Introduced plants of Lupinus polyphyllus are larger but flower less frequently than conspecifics from the native range: Results of the first year

Abstract Introduced species, which establish in novel environments, provide an opportunity to explore trait evolution and how it may contribute to the distribution and spread of species. Here, we explore trait changes of the perennial herb Lupinus polyphyllus based on 11 native populations in the western USA and 17 introduced populations in Finland. More specifically, we investigated whether introduced populations outperformed native populations in traits measured in situ (seed mass) and under common garden conditions during their first year (plant size, flowering probability, and number of flowering shoots). We also explored whether climate of origin (temperature) influenced plant traits and quantified the degree to which trait variability was explained collectively by country and temperature as compared to other population‐level differences. Three out of four plant traits differed between the native and introduced populations; only seed mass was similar between countries, with most of its variation attributed to other sources of intraspecific variation not accounted for by country and temperature. Under common garden conditions, plants originating from introduced populations were larger than those originating from native populations. However, plants from the introduced range flowered less frequently and had fewer flowering shoots than their native‐range counterparts. Temperature of a population's origin influenced plant size in the common garden, with plant size increasing with increasing mean annual temperature in both native and introduced populations. Our results of the first year reveal genetic basis for phenotypic differences in some fitness‐related traits between the native and introduced populations of L. polyphyllus. However, not all of these trait differences necessarily contribute to the invasion success of the species and thus may not be adaptive, which raises a question how persistent the trait differences observed in the first year are later in individuals’ life for perennial herbs.


| INTRODUC TI ON
Introduced species can be free from many of the ecological and evolutionary constraints present in their native range, such as natural enemies and competitors that are adapted to their presence (e.g., Callaway & Aschehoug, 2000). As a consequence, introduced species may provide valuable information on trait evolution and species responses to novel environments (e.g., Colautti & Lau, 2015;Parker et al., 2003). Numerous studies that have compared plant traits between native and introduced populations have reported differences (e.g., Buckley et al., 2003;Ebeling et al., 2008;Turner et al., 2014;Zhang et al., 2018), with introduced plants tending to be larger and more fecund than their conspecifics from the native range (reviewed in Parker et al., 2013). Given that these traits can affect population dynamics, such trait changes can contribute to a higher population growth rate and faster population spread in species' introduced ranges. Understanding the causes of trait changes in introduced species is therefore essential for predicting shifts in plant communities under global environmental change.
One potential explanation of differences in plant traits observed in situ between native and introduced populations could be that they arise from rapid evolution in the introduced range in response to changes in the abiotic and biotic environment (Colautti & Lau, 2015).
For example, a lack of specialist enemies (e.g., herbivores, pathogens) in new environments might enable individuals to invest more in growth and fecundity instead of defense against enemies (Blossey & Nötzold, 1995;Joshi & Vrieling, 2005). Since many invasive plant species are originally introduced as ornamentals (Hulme et al., 2018), trait differences can also result from pre-introduction evolution due to horticultural breeding (Kitajima et al., 2006;te Beest et al., 2012).
Alternatively, phenotypic differences between native and introduced populations could be environmental in origin, arising as a result of more favorable growth conditions in the introduced range (Moloney et al., 2009). As an example, in the annual Conyza canadensis, individuals in the introduced range outperformed those in the native range, but the two groups of plants performed similarly when they were exposed to a common environment (Rosche et al., 2019), suggesting that trait differences were phenotypic rather than genetic. As these results indicate, in order to reveal the potential genetic basis (if any) for the success of invasive populations, a comparison under standardized conditions is necessary (Moloney et al., 2009). Plant traits are highly variable within species, with a significant proportion of the variability arising from differences among populations within both native and introduced ranges (e.g., Ebeling et al., 2008;Rosche et al., 2019); for this reason, it is necessary to sample multiple populations from each range. Climate can be the main selective force for plant traits, resulting in phenotypic variation along climatic clines. Such clinal variation is expected particularly for native populations due to their long evolutionary history.
Introduced populations, in turn, may either breakdown environmental constraints (Endriss et al., 2018;Smith et al., 2020) or may adapt to local abiotic conditions rapidly, showing parallel climatic clines to their native counterparts (e.g., Hodgins & Rieseber, 2011;McGoey et al., 2020). It is therefore preferable that analyses of plant performance also consider differences in abiotic conditions that may contribute to trait variability (Colautti et al., 2009;Rosche et al., 2019).
Here, we explore trait differences in the perennial herb Lupinus polyphyllus (Lindl.) based on 11 native populations in the western USA and 17 introduced populations in Finland. More specifically, we investigated variation in seed mass in situ, and variation in plant size, flowering probability, and number of flowering shoots during the first growing season under common garden conditions in the introduced range. All four traits are key components of the population dynamics of the species and are thus closely related to its fitness (Ramula, 2014;Sõber & Ramula, 2013). We asked three questions: (a) Do plants from the introduced populations differ those from the native populations in terms of plant traits? (b) Does climate of origin influence plant traits similarly in both native and introduced populations? (c) How much trait variability is explained collectively by country and climate as compared to other sources of intraspecific variation? We predicted that plants from the introduced populations would be larger in size and would have higher flowering probability and flowering shoot production than those from the native population. Moreover, due to their longer evolutionary history, we predicted that plants from the native populations would exhibit a stronger association with climate of their origin than plants from the introduced populations. Finally, we predicted that all traits would exhibit population-level variation not explained by country and climate.

| Study system, DNA barcoding, and climate variables
Lupinus polyphyllus (garden lupin, Fabaceae) is a 50-100 cm high, short-lived perennial herb that is native to parts of western North America that have mostly an oceanic climate; it is invasive in Europe, southern Australia, New Zealand, and Chile (Fremstad, 2010;Meier et al., 2013). In Finland, the species was recorded as a garden escaper in the southern parts of the country in the late 1800s (Fremstad, 2010), and it is currently associated with declines in local flora and insect fauna (Valtonen et al., 2006;Ramula & Pihlaja, 2012;Ramula & Sorvari, 2017). It inhabits moist meadows and river banks in the native range (Beuthin, 2012), and road verges, wastelands, and forest understories in the introduced range (Fremstad, 2010). The species reproduces mostly by seed but vegetative reproduction via rhizomes is possible (Li et al., 2016). An individual plant is able to produce hundreds of seeds (Aniszewski et al., 2001;Ramula, 2014) which are dispersed ballistically up to a few meters from the mother plant (Jantunen et al., 2005) and may remain viable in the soil for decades (Fremstad, 2010).
In July-August 2018, we collected seeds from 16 putatively native populations in the western USA and 17 introduced populations in Finland across a latitudinal gradient. In the native range, unlike in Finland, several Lupinus species co-exist. Because the American populations were visited only once for seed collection (outside the flowering season), we used a standard molecular barcoding of leaf samples collected from the seedlings in the greenhouse (see below) to confirm species identity. Species identity for the Finnish populations was determined in the field during the flowering period. DNA of 3-4 individuals per population (66 samples in total) was extracted from frozen leaf samples using NucleoSpin Plant II-kit (Macherey-Nagel) following the manufacturer's instructions. DNA barcoding was done by amplifying a short fragment of the ITS region using primers ITS2_S2F and ITS2_S3R (Chen et al., 2010). The PCR consisted of 1X QMP Master Mix (Qiagen), forward and reverse primer each at a final concentration of 0.2 μM, 1 μl template DNA and PCR grade water, in a total reaction volume of 12 μL. The amplification profile included an initial denaturation at 95°C for 15 min., followed by 35 cycles of denaturation at 98°C for 60 s, annealing at 58°C for 90 s, and extension at 72°C for 60 s. Prior to sequencing, the PCR products were enzymatically purified with A'SAP PCR clean up kit following the manufacturer's protocol (ArcticZymes). The purified samples were sent to Macrogen Europe for Sanger sequencing.
We retrieved 310 ITS1 sequences from Genbank for 117 available Lupinus species, subspecies or varieties. These sequences were aligned and trimmed in Geneious 2019.2.3 to build an UPGMA tree from pairwise distances based on the HKY substitution model. This tree, together with the phylogeny established by Eastwood et al. (2008), was used to restrict the dataset to 159 samples from 56 more closely related species, subspecies or varieties. To establish the relationships between individuals from these 56 species and our samples, we built a sequence tree in Beast v. 2.4.8 (Bouckaert et al., 2019), based on the birth-only (Yule) model-this choice being motivated by the relatively shallow evolutionary history of the Lupinus genus. This model was parametrized with the HKY substitution model (with 4 gamma-distributed rate classes and a non-null proportion of invariant sites), and a strict molecular clock (due to a small number of informative sites in our sequence data). The MCMC algorithm was run for 100,000,000 generations, sampling every 1,000th state, which was largely sufficient to reach convergence.
Resulting trees were summarized through maximum clade credibility, locating nodes at their common ancestor height in the posterior DNA barcoding revealed that assignment to L. polyphyllus was unclear for five American populations (i.e., ITS1 alone did not allow us to conclude on species identity in our dataset); these populations were therefore omitted from analyses and we were left with 11 populations from the native range (Table 1). The range of between-population distances was 3-1061 km (mean = 441 km) in the USA and 1-441 km (mean = 213 km) in Finland. The native populations generally inhabited lower latitudes and higher altitudes than the introduced populations (Table 1).
To explore whether climate of origin influences plant traits, we obtained data on mean annual temperature, mean temperatures of warmest and coldest quarters, and mean annual precipitation for each population from WorldClim version 2 (Fick & Hijmans, 2017) using the package raster (Hijmans, 2019)

| Trait variability
After air-drying the seeds for about two months at room temperature, we individually weighed 10 fully developed, randomly chosen seeds per mother plant (if possible) to quantify seed mass and stored them in paper bags for later use. Flat, wrinkled seeds were not considered. In mid-January 2019, we chose two seeds from each mother plant for a growing experiment at the Ruissalo Botanical Garden of the University of Turku (lat, long = 60.43, 22.18). To promote germination, we scarified each seed by nicking the seed coat with a scalpel (Beuthin, 2012). The scarified seeds were sown individually into plastic trays (16 × 16 pots of 2 × 2 cm) filled with a commercial potting mix suitable for seedlings (Kekkilän taimimulta). The trays were kept in a greenhouse at 15°C in the daytime and 12°C at night with a photoperiod of 16 hr light and 8 hr dark and were watered when necessary. Two weeks later, seedlings that emerged were replanted into larger plastic pots of 8 × 8 cm (volume of 0.3 L; preferably one seedling per mother plant) filled with a commercial potting mix for garden plants (Kekkilän karkea ruukutusseos) and were kept in the greenhouse. Due to poor seed germination in some of the populations, we repeated seed scarification and seed sowing for two more seeds from 1 to 20 mother plants from 15 populations at the end of January 2019. This time, the scarified seeds were placed on a moist paper towel in petri dishes in the greenhouse for a week, and seedlings were then planted individually in 8 × 8 cm plastic pots.
After excluding three native populations and one introduced population due to poor germination (see Table 1

| Statistical analyses
To ensure that the data on four plant traits considered (seed mass, plant size, flowering probability, and number of flowering shoots) were not spatially structured, we assessed potential spatial autocor- To quantify how much trait variability was explained collectively by country and temperature as compared to other sources of intraspecific variation, we calculated marginal and conditional R 2 values for each trait (using MuMIn::r.squaredGLMM; Bartón, 2019). The former describes the proportion of the total variance explained by the fixed factors (country and temperature), while the latter describes the proportion of the total variance explained by both the fixed and random (i.e., population and mother plant) factors.

| RE SULTS
Three out of four plant traits (plant size, flowering probability, and number of flowering shoots) differed between countries, while seed mass in situ was similar in both groups (Table 2, Figure 1). Under common garden conditions, plants originating from the introduced populations were larger in size than those originating from the native populations; however, after adjusting for initial size differences, their flowering probability and the number of flowering shoots were smaller (Table 2, Figure 1). Native and introduced populations showed a similar response (if any) to temperature of their origin as indicated by the lack of significant interactions between country and temperature for all four traits considered ( Table 2). Temperature of a population's origin explained a significant proportion of the total variation in plant size, with size increasing with increasing mean annual temperature (intercept = 12.873, slope = 0.085 ± 0.209 (SD), Table 2).
Moreover, seed mass tended to decrease with increasing mean annual temperature (intercept = 3.186, slope = −0.021 ± 0.036 (SD), Table 2). Instead, flowering probability and the number of flowering shoots were not associated with temperature of origin ( Table 2).  polyphyllus from the introduced populations were larger in size than those originating from the native populations. This finding is in line TA B L E 2 Results from general and generalized linear mixed models for four traits of the perennial herb Lupinus polyphyllus. Population was used as a random factor in all models, and mother plant was further nested within population in the model of seed mass. df and ddf denote the degrees of freedom in the numerator and in the denominator, respectively with the evolution of increased competitive ability (EICA) hypothesis, which proposes that larger size is selected for due to intense intraspecific competition or reduced herbivore pressure in the new range (Blossey & Nötzold, 1995), that it is preferable to use common gardens in both ranges when comparing plant performance between native and introduced populations (Maron et al., 2004;Moloney et al., 2009). For example, in the perennial herb Hypericum perforatum, plants performed best in common gardens that were located at the same latitudes as their source populations (Maron et al., 2004). However, the location of the common garden seems a less likely explanation for the size difference in the present study because the native populations over-performed the introduced populations in flowering probability and the number of flowering shoots. Finally, the size difference could be due to maternal effects, although this possibility is unlikely because maternal effects tend to be strongest early in life (e.g., Rossiter, 1996) and here plant size was measured after about six months of plant establishment. Regardless of the exact mechanism behind the size difference between the native and introduced populations, large size provides a competitive advantage in resource uptake in the intro- Despite the fact that plants originating from the introduced populations were larger, we observed that their flowering probability and the number of flowering shoots were lower than those of the native plants under common garden conditions. This observation F I G U R E 1 Traits of the perennial Lupinus polyphyllus (back-transformed least square mean ± SE) in native (USA) and introduced (FIN) populations. Seed mass was estimated in situ, while the other traits were estimated under common garden conditions in the first summer. Plant size was measured as height × base diameter in cm. An asterisk denotes a significant difference between countries (p-value < .05) based on a Wald's test contradicts previous findings from perennial herbs that individuals tend to be more fecund in the introduced range than in the native range (e.g., reviewed in Parker et al., 2013, but see Elst et al., 2016) or that fecundity does not differ between ranges (Parker et al., 2013;Sun & Roderick, 2019). The inconsistent findings of the present study could be due to the unbalanced study design (i.e., fewer native populations than introduced populations). However, this explanation seems unlikely, given that we sampled a latitudinal gradient of hundreds of kilometers in both ranges. Alternatively, the lower flowering probability and flowering shoot production of introduced plants may not be adaptive, but could have resulted from nonadaptive changes, such as founder effects (Keller & Taylor, 2008). It is also possible that the differences in flowering were due to different photoperiod adaptation. As the native populations represented lower latitudes (Table 1), they were adapted to somewhat shorter days than the introduced populations (daylength = 14.8 hr at latitude 40 and 18.4 hr at latitude 60 in early July). Exposing individuals from lower latitudes to longer days generally hastens flowering of Lupinus species (Dracup et al., 1998). Overall, our findings should be interpreted with caution because only two fecundity-related traits were measured in the common garden, with no direct estimate of seed production. Plants in the field populations of L. polyphyllus in Finland can be extremely fecund, producing up to hundreds of seeds per inflorescence (Aniszewski et al., 2001;Ramula, 2014). Therefore, we cannot rule out the possibility that plants in the introduced range produce fewer inflorescences per plant but that these are on average taller and contain more seeds than those in the native range. might be expected to exert selective pressure for smaller seed size because plants with lighter seeds can escape predation through better dispersal (Janzen, 1969). On the other hand, seed mass exhibits little variation in some plant species (Harper et al., 1970) and may not necessarily differ in relation to invasion status (e.g., Buckley et al., 2003). For L. polyphyllus, seed mass is indeed remarkably similar across introduced populations in different habitat types, although it does vary among individual plants within populations (Sõber & Ramula, 2013). The present study confirms the existence of intraspecific variation in seed mass that is not explained by country or temperature of a population's origin and suggests that seed mass is similar between native and introduced populations. As germination probability increases with increasing seed mass for the study species (Sõber & Ramula, 2013), native and introduced populations might be expected to exhibit a similar germination rate, given their equal seed mass. However, this was not the case in the present study, in which three out of 11 native populations (27%) had poor establishment as compared to one out of 17 introduced populations (6%) under the greenhouse conditions. We do not have an explanation for this difference in seedling establishment; it could be due to either intrinsic (e.g., differences in seed dormancy) or extrinsic (e.g., different photoperiodic adaptation) factors.

| D ISCUSS I ON
Previous comparisons of plant performance between ranges have emphasized the importance of considering climatic differences among populations (Colautti et al., 2009;Rosche et al., 2019). As predicted, we observed that trait variability was partially explained by temperature of a population's origin, with plant size increasing and seed mass tending to decrease with increasing mean annual F I G U R E 2 Contributions of the fixed (country and temperature) and random (population) factors to trait variability of the perennial Lupinus polyphyllus expressed as marginal and conditional R 2 values, respectively. For seed mass, mother plant nested within population was also included as a random factor, size was estimated as height × base diameter in cm.  (Baker, 1965). However, the plants in the common garden were grown without competitors, and it therefore remains to be tested

ACK N OWLED G EM ENTS
We are extremely grateful to Robin Cristofari for helping with the DNA barcoding analysis, to Jacob Elias for collecting seeds, to the staff of the Ruissalo Botanical Garden for taking care of the plants, and to the Academy of Finland for funding (#319241, #331046 to SR). We also thank the US Forest Service for seed collection permits, Sagehen Creek Field Station for help with coordinating fieldwork and allowing collections on the premises and for three anonymous reviewers for their comments. The Kone Foundation and Finnish Cultural Foundation, V-S regional fund, supported seed collection in the USA (grants to AK).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.