plants as pollen receiver 130 (RxB, n = 84 crosses) and the red plants as pollen donors and the blue as receiver (BxR, n = 131 88). The F1 seeds obtained were put to germinate in Petri dishes in germination chambers 132 under 16h of light at 22°C and 8h of darkness at 15°C and seedlings were grown in the 133 greenhouse. In this F1, different types of pollinations were made to obtain the F2. Some F1 134 plants were self-pollinated (N=149 pollinations), others were crossed with each other (N = 43 135 crosses), others were backcrossed with blue parental (N = 34 crosses), and others with red 136 parental (N = 39 crosses). All seeds produced by this F1 were germinated and the resulting 137 seedlings were grown in greenhouses until flowering (2907 plants). 138 139 Flower colour characterization 140 To characterize quantitatively floral colours of L. arvensis plants obtained from the crossing 141 program previously described, the reflectance spectra of the petals of a subsample of plants 142 were measured. The reflectance was measured in 88 parental plants (44 B and 44 R), 38 F1 143 plants (S thereafter; 15 from BxR and 23 from RxB) and 41 F2 plants obtained from self- 144 pollination of the F1. Reflectance was also measured in 53 plants from the F1 backcrosses 145 with both parents (19 from SxB, 5 from BxS, 14 from SxR and 15 from RxS). In each plant, 146 the reflectance of the adaxial surface of a petal was measured, discarding the basal part 147 corresponding to the centre of the flower (bull’s-eye). To do that, a JAZ A1465 double-beam 148 spectrophotometer from Ocean Optics, equipped with a UV-visible light source and capable 149 of measuring reflectance between 190 and 890 nm was used. Reflectance spectra of the 150 measured flowers are deposited at the open repository of the Universidad de Sevilla 151 (https://idus.us.es/xmlui/). 152 153 Model of flower colour vision 154 To assess how petals are perceived by bees, the reflectance values between 300 and 700 nm 155 obtained in each measurement were elaborated and represented in the colour hexagon model. 156 This model was developed by Chittka (1992) integrating experimental data related to the 157 reception of visual signals by bees and the translation of these signals in the bee brain. The 158 colour hexagon is a two-dimensional representation in which each reflectance spectrum 159 corresponds to a point defined by its Cartesian coordinates; a detailed description of how to 160 transfer the reflectance data to the colour hexagon can be seen in Chittka & Kevan (2005). 161 This model allows quantifying the contrast of a flower with the general green background as 162 the Euclidean distance between the point generated by the flower spectrum and the centre of 163 the hexagon; in addition, it allows the categorization in a conventional manner of the colours 164 perceived by bees placing them in six colour categories (Chittka 1992). 165 The Chittka model also allows quantifying the colour contrast of colour between two flowers 166 perceived by the bees as the Euclidean distance in the hexagon between the points generated 167 by their colour spectra, 0.1 being the threshold value for colour discrimination. However, 168 recent behavioural studies modelled by particular bee species have reported that colour 169 discrimination depends on context (Dyer and Chittka 2004; Dyer 2006) and follow sigmoidal- 170 shaped functions (Garcia et al. 2017, 2018). To assess the capacity of bees for discrimination 171 between both parental morphs (blue and red) and both F1-hybrid types (BxR and RxB), 172 Euclidean distances were calculated for all possible pairs of flowers between twelve flowers 173 of each of those four classes (blue, red, BxR and RxB). From those data, the discrimination 174 capacity by bees for those pairs of flowers were calculated by using the 3-parameter logistic 175 function described by Garcia et al. (2017). Given that the main pollinators of L. arvensis are 176 Apoideae species and its flowers have blue anthocyanins, we selected the models for blue 177 stimuli for both Apis mellifera L. and Bombus terrestris L. We used the median values of K, r 178 and Mo parameters for those models from S-4 supplementary material from Garcia et al. 179 (2017). 180 181 Pollinator preferences on colour morphs 182 To ascertain pollinator preferences on colour morphs, and so their discrimination capacity, we 183 recorded pollinator visitation to parental and F1 hybrid plants during two reproductive 184 seasons. We constructed artificial stands with a similar number of flowers of each of the three 185 colours that where intermingled. Each stand occupied an area of 0.5 m2 and insect visitations 186 were recorded by observing each stand for 10-min periods. All observations were made 187 during sunny conditions between 9:00 and 15:00h to totalize 9 hours of censuses per year. In 188 each census the number of flowers of each morph visited and the transitions between colour 189 morphs made by pollinators were recorded. Differences in the number of visits per morph and 190 census were analysed by means of a GLM model with Poisson distribution of errors and log 191 link function with morph colour and year as main factors and considering their interaction. 192 Differences in the frequency of transitions made by pollinators among colour morphs each 193 year were analysed by pooling together data from all the censuses and using chi-square tests 194 of frequencies. 195 RESULTS 196 Heritability of flower colour 197 All offspring obtained from crosses BxB and RxR was homogeneous and showed the same 198 colour as the parents (N = 350 individuals observed in each case), which confirms the purity 199 of the blue and red lines selected as parental. The crosses between plants of different colour, 200 BxR and RxB, also originated a homogeneous offspring salmon in colour (N = 1199 201 individuals analysed, Fig. 1). In addition, these individuals presented a bull’s-eye (ring of 202 colour at the base of the petals) similar in size to that of the blue morph but larger than that of 203 the red morph (Fig. 1). The self-pollination of the F1 originated 707 blue plants, 926 red and 204 452 salmon, but also appeared 51 individuals with intermediate colours between red and 205 salmon (Fig. 1). The backcrosses of the F1 with each of the parents also gave rise to these 206 four phenotypes, but in different proportions. When the backcross was performed with the 207 blue parent, offspring showed mainly blue flowers (n = 407 plants), whereas when it was 208 carried out with the red parent offspring was predominantly red (n = 496). 209 210 Chittka L. 1992. The colour hexagon: a chromaticity diagram based on photoreceptor 361 excitations as a generalized representation of colour opponency. J Comp Physiol A: 362 Neuroethol Sens Neural & Behav Physiol. 170: 533-543. 363 Chittka L, Kevan PG. 2005. Flower colour as advertisement. In Dafni A, Kevan P, Husband 364 BC. (Eds.). Practical Pollination Biology. Enviroquest Ltd.: Canada. 157-196. 365 Chittka L, Menzel R. 1992. The evolutionary adaptation of flower colours and the insect 366 pollinators' colour vision. J Comp Physiol A: Neuroethol Sens Neural & Behav Physiol. 367 171: 171-181. 368 Chittka L, Spathe J, Schmidt A, Hickelsberger A. 2001. Adaptation, constraint and chance in 369 the evolution of flower colour and pollinator colour vision. In Thomson JD. (Ed.). 370 Cognitive Ecology of Pollination. Cambridge University Press: Cambridge (UK). 106– 371 126. 372 Dyer AG. 2006. Discrimination of flower colours in natural settings by the bumblebee species 373 Bombus terrestris (Hymenoptera: Apidae). Entomol Gen. 28: 257–268. 374 Dyer AG, Chittka L. 2004. Biological significance of distinguishing between similar colours 375 in spectrally variable illumination: Bumblebees (Bombus terrestris) as a case study. J 376 Comp Physiol A. 190:105–114. 377 Faegri K, van der Pijl L. 1966. The principles of pollination ecology. Pergamon Press. 378 London (UK). 379 Fenster CB, Armbruster WS, Wilson P, Dudash MR, Thomson DJ. 2004. Pollination 380 syndromes and floral specialization. Ann Rev Ecol Evol Syst. 35: 375–403. 381 Freyre R, Griesbach RJ. 2004. Inheritance of flower color in Anagallis monelli L. 382 HortScience 39: 1220-1223. 383 Garcia JE, Spaethe J, Dyer AG. 2017. The path to colour discrimination is S-shaped: 384 behaviour determines the interpretation of colour models. J Comp Physiol A. 203: 983- 385 997. 386 Garcia JE, Shrestha M, Dyer AG. 2018. Flower signal variability overwhelms receptor-noise 387 and requires plastic color learning in bees. Behav Ecol. 29: 1286-1297. 388 Gettys LA. 2012. Genetic Control of White Flower Color in Scarlet Rosemallow (Hibiscus 389 coccineus Walter). J Hered.103: 594-597. 390 Gigord LD, Macnair MR, Smithson A. 2001. Negative frequency-dependent selection 391 maintains a dramatic flower color polymorphism in the rewardless orchid Dactylorhiza 392 sambucina (L.) Soo. Proc Natl Acad Sci. 98: 6253-6255. 393 Grant V. 1993. Effects of hybridization and selection on floral isolation. Proc Natl Acad Sci. 394 90: 990-993. 395 Harborne JB. 1968. Comparative biochemistry of the flavonoids—VII: Correlations between 396 flavonoid pigmentation and systematics in the family Primulaceae. Phytochem. 7: 1215- 397 1230. 398 Hoffmann HKH. 1879. Nachträge zur Flora des Mittelrhein-Gebietes. Schmitz. 399 Hopkins R, Rausher MD. 2011. Identification of two genes causing reinforcement in the 400 Texas wildflower Phlox drummondii. Nat. 469: 411. 401 Huxley JS. 1955. Morphism and evolution. Hered. 9: 1-51. 402 Ishiguro K, Taniguchi M, Tanaka Y. 2012. Functional analysis of Antirrhinum kelloggii 403 flavonoid 3′-hydroxylase and flavonoid 3′, 5′-hydroxylase genes; critical role in flower 404 color and evolution in the genus Antirrhinum. J Plant Res. 125: 451-456. 405 Ishikura N. 1981. Flavonoids in the petal cells of Anagallis arvensis f. coerulea containing a 406 blue crystalline anthocyanin. Z für Pflanzenphysiol. 103: 469-473. 407 Kay QON, Daoud HS, Stirton CH. 1981. Pigment distribution, light reflection and cell 408 structure in petals. Bot J Linn Soc. 83: 57–83. 409 Lagomarsino LP, Forrestel EJ, Muchhala N, Davis CC. 2017. Repeated evolution of 410 vertebrate pollination syndromes in a recently diverged Andean plant clade. Evol. 71: 411 1970-1985. 412 Malerba R, Nattero J. 2012. Pollinator response to flower color polymorphism and floral 413 display in a plant with a single-locus floral color polymorphism: consequences for plant 414 reproduction. Ecol Res. 27: 377-385. 415 Marsden‐Jones EM, Weiss FE. 1938. The essential differences between Anagallis arvensis L. 416 and Anagallis foemina Mill. Proc Linn Soc Lond. 150: 146-155. 417 Narbona E, Wang H, Ortiz PL, Arista M, Imbert E. 2018. Flower colour polymorphism in the 418 Mediterranean Basin: occurrence, maintenance and implications for speciation. Plant Biol. 419 20: 8-20. 420 Ortiz PL, Berjano R, Talavera M, Rodríguez-Zayas L, Arista M. 2015. Flower colour 421 polymorphism in Lysimachia arvensis: How is the red morph maintained in Mediterranean 422 environments? Perspect Plant Ecol Evol Syst. 17: 142-150. 423 Pax F, 1905. Primulaceae/Primula. In Engler A. (ed.). Das Pflanzenreich, Band IV/237, H. R. 424 Engelmann, Leipzig, Germany. 1–160. 425 Rausher MD, Miller RE, Tiffin P. 1999. Patterns of evolutionary rate variation among genes 426 of the anthocyanin biosynthetic pathway. Mol Biol Evol. 16: 266-274. 427 Rausher MD. 2008. Evolutionary transitions in floral color. Int J Plant Sci. 169: 7-21. 428 Schemske DW, Bradshaw HD. 1999. Pollinator preference and the evolution of floral traits in 429 monkeyflowers (Mimulus). Proc Natl Acad Sci. 96: 11910-11915. 430 Servedio MR, Van Doorn GS, Kopp M, Frame AM, Nosil P. 2011. Magic traits in speciation: 431 ‘magic’ but not rare? Trends Ecol Evol. 26: 389-397. 432 Shrestha M, Dyer AG, Boyd‐Gerny S, Wong BB, Burd M. 2013. Shades of red: bird‐ 433 pollinated flowers target the specific colour discrimination abilities of avian vision. N 434 Phytol. 198: 301-310. 435 Shrestha M, Lunau K, Dorin A, Schulze B, Bischoff M, Burd M, Dyer AG. 2016. Floral 436 colours in a world without birds and bees: the plants of Macquarie Island. Plant Biol. 18: 437 842-850. 438 Smith SD, Rausher MD. 2011. Gene loss and parallel evolution contribute to species 439 difference in flower color. Mol Biol Evol. 28: 2799-2810. 440 van der Kooi CJ, Elzenga JTM, Staal M, Stavenga DG. 2016. How to colour a flower: on the 441 optical principles of flower coloration. Proc R Soc B. 283: 20160429. 442 van der Kooi CJ, Dyer AG, Kevan PG, Lunau K. 2018. Functional significance of the optical 443 properties of flowers for visual signalling. Ann Bot. 123: 263-276. 444 Waser NM, Price MV. 1981. Pollinator choice and stabilizing selection for flower color in 445 Delphinium nelsonii. Evol. 35: 376-390. 446 Weiss MR. 1995. Floral color change: a widespread functional convergence. Am J Bot. 167- 447 185. 448 Wessinger CA, Rausher MD. 2014. Predictability and irreversibility of genetic changes 449 associated with flower color evolution in Penstemon barbatus. Evol. 68: 1058-1070. 450 Zufall RA, Rausher MD. 2004. Genetic changes associated with floral adaptation restrict 451 future evolutionary potential. Nat. 428: 847–850. 452 453 Captions of figure 454 Figure 1. Flowers of Lysimachia arvensis. A: blue (top left), red (top right) and salmon F1 455 (bottom) morphs. B: sample of individuals resulting from the self-pollination of F1 or from 456 the backcrosses of F1 with its parents. 457 458 Figure 2. Reflectance spectra of the flower color of Lysimachia arvensis. A: blue and red 459 morphs. B: F1 resulting from the cross between red and blue morphs (grey BxR, black RxB). 460 C, D: offspring resulting from the backcross between the F1 and the blue morph acting as 461 pollen receiver (C) or pollen donnor (D). E, F: offspring resulting from the backcross between 462 the F1 and the red morph acting as pollen receiver (E) or pollen donnor (F). G: F2 offspring 463 resulting from self-pollination of F1. Means and standard deviations are shown. In panels A, 464 C, D, E, F and G, grey lines correspond to reddish flowers under human vision and black lines 465 to bluish flowers under human vision. 466 467 Figure 3. Representation of the flower colour of Lysimachia arvensis in the hexagon model 468 proposed by Chittka (1992) based on the perception of color by bees. A: Blue and red 469 morphs. B: F1 resulting from the cross between blue and red morphs. C: backcross between 470 F1 and the blue morph. D: backcross between the F1 and the red morph. E: F2 offspring 471 resulting from self-pollination of F1. 472 473 Figure 4. Euclidean distances and discrimination probabilities between pairs of flowers of L. 474 arvensis. Median and range values are shown. Euclidean distances were calculated according 475 to hexagon model by Chittka (1992) and discrimination probabilities according to sigmoidal- 476 shaped functions by Garcia et al. (2017) (see text for details). 477 478 Figure 5. Pollinator preferences and transition between flowers in experimental stands during 479 two consecutive years. A: Mean number of pollinator visits per census at blue, red or salmon 480 flowers. B: Transition between flower colours made for pollinators each year. Each pie shows 481 the transitions from blue, red or salmon flowers to blue, red or salmon flowers each year. In 482 each pie, different letters indicate significant differences. 483 484 Figure 1 Figure 2 Figure 3