Rapid local adaptation in both sexual and asexual invasive populations of monkeyflowers (Mimulus spp.)

Annals of Botany 127: 655–668, 2021 doi: 10.1093/aob/mcab004, available online at www.academic.oup.com/aob © The Author(s) 2021. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. Rapid local adaptation in both sexual and asexual invasive populations of monkeyflowers (Mimulusspp.) VioletaI.Simón-Porcar1,*, JoseL.Silva2 and MarioVallejo-Marín3 1Departmento de Biología Vegetal y Ecología, Universidad de Sevilla, Apartado 1095, E-41080 Sevilla, Spain, 2Pyrenean Institute of Ecology (CSIC), Avenida Montañana 1005, 50059 Zaragoza, Spain, and 3Biological and Environmental Sciences, Faculty of Natural Sciences, University of Stirling, Stirling, Scotland FK9 4LA, UK *For correspondence. E-mail [email protected] Received: 22 December 2020 Returned for revision: 16 December 2020 Editorial decision: 5 January 2021 Accepted: 7 January 2021 Electronically published: 19 February 2021 • Background and Aims Traditionally, local adaptation has been seen as the outcome of a long evolutionary history, particularly with regard to sexual lineages. By contrast, phenotypic plasticity has been thought to be most important during the initial stages of population establishment and in asexual species. We evaluated the roles of adaptive evolution and phenotypic plasticity in the invasive success of two closely related species of invasive monkeyflowers (Mimulus) in the UK that have contrasting reproductive strategies: M.guttatus combines sexual (seeds) and asexual (clonal growth) reproduction while M. × robertsii is entirely asexual. • Methods We compared the clonality (number of stolons), floral and vegetative phenotype, and phenotypic plasticity of native (M.guttatus) and invasive (M.guttatus and M. × robertsii) populations grown in controlled environment chambers under the environmental conditions at each latitudinal extreme of the UK. The goal was to discern the roles of temperature and photoperiod on the expression of phenotypic traits. Next, we tested the existence of local adaptation in the two species within the invasive range with a reciprocal transplant experiment at two field sites in the latitudinal extremes of the UK, and analysed which phenotypic traits underlie potential local fitness advantages in each species. • Key Results Populations of M.guttatus in the UK showed local adaptation through sexual function (fruit production), while M. × robertsii showed local adaptation via asexual function (stolon production). Phenotypic selection analyses revealed that different traits are associated with fitness in each species. Invasive and native populations of M.guttatus had similar phenotypic plasticity and clonality. M. × robertsii presents greater plasticity and clonality than native M.guttatus, but most populations have restricted clonality under the warm conditions of the south of theUK. • Conclusions This study provides experimental evidence of local adaptation in a strictly asexual invasive species with high clonality and phenotypic plasticity. This indicates that even asexual taxa can rapidly (<200years) adapt to novel environmental conditions in which alternative strategies may not ensure the persistence of populations. Key words: Asexual, introduced species, local adaptation, Mimulus guttatus, M. × robertsii, M.luteus, phenotypic plasticity, reciprocal transplants. INTRODUCTION Populations of broadly distributed species adapt to local conditions through genetic differentiation (Williams, 1966; Kawecki and Ebert, 2004; Hereford, 2009) and phenotypic plasticity (Bradshaw, 1965; Donohue, 2018). These two mechanisms are universal, interacting and non-mutually exclusive (Price etal., 2003; de Jong, 2005; West-Eberhard, 2005; Kelly, 2019). Yet, the traditional view was that local adaptation has a greater importance in sexual populations with a long evolutionary history (i.e. those with a greater number of recombination events behind; Weissmann, 1889; Crow and Kimura, 1965; Maynard Smith, 1968; Burt, 2000; Rushworth etal., 2020). In contrast, clonal propagation has been considered to reduce the opportunities for local adaptation (Schon etal., 1998; Rouzine etal., 2003; Schiffels et al., 2011) although this mechanism can theoretically occur through selection on genes or genotypes (Vrijenhoek, 1979; Lushai et al., 2003). Given the expected reduction in genotypic diversity, phenotypic plasticity has been attributed a more important role in asexual lineages (Lynch, 1984; Van Kleunen and Fischer, 2001; Oplaat and Verhoeven, 2015; Fazlioglu and Bonser, 2016; Geng et al., 2016) and during the initial stages of population establishment (Davidson etal., 2011; Liao etal., 2016). Introduced species often evolve to cope with novel biotic and abiotic conditions in non-native ranges (Bossdorf et al., 2005; Vandepitte et al., 2014; Oduor et al., 2016; Mitchell and Whitney, 2018; Liu et al., 2020), and thus constitute an excellent model system to study adaptive evolution occurring over short periods of time (Thompson, 1998; Colautti and Lau, 2015). In addition, phenotypic plasticity seems to make a major contribution to the establishment and spread of introduced species in novel environments (Ghalambor etal., 2007; Riis etal., 2010; Ebeling etal., 2011; Pahl etal., 2013; Liao etal., 2016; Liu etal., 2016). Interestingly, clonal propagation Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 656 is an advantageous trait for plant invasions, and numerous invasive plant species combine both sexual and asexual modes of reproduction or are mostly asexual (Pyšek, 1997; Silvertown, 2008; Roiloa, 2019). However, although an increasing number of studies have shown evolution at a contemporary scale in invasive plants with sexual (Maron etal., 2004; Leger and Rice, 2007; Novy etal., 2013; Lucek etal., 2014; Li etal., 2015; Bhattarai etal., 2017; Marchini etal., 2018) or mixed reproductive systems (Michel etal., 2004; Lambertini etal., 2010), field tests of local adaptation and phenotypic plasticity are rare for obligately asexual flowering plants (Lovell et al., 2014; Rushworth etal., 2020). In this study, we investigate the evolutionary strategies of two invasive Mimulus (Phrymaceae) species that differ in their ability to reproduce sexually: Mimulus guttatus (which combines sexual and asexual reproduction) and M. × robertsii (strictly asexual). We evaluate the roles of adaptive evolution and phenotypic plasticity in the invasive success of Mimulus at two nested levels: (1) between native and introduced populations, and (2) among introduced populations. In a first experiment, we assess phenotypic differences and compare the clonality and plasticity of ancestral native (M.guttatus) and invasive (M.guttatus and M. × robertsii) populations under the environmental conditions at each latitudinal extreme of the UK, discerning the roles of temperature and photoperiod in a fullcrossed design implemented in controlled environment chambers. Our hypothesis here is that asexual M. × robertsii should display levels of clonality and plasticity equal to or higher than those of the sexual taxa (native M.guttatus and invasive M.guttatus). In a second experiment, we test the existence of local adaptation of the two species within the invasive range with a reciprocal transplant experiment at two field sites in the latitudinal ends of the UK and analyse which phenotypic traits underlie the local fitness advantage in each species. We predict that if sexual reproduction boosts adaptation, M.guttatus would be more likely to be locally adapted than M. × robertsii. To explore the possible mechanisms driving local adaptation, we also carry out phenotypic selection analyses to identify the phenotypic traits related to fitness in each species. MATERIALS ANDMETHODS Studysystem Mimulus guttatus DC. (2n = 2x = 28) is an herbaceous, annual or perennial plant native to Western North America (Grant, 1924; Wu etal., 2008; Lowry and Willis, 2010). M.guttatus was introduced to the UK for ornamental purposes 200years ago (Roberts, 1964; Parker, 1975; Puzey and Vallejo-Marín, 2014) and perennial forms, which combine reproduction via seeds (sexual) and stolons (asexual), became naturalized in wetlands, riverbanks and wet ditches across the entire country (Preston et al., 2002), as in other areas in Europe and New Zealand (Howell and Sawyer, 2006; Truscott et al., 2006; Da Re et al., 2020). The second taxon, Mimulus × robertsii Silverside, is a triploid (2n = 3x = 44–46) that originated in the UK, the product of an unknown number of hybridization events between introduced populations of the diploid M. guttatus and the closely related South American tetraploid M. luteus L.(2n = 4x = 60–62). M.luteus was introduced in the UK soon after M.guttatus but is currently rare (Vallejo-Marín and Lye, 2013). The hybrid M × robertsii, which is perennial and sexually sterile (Parker, 1975; Meeus etal., 2020) but capable of extensive clonal reproduction via stolons, has become well established across the UK, though it is far less abundant than M.guttatus in the south of the country (Preston etal., 2002; Stace, 2010; Vallejo-Marín and Lye, 2013; Da Re etal., 2020). M.guttatus and M. × robertsii are very similar in their morphology, phenology and habitat in the UK. Both species bear high genetic diversity and low genetic structure (Vallejo-Marín and Lye, 2013; Pantoja etal., 2017), suggesting that metapopulation dynamics with high gene flow are important in the spatial structuring of the introduced range. Experiment 1: controlled environment chambers Plant material. For M.guttatus, we used seeds from five native populations from North America and from five introduced populations in the UK (Supplementary Data Table S1). We follow Lowry etal. (2019) and use the classic taxonomic definition of M.guttatus (Grant, 1924), rather than the recent nomenclature proposed by Nesom (2014). All seeds were field-collected, except seeds from accessions in the Alaskan range, the putative ancestral range of UK populations (Puzey and Vallejo-Marín, 2014; Vallejo-Marin etal., 2020). Three Alaskan accessions were retrieved from herbarium specimens preserved at University of Alaska Museum Herbarium (accessions V153408, V127607, V142998). As each accession represents a single sampled individual and locality, these three accessions were pooled into a single Alaskan ‘population’ (ALA). From each population, we selected three to five maternal seed families (seeds collected from the same maternal parent). In total, we had 43 families from ten populations. For the sexually sterile hybrid M. × robertsii, we collected in the field vegetative fragments (clones) from five UK populations (Table S1). In each population, we sampled 1–5 ramets (limited by population size) separated by at least 1m to reduce the probability of sampling the same genet multiple times (15 ramets in total from five populations). All maternal families in both species were randomly collected with regard to their clonality and phenotypic traits. Native and invasive populations of M.guttatus cover a wide latitudinal range of their distribution, while M. × robertsii populations came from the centre and north of the species’ narrower range. Experimental treatments. We used the Controlled Environment Facility at the University of Stirling to create environmental conditions that resembled the UK Mimulus growing season (Fig. 1A). To model the conditions, we used two opposite localities that encompass the latitudinal range of Mimulus in the UK: Newport, in the Isle of Wight (50.70°N, 1.29°W), and Baltasound, in the Shetland Isles (60.76°N, 0.86°W). For each of these localities and for every 2-week period between April and September (the UK Mimulus growing season), we calculated the photoperiod with the package geosphere (Hijmans etal., 2014) in R (R Core Team, 2019)and obtained maximum and minimum temperatures from the WorldClim database (Hijmans Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 657 etal., 2005). Photoperiod and temperature temporal series were combined in a full-crossed design to create four experimental treatments that allowed us to disentangle the effects of temperature and photoperiod on plant performance: a short-day, warm temperature treatment (SW; the natural conditions in Newport), a long-day, cold temperature treatment (LC; the natural conditions in Baltasound), a short-day, cool temperature treatment (SC) and a long-day, warm temperature treatment (LW) (Supplementary Data Fig. S1 and Table S2). The different growth conditions in each temporal series were substituted every 10d to allow completion of the experiment in 120d. Each experimental treatment was implemeted in one Snijder Scientific (Tilburg, Netherlands) MC1750E controlled environmental chamber. Plant growth. We planted seeds from each of 43 maternal families of M.guttatus into four 0.5-L pots (172 pots in total) filled with modular seed growing medium (Sinclair, Lincoln, UK), and placed them in the dark at 4°C for 1week. For M. × robertsii, we planted individually eight cuttings from each of 15 maternal families in 0.5-L pots (120 pots in total) filled with All-Purpose growing medium (Sinclair). All cuttings had a similar size, two small leaves and ~2cm of roots. We moved one pot per maternal family of M.guttatus and two pots per maternal family of M. × robertsii to each chamber on 1 May, 2014. We noted the day of first germination for each M.guttatus pot and, 4weeks after first germination, we selected and thinned the two biggest seedlings to one per pot filled with All-Purpose growing medium to obtain two replicates per maternal family in each chamber. The maximum difference in transplant time among pots was 1week within each chamber and 2weeks across the entire experiment. Pots were randomly repositioned within each chamber every other day throughout the experiment. Measurements and statistical analyses. Most individuals survived until the end of the experiment, and all measurements were taken at this moment except where otherwise specified. To investigate the phenotypic differences in clonality among the three population types (classified by their origin and species, i.e. native M.guttatus, invasive M.guttatus and M. × robertsii), we recorded the total number of stolons produced by individuals in the four environmental chambers. To compare their phenotypes, we recorded days to flower since germination or planting of the clonal fragment, corolla width of the first flower (measured with a digital calliper to the nearest 0.1mm on the second day after anthesis), whether plants flowered or not, the number of branches, floral stems and flowers, plant height (from the soil surface to the highest meristem, measured to the nearest cm), and length and diameter of the first internode. Finally, the entire individuals (above and below ground) were harvested, washed gently in water and dried at 60°C in individual paper bags to estimate final total dry biomass. Despite being sterile, we consider M. × robertsii flowering as indicative of individual performance. To avoid over-parameterization in subsequent analyses, we averaged the two values from each family (cuttings in M. × robertsii, siblings in M.guttatus), for each trait under each of four treatments (except for germination time in M.guttatus, which had a single data point per family). Preliminary analyses showed low correlation between most phenotypic traits (Supplementary Data Fig. S2) and thus each variable was analysed separately. We used Generalized Linear Mixed Models (GLMMs) to analyse the variation in clonal reproduction and in phenotypic traits as a function of the population type (native M.guttatus, introduced M.guttatus and M. × robertsii), treatment photoperiod (Short vs. Long), treatment temperature (Warm vs. Cold), and their two- and three-way interaction terms. Analogous AB C Fig. 1. Experimental set-ups to test for phenotypic plasticity and local adaptation of Mimulus in the UK. Seedlings growing in a controlled environmental chamber at the University of Stirling in 2014 (A). Common gardens set in the Isle of Wight (B) and Shetland (C) in summer 2015. Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 658 GLMMs were also carried out for each population type separately. In all models, population was included as a random effect. We used a Poisson error distribution for number of stolons, branches, floral stems and flowers, a binomial model for flowering, and a Gaussian model for germination time, flowering time, corolla width, plant height, internode length, internode diameter and dry mass. The survival of plants was above 96% and thus this variable was not modelled. The significance of the fixed effects and their interactions were assessed by type-III Wald χ 2 tests on the corresponding GLMMs. Where the interactions were not significant, we removed them and tested also the effect of the main effects alone with type-II Wald χ 2 tests. To account for multiple tests, we applied a Bonferroni correction, dividing the significance alpha level by the number of variables analysed (corrected P-value = 0.004). Where population type or any interaction of fixed factors were significant, we performed post hoc contrasts based on estimated marginal means (EMMs) of the corresponding model. These procedures were repeated for all GLMMs in this study. All analyses were performed in R 3.4.0 (R Core Team, 2019)with the packages lme4 (Bates etal., 2015), car (Fox etal., 2012) and emmeans (Lenth etal., 2018). To investigate differences in phenotypic plasticity among population types, we estimated the Relative Distances Plasticity Index (RDPI; Valladares etal., 2006) for each trait measured in the chambers, for each family. RDPI values were first estimated from trait distances between the two temperature (RDPIt) and the two photoperiod (RDPIp) treatments separately, pooling data from two chambers in each treatment. Trait distances were calculated as the absolute value of the difference of trait values of the same family (the average of the two individual replicates) at each of two treatments, divided by the maximum of the two trait values. RDPI values were also estimated for each family across the four environmental chambers (RDPItp) as the average of the six trait distances between each pair of chambers. We analysed the variation in phenotypic plasticity with GLMMs modelling RDPIt, RDPIp and RDPItp estimates as a function of the population type. In addition, we ran multivariate analyses of variance (MANOVAs) with RDPIt, RDPIp or RDPItp estimates for all phenotypic traits as response variables and population type as the independent variable. RDPI estimates for germination day were excluded for multivariate analyses because the lack of data for M. × robertsii. Finally, to test for differences in plasticity in response to temperature and photoperiod, we pooled RDPIt and RDPIp estimates and used a GLMM to model RDPI values as a function of the RDPI type, population type and their interaction. All RDPI GLMMs used a Gaussian distribution of errors. Experiment 2: Reciprocal transplants Population survey and plant material. We used the distribution database of the Botanical Society of Britain and Ireland (BSBI; http://bsbidb.org.uk/) and personal records to design a survey of the northern and the southern extremes of the distribution of M.guttatus and M. × robertsii in the UK in summer 2014. We focused on BSBI records from the year 2000 with a precision of at least 100m2. In total, we visited 60 localities between 50.1132° and 51.1489°N for the south of the country and 57.4963° and 60.8087°N for the north and found 39 populations. Because we were interested in identifying potential ecotypes adapted to the latitudinal extremes of the UK, we prioritized sampling fewer individuals in a higher number of populations instead of large numbers of individuals in fewer populations. This strategy has shown great statistical power (Blanquart etal., 2013) and was suited to our study system as many populations of Mimulus were small and probably contained few genets. To avoid sampling clones more than once, collected plants were at least 1 m apart from each other. Cuttings were transported and planted at the glasshouse of the University of Stirling within 1week of collection. In total, we sampled 155 cuttings from 36 populations for this study (Fig. 2; Supplementary Data Table S3). M.guttatus and M. × robertsii are morphologically very similar and sometimes difficult to distinguish (Vallejo-Marín and Lye, 2013). To verify the species identity of each sampled individual, we determined their relative genome size with flow cytometry (see methods in Simón- Porcar etal., 2017). To allow us to compare M. × robertsii with both parental species, the only available population of M.luteus in the UK for which we had seeds was included in this experiment (Fig. 2; Table S3). For M. luteus, field-collected seeds from 25 different maternal individuals were planted and, once germinated, one seedling per maternal family was transplanted and grown untiladult. We kept individual plants in 9-cm-diameter pots filled with All-Purpose growing medium until the next summer season. To buffer maternal resources effects, we transplanted a similar size fragment from each individual and randomized its position within the glasshouse at least four times between summer seasons. We cloned each individual four times to obtain replicates by April 2015, 1month before setting up the experiment. All cuttings had similar architecture and size, and they were weighed prior to planting to evaluate possible maternal resources effects on subsequent measures of fitness. Clones were allowed to establish and develop roots and below-ground biomass, similarly to how they naturally persist between growing seasons, but we restricted aerial biomass to the initial status by pruning elongating branches until the start of the experiment. Experimental design. Two replicates per individual, for a total of 360 plants, were transplanted into each of two common gardens (Supplementary Data Table S3). We established one common garden in the southernmost extreme of the UK at Ventnor Botanic Garden (Ventnor, Isle of Wight, England; 50.5890°, −1.2285°; hereafter IOW; Figs 1B and 2) on 14 May, 2015, and a second common garden in the northernmost extreme of the UK at Da Gairdins i Sand (Sand, Shetland, Scotland; 60.2112°, −1.3761°; hereafter SHE; Figs 1C and 2) on 18 May, 2015. The two common gardens were set up to be identical, consisting of a 100-m2 square pond built up with a PVC pond liner (Aquatex, LBS Horticultural, UK), filled with 1cm of gravel to imitate natural conditions and provide an appropriate environment for root growth. Individual clones were planted in 10-L pots filled with 7L of all-purpose commercial growing medium (LBS Horticultural), which were placed in the pond in a regular grid with individuals from different species and origins completely randomized. Pots were 25cm apart and pot walls precluded stolons from exiting the pot, avoiding Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 659 mingling or competition. The ponds were permanently flooded at a level of 10cm so that plants were always moist as in natural habitats. The experiment was terminated at the end of the growing season, after senescence of the aerial parts of all individuals on 24 and 30 August in IOW and SHE, respectively. Measurements and statistical analyses. To explore local adaptation, we assessed the fitness of individuals at each site recording their survival and reproductive success (number of stolons, and fruits in M.guttatus) at the end of the experiment. We also explored phenotypic differentiation and the traits contributing to local fitness differences within each species through phenotypic selection. For this we recorded plant height and stomata density (in the 6th week of the experiment, when plants seemed to have achieved their maximum vigour); whether plants had flowered or not, plant cover, total dry biomass, and total number of branches, flowering stems and flowers produced (at the end of the experiment); and days to flower, first flowering node and corolla width of the first flower (at flowering of each individual). Stomata density, a trait involved in the hydric balance of plants and thus a potential indicator of physiological variations (Raven, 2002), was calculated under a 50× microscope from stomata imprints taken with transparent nail paint, adhesive tape and microscope slides from the beam of three new unshaded leaves per individual. Plant cover was measured over scaled overhead view photographs of each individual that were analysed with Mimulus x robertsii Mimulus luteus Mimulus guttatus 0 100 200 km SHE IOW Fig. 2. Map of populations (circles) and experimental sites (triangles) used in the reciprocal transplants. Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 660 the software ImageJ (Abramoff etal., 2004). To estimate dry biomass, the entire individuals (above- and below-ground) were harvested, washed and dried at 60°C in individual paperbags. To ascertain the existence of local adaptation in M.guttatus and M. × robertsii, we analysed the variation in the sexual and asexual fitness measures (number of fruits and stolons) of each species with GLMMs, including Site, Origin and their interaction as fixed factors in the models for each variable. The models used a Poisson distribution and included initial cutting weight as a covariate and population and individual nested within population as random factors. We consider a pattern of fitness advantage at home sites jointly with a significant effect of the interaction Site × Origin as evidence of local adaptation. The survival of plants was nearly 100% (see Results) and thus this variable was not modelled. To explore the phenotypic differentiation between possible latitudinal ecotypes and compare the natural environmental effects on the growth of plants with the effects found in the environmental chambers experiment, we carried out similar GLMMs for each species and phenotypic trait measured. APoisson model was used to analyse the number of branches, floral stems and flowers, and a Gaussian model was used for the remaining variables. We consider a significant effect of Origin as evidence of genetic differentiation, and a significant effect of Site as evidence of strong environmental effects (i.e. plasticity) on the development and growth ofplants. To investigate the phenotypic traits contributing to local fitness, we carried out phenotypic selection analyses by regressing the sexual and asexual fitness measures (number of fruits and stolons produced) on standardized phenotypic traits separately for each species and site. Only selection gradients were estimated to determine the magnitude and sign of directional and stabilizing selection on each trait, excluding indirect selection on correlated traits (Lande and Arnold, 1983). Separately for each species and site, we calculated the relative fitness (individual fitness divided by mean fitness) and standardized trait values (with a mean of 0 and a variance of 1). To improve the normality of the residuals in the regression models, the relative numbers of fruits and stolons were root squared. Because preliminary analyses had showed a correlation between various phenotypic traits for each species in this experiment (Supplementary Data Fig. S3), we calculated the variance inflation factors (VIFs) in each model and excluded those traits with VIF > 5 (i.e. number of branches and number of floral stems in models for SHE). Quadratic regression coefficients were doubled to estimate the stabilizing/disruptive selection differentials (Stinchcombe etal.,2008). To investigate the causes of the low occurrence of M.luteus in the UK and compare the patterns found in the hybrid M. × robertsii with both parental species, we assessed the sexual and asexual fitness and the phenotypic patterns of the single population included in our experiment. We recognize that the study of a single population does not allow robust inferences on the species patterns but given the great scarcity of M.luteus populations in the UK, we still consider this approach worthy and valuable for species comparisons. The production of fruits and stolons, and each phenotypic trait measured, were analysed as a function of experimental site with GLMMs including initial weight as a covariate and individual as a random factor. Then, we compared the fitness of M.luteus with the other two Mimulus species and tested the phenotypic similarity of M. × robertsii and M.luteus with GLMMs including species, experimental site, and their interaction as fixed factors, initial weight as a covariate, and population and individual nested within population as random factors. Because M.luteus had a single population in the north, the southern populations of M. × robertsii and M.guttatus, and the variable ‘population origin’ were excluded from these analyses. Finally, we carried out phenotypic selection analyses on M.luteus as explained above. RESULTS Experiment 1: controlled environment chambers Clonality. The population types differed in clonality (χ 2 = 17.974; P < 0.001), with M. × robertsii producing the most stolons, significantly more than native M.guttatus (Fig. 3). Overall, clonal reproduction was not affected by photoperiod but it was affected by temperature (χ 2 = 8.670; P = 0.003). The significant interaction of population type and temperature (χ 2 = 32.035; P < 0.001) reflected that warm treatments increased clonality in both M.guttatus groups, but decreased clonality in M. × robertsii (Fig. 3; Supplementary Data Tables S4 and S5). 5 10 15 LW SW LC SC #Stolons M. x robertsii Invasive M. guttatus Native M. guttatus Treatment Fig. 3. Clonality of Mimulus × robertsii, introduced M.guttatus and native M. guttatus populations grown in four different controlled environmental chambers with contrasting photoperiods (L: long; S: short) and temperatures (C: cold; W: warm) in a crossed design. Mean values and standard errors of the variables measured are indicated by dots and error bars, respectively. Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 661 Table 1. Comparison of phenotypic plasticity among population types. Results of the MANOVAs and post hoc tests analysing RDPI phenotypic plasticity indexes for all traits measured in the controlled environmental chambers as a function of the population type. Mean ± s.d. RDPI values across traits for each population type are provided jointly with the results of the post hoc tests MANOVA Mean ± sd; post hoc d.f. d.f. residuals Pillai’s trace Approx. Fnum d.f. den d.f. P M.guttatus native M.guttatus invasive M. × robertsii RDPIt2 50 0.642 2.261 18 86 0.006** 0.312 ± 0.052 a 0.291 ± 0.042 a 0.318 ± 0.071 a RDPIp2 47 0.830 3.154 18 80 <0.001*** 0.218 ± 0.041 a 0.247 ± 0.056 ab 0.354 ± 0.078 b RDPItp 2 32 0.903 2.287 18 50 0.011* 0.299 ± 0.038 a 0.317 ± 0.046 a 0.346 ± 0.065 a Different letters indicate significant differences. *P < 0.05; **P < 0.01; ***P < 0.001. Phenotypes. Overall, M. × robertsii plants were shorter and thinner, and produced fewer branches, floral stems and flowers than both M.guttatus groups. Invasive M.guttatus produced fewer floral stems and flowers than native M.guttatus (χ 2 > 13.928; P < 0.001) (Supplementary Data Fig. S4 and Tables S4 and S5). Warm treatments strongly accelerated germination and flowering in all population types, increased flower production, most significantly in native M. guttatus, and decreased corolla width, most significantly in invasive M. guttatus. Warm treatments also increased plant height in all groups, more sharply in M.guttatus than in M. × robertsii, increased internode length and decreased internode diameter and dry mass, most significantly in invasive M.guttatus, and increased the number of branches in both M.guttatus groups (χ 2 > 9.094; P < 0.002) (Fig. S4 and Tables S4 and S5). Short days delayed flowering in both M.guttatus groups, and strongly decreased the probability of flowering in invasive M.guttatus and M. × robertsii, the production of stems in M.guttatus, and flower production in all groups, more markedly in M.guttatus than in M. × robertsii. Short days also reduced plant height, most significantly in invasive M. guttatus and M. × robertsii, decreased internode length in both M.guttatus groups, and decreased internode diameter and dry mass, most significantly in invasive M.guttatus (χ 2 > 8.911; P < 0.002) (Fig. S4 and Tables S4 and S5). The interaction of temperature and photoperiod had an effect on the production of flowers in M.guttatus, with SC and LW treatments having the lowest and greatest flower production, respectively (χ 2 > 14.412; P < 0.001). The three-way interaction of factors was always non-significant (Fig. S4 and Tables S4 and S5). Phenotypic plasticity. The overall values for RDPIt, RDPIp and RDPItp were 0.307 ± 0.031, 0.27 ± 0.034 and 0.32 ± 0.028 (mean ± s.d.), respectively. RDPIt estimates did not differ among groups for any trait after Bonferroni correction (χ 2 < 8.678; P > 0.013). The RDPIp estimates for flowering day, number of flowers, floral stems, branches and plant height varied significantly among population types (χ 2 > 11.991; P < 0.002). In most cases the post hoc tests indicated significantly greater plasticity in M. × robertsii than in the other groups (Supplementary Data Table S6). RDPItp estimates for dry mass were also significantly greater in M. × robertsii than in the other groups (χ 2 = 13.655; P = 0.001) (Table S6). MANOVAs found significant differences among population types for RDPItp, RDPIt and RDPIp estimates (Pillai’s trace = 0.642–0.903; F > 2.261; P < 0.01; Table 1). M. × robertsii showed the greatest RDPI values, although the post hoc analyses showed only significantly differences in RDPIp between M. × robertsii and native M.guttatus (Table 1). In the GLMM pooling RDPIt and RDPIp estimates, all fixed factors (RDPI type, population type and their interaction) were significant (χ 2 > 6.716; P < 0.02). M. guttatus had higher RDPIt than RDPIp estimates and the opposite was found in M. × robertsii (differences were significant only within native M.guttatus). RDPIp estimates of M. × robertsii were significantly higher than RDPIt estimates of native M.guttatus. All RDPI estimates for the production of stolons were similar for all population types (χ 2 < 5.034; P > 0.081) (Table S6). Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 662 Experiment 2: reciprocal transplants Local adaptation. The survival and flowering of plants was respectively above 98% and 96% across the experiment. The fruit set of M.guttatus populations was significantly dependent on the experimental site and the interaction of experimental site and population origin (χ 2 > 50.669; P < 0.001; Table 2). This species produced fewer fruits in SHE than in IOW, with a significantly greater decrease for southern populations (Fig. 4). The production of stolons was not significantly dependent on any modelled factor in M.guttatus (Table 2), but it was also dependent on the experimental site and the interaction of experimental site and population origin in M. × robertsii (χ 2 > 14.81; P < 0.001; Table 2). Overall, the production of stolons in M. × robertsii was higher in SHE than in IOW, and this was based on a high increase in northern populations. By contrast, southern populations showed a slightly lower production of stolons in SHE than in IOW (Fig. 4). Phenotypic differentiation. Across most traits and for both species, plants were similar regardless of their latitudinal origin. In M.guttatus, northern individuals produced flowers with bigger corollas than southern individuals (χ 2 = 6.566; P = 0.01) (Supplementary Data Table S7 and Fig. S5). Experimental site had a strong effect on the development of plants. M. guttatus individuals flowered later, produced fewer flowers, had lower stomata density, and grew less according to plant cover and final dry mass in SHE than in IOW (χ 2 > 24.524; P < 0.001). M. × robertsii flowered later, M. guttatus M. x robertsii 8 7 6 5 4 30 25 20 15 IOW SHE IOWSHE North South ORIGIN Fruits Stolons Fig. 4. Fitness of Mimulus guttatus and M. × robertsii individuals included in the reciprocal transplant experiment between different latitudes in the UK. Mean values and standard errors of the number of fruits and stolons are indicated by dots and error bars, respectively. Table 2. Results of the GLMMs modelling the effects of experimental site, population origin (O) and their interaction, and initial weigth (W0), in the sexual and asexual fitness traits recorded in a reciprocal transplant experiment with introduced Mimulus guttatus and M. × robertsii populations. M.guttatus M. × robertsii Trait Fixed factor Estimate (s.e.) χ 2Estimate (s.e.) χ 2 Fruits Intercept 2.776 (0.266) 108.875 *** W00 (0.001) 0.003 O (South) 0.317 (0.353) 0.805 Site (SHE) −0.599 (0.038) 248.638*** Site:O (SHE:South) −0.357 (0.05) 50.669*** Stolons Intercept 1.718 (0.14) 150.518 *** 1.409 (0.127) 122.591 *** W00.001 (0.001) 0.235 0 (0) 0.016 O (South) 0.132 (0.119) 1.235 0.314 (0.233) 1.809 Site (SHE) 0.07 (0.061) 1.295 0.467 (0.072) 41.635*** Site:O (SHE:South) −0.033 (0.073) 0.212 −0.522 (0.136) 14.81*** Model estimates and standard errors for each fixed factor and interaction are provided jointly with results of the type-III Wald χ 2 tests. χ 2 values and indications of their associated P-values are provided. *P < 0.05; **P < 0.01; ***P < 0.001. Significant effects after Bonferroni correction of P-values are indicated in bold. χ2 degrees of freedom = 1. Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022 Simón-Porcar etal. — Adaptation in sexual and asexual Mimulus 663 produced fewer floral stems and flowers, had lower dry mass, and produced more stolons in SHE than in IOW (χ 2 > 8.4; P < 0.01) (Table S7 and Fig. S5). The interaction site × origin was significant for the number of branches, stems and flowers in M. guttatus, with negative estimates for south plants in SHE. In M. × robertsii, site × origin was significant for the number of flowers, with positive estimates for south plants in SHE (χ 2 > 9.933; P < 0.01) (Table S7 and Fig. S5). Phenotypic selection. The selection gradients differed between species, fitness traits and sites, suggesting diverse mechanisms of local adaptation in each species. In M. guttatus, fruit set (sexual fitness) in IOW showed significant positive linear selection and stabilizing selection on flowering day and dry mass, and significant negative linear selection and disruptive selection in height (t > 2.453; P < 0.015). In SHE there was significant positive linear selection and stabilizing selection in corolla width (t > 2.09; P < 0.037; Fig. 5; Supplementary Data Table S8). For M.guttatus stolons (asexual fitness) we found only significant negative linear selection and disruptive selection in dry mass in SHE (t > 2.02; P < 0.044; Fig. 5). For M. × robertsii stolons we found positive linear selection in dry mass in IOW, and significant positive linear selection and stabilizing selection in corolla width in SHE (t > 2.073; P < 0.042; Fig. 5; Table S8). M.luteus produced significantly more fruits in IOW than in SHE (χ 2 > 96.962; P < 0.001) and a similar number of stolons in both sites (χ 2 = 3.329; P = 0.068). There were no differences between M. luteus and north M. guttatus in sexual or asexual fitness overall (χ 2 < 4.336; P > 0.1), but M.luteus produced relatively more fruits than north M.guttatus in SHE (χ 2 = 11.018; P < 0.001). The production of stolons was higher in M.luteus than in north M. × robertsii overall, but it was lower in SHE (χ 2 > 9.078; P < 0.003). The models of phenotypic traits showed that M.luteus flowered later, produced more branches and floral stems, and had lower stomata density and dry mass in SHE than in IOW (χ 2 > 8.234; P < 0.01). The phenotypic traits of north M. × robertsii and M.luteus did not differ significantly, but north M. × robertsii produced relatively fewer branches, floral stems and flowers than M.luteus in SHE (negative coefficients for north M. × robertsii in SHE; χ 2 > 9.596; P < 0.01) (Supplementary Data Table S9). The phenotypic selection analyses through fruit set in M. luteus showed significant positive linear selection in dry mass, stabilizing selection in the number of branches and dry mass, and disruptive selection in stomata density in IOW (t > 2.169; P < 0.04). The models regressing the number of stolons indicated positive linear and stabilizing selection on stomata density in IOW (t > 2.232; P < 0.035) (Table S8 and Fig. S6)). DISCUSSION Clonality, phenotypic and plasticity changes in invasiveMimulus The reproductive systems of native M. guttatus, invasive M. guttatus and invasive M. × robertsii showed a transition from a relatively higher investment in sexual organs (i.e. floral Plant cover Height Flowers Flo wering node Flowering day Floral stems Dry mass Corolla Branches IOW Fruits Stolons SHE Estimate –4 –2 024–4–2024–4 –2 02 4–4–2024 IOW SHE Stomata Stomata Plant cover Height Flowers Flo wering node Flowering day Floral stems Dry mass Corolla Branches M. guttatus M. x robertsii Fig. 5. Estimates and 95% confidence intervals for the linear (black) and quadratic (gray) phenotypic selection coefficients on each trait included in the selection gradient analyses of Mimulus guttatus and M. × robertsii in the reciprocal transplants experiments at each site. *P < 0.05; **P < 0.01; ***P < 0.001. Downloaded from https://academic.oup.com/aob/article/127/5/655/6144755 by UNIVERSIDAD DE SEVILLA user on 18 March 2022