Present address: Department of Biology, Columbus State University, Columbus, GA 31907, USA.
Artificial selection on flowering time: influence on reproductive phenology across natural light environments
Article first published online: 19 MAR 2012
© 2012 The Authors. Journal of Ecology © 2012 British Ecological Society
Journal of Ecology
Volume 100, Issue 4, pages 852–861, July 2012
How to Cite
Galloway, L. F. and Burgess, K. S. (2012), Artificial selection on flowering time: influence on reproductive phenology across natural light environments. Journal of Ecology, 100: 852–861. doi: 10.1111/j.1365-2745.2012.01967.x
- Issue published online: 15 JUN 2012
- Article first published online: 19 MAR 2012
- Received 14 October 2011; accepted 20 February 2012 Handling Editor: Honor C. Prentice
- artificial selection;
- correlated response;
- ecological genetics and ecogenomics;
- flowering time;
- fruit maturation time;
- life history evolution;
- maternal effects;
- phenotypic plasticity;
1. Flowering time is frequently under selection due to a combination of abiotic, biotic and intrinsic factors. Evolution in response to this selection is likely to have broad effects, altering not only flowering time but reproductive phenology and, potentially, traits throughout the life cycle. We know little about the broader phenotypic changes associated with evolutionary shifts in flowering time, and the extent to which expression of these changes depends on local environmental conditions.
2. After three generations of selection for early- and late-flowering, we grew plants of the herb Campanulastrum americanum in contrasting light environments (light gap and understorey) in its home population.
3. Response to selection on flowering time and correlated responses in reproductive phenology were expressed across light environments with the reproduction of early-flowering lines being over 2 weeks ahead of late-flowering lines. Plants in the understorey delayed initiation of flowering but accelerated flower deployment, fruit maturation and the end of reproduction, resulting in a condensed reproductive period.
4. Timing of seed dispersal influences whether offspring grow as annuals or biennials in C. americanum. Because evolution of flowering time shifted reproductive phenology, it is likely to alter life history frequency. In contrast, understorey habitats both delayed flowering and accelerated reproductive phenology, yielding no expected life history change.
5. Synthesis. Evolution of flowering time altered the phenology of all subsequent reproductive traits and is also likely to affect offspring traits. This ripple effect of changes in flowering time indicates that it is essential to recognize genetic and functional linkages among traits to understand potential life cycle consequences of selection on a single character.
Flowering time is central to plant reproductive success. The number of flowers produced (Picó & Retana 2000), the abundance of pollinators (Brody 1997; Rafferty & Ives 2011), environmental conditions during fruit maturation (Giménez-Benavides, Escudero & Iriondo 2007) and levels of fruit herbivory (Pilson 2000; Lacey et al. 2003) may all vary with the timing of flowering. As a consequence of this key role, flowering time is expected to be under strong selection. Selection may be due to abiotic factors such as temperature or water availability (Etterson 2004; Franks, Sim & Weis 2007; Giménez-Benavides, Escudero & Iriondo 2007; Giménez-Benavides et al. 2011), biotic factors such as the presence of pollinators (Elzinga et al. 2007; Kilkenny & Galloway 2008; Sandring & Ågren 2009) or seed predators (Elzinga et al. 2007; Parachnowitsch & Caruso 2008), or intrinsic factors such as plant size (Ollerton & Lack 1998; Colautti & Barrett 2010). These distinct selective forces may have opposing effects on flowering time (e.g. Ehrlen & Munzbergova 2009) and may vary in magnitude and direction over time (Ollerton & Lack 1998). Although we are far from fully understanding selection on flowering time, a recent meta-analysis found consistent phenotypic selection favouring early-flowering, especially in temperate regions, supporting the importance of abiotic factors (Munguía-Rosas et al. 2011).
Flowering time in temperate populations of many taxa is advancing as warmer climates result in earlier growing seasons (e.g. Bradley et al. 1999; Fitter & Fitter 2002; Peñuelas, Filella & Comas 2002; Parmesan & Yohe 2003; Parmesan 2007; Gordo & Sanz 2009). These changes in flowering initiation likely represent plastic responses to warmer temperatures (Gordo & Sanz 2010). To date, little is known about whether warmer climates also result in the selection for earlier reproduction (but see Bradshaw & Holzapfel 2006 for animal studies). Such a pattern of selection would be predicted if observed plastic responses to temperature are adaptive. For example, in Campanulastrum americanum, earlier flowering was found in plants that experienced the earlier spring of an experimentally expanded growing season and this early flowering was favoured by selection (Haggerty & Galloway 2011). In summary, flowering time is frequently under selection, and this selection is expected to intensify as populations respond to the altered environmental conditions associated with warming climates.
Although date of first flower has been the focus of most research, it is only one stage in the seasonal reproductive process that starts with the differentiation of reproductive meristems and finishes with the maturation of the final fruit. A shift in the initiation of flowering in response to selection is likely to influence the expression of subsequent reproductive traits. In particular, the functionally related traits that comprise reproductive timing including seasonal patterns of flower production, fruit maturation and seed dispersal (i.e. phenology) are likely to be phenotypically integrated (e.g. Pigliucci 2003; Edwards & Weinig 2011). In turn, this reproductive phenology can have transgenerational effects, influencing offspring trait expression. Therefore, changes in flowering time are expected to have ripple effects throughout the life cycle (Badyaev & Uller 2009; Donohue 2009). For example, reproductive phenology can determine the abiotic environment during fruit maturation, specifically temperature and photoperiod, and these may influence offspring germination timing, survival and phenotype (e.g. Lacey 1996; Donohue et al. 2005a,b; Walck et al. 2011). Reproductive phenology also determines dispersal timing, which in turn influences offspring germination opportunities and the presence of seed predators (reviewed in Lacey et al. 2003; Donohue 2005; also Lacey & Pace 1983; Picó & Retana 2000; Parachnowitsch & Caruso 2008; Galloway & Burgess 2009). Therefore, evolution of flowering time in response to either the observed wide-scale selection for earlier flowering, the diversity of selection imposed by local biotic and abiotic factors or potential selection associated with warmer climates is likely to have broad effects on a species’ life history. However, we know little about the response of reproductive phenology to evolutionary changes in flowering time.
Artificial selection is a powerful experimental tool to investigate the life cycle consequences of natural selection on a specific trait, such as flowering time (Conner 2003). In particular, if the traits that comprise reproductive phenology are genetically correlated, evolution of one character will alter the expression of the correlated traits. Thus, selection on flowering time may alter the entire reproductive phenology. Experiments have shown that environmental manipulations that affect flowering time, such as warming, frequently shorten, but may also lengthen the reproductive cycle (Sherry et al. 2007; Post et al. 2008; Haggerty & Galloway 2011). This suggests that even if reproductive phenology is not genetically correlated with flowering time, the altered environmental conditions associated with a shift in flowering time may result in plastic changes in reproductive phenology. Furthermore, in natural habitats, many environmental factors influence reproductive timing; for example, light, nutrients and water (reviewed in Kelly & Levin 2000; Forrest & Miller-Rushing 2010; also Dahlgren, von Zeipel & Ehrlen 2007; Jentsch et al. 2009; Crimmins, Crimmins & Bertelsen 2010; Munguía-Rosas et al. 2012). Testing artificial selection lines in natural environments permits a thorough evaluation of potential plastic changes in reproductive phenology associated with altered flowering times. Such plastic changes may reduce or magnify the genetic changes that represent correlated responses to selection. Despite the potential for artificial selection to illuminate the broad consequences of selection on specific traits, its use in natural contexts has been limited.
Here, we use the herbaceous plant C. americanum, selected for early- and late-flowering, to evaluate correlated responses in reproductive phenology and whether their expression depends on local environmental conditions. Campanulastrum americanum’s life history schedule depends on the season of germination. Fall-germinating plants grow as annuals, while spring-germinating seedlings grow as biennials. In this forest-edge species, local light environment is often heterogeneous and influences life history frequency through the effects of the maternal and offspring environments on germination season (Galloway & Etterson 2007). In addition, the timing of maternal flowering and seed maturation influences the frequency of annuals and biennials (Galloway & Burgess 2009).
The association of flowering time and life history schedule may differ between light environments for C. americanum, which typically flowers in the later part of the summer. Early-flowering plants reproduce under warmer conditions than late-flowering plants. Light gaps, which are typically warmer than the understorey (Kilkenny & Galloway 2008), further increase the temperature, potentially accelerating the reproduction of early-flowering plants. Therefore, we expect the evolution of early-flowering to alter reproductive phenology more in light gaps than under the canopy. In contrast, reproductive phenology of late-flowering plants may be more dramatically altered in the understorey where the initiation of flowering is already delayed (Galloway & Etterson 2009), and the combined genetic and environmental effects on flowering time place plants in a cooler part of the growing season. Cooler temperatures may slow reproductive phenology and reduce fecundity. Therefore, to understand potential life cycle consequences of the evolution of flowering time, correlated responses to selection must be studied in ecologically relevant habitats.
To evaluate the interplay between the timing of flowering and local environmental conditions on reproductive phenology, we grew early-flowering, late-flowering and random-flowering C. americanum in understorey and light-gap regions of a natural population. We then evaluated reproductive phenology to address the following questions. (i) Are genetic changes in flowering time expressed under field conditions? (ii) Does the response to selection differ between light-gap and understorey habitats? (iii) Do genetic changes in the initiation of flowering influence subsequent reproductive phenology under natural conditions? (iv) Are patterns of reproductive phenology similar between light-gap and understorey habitats? We then combine these results with those of earlier studies to assess the probability that genetic and environmental influences on flowering time will have transgenerational effects, influencing offspring life history schedule.
Materials and methods
Campanulastrum americanum Small (=Campanula americana L., Campanulaceae) is a predominately outcrossing autotetraploid herb (Gadella 1964; Galloway, Etterson & Hamrick 2003; Galloway & Etterson 2005). The species is widely distributed throughout disturbed areas in eastern North America. Across this broad distribution, populations are typically located in areas with heterogeneous light availability; for example, forest edges, roadsides and stream banks. The study population is located on a steep hillside on Beanfield Mountain, near Mountain Lake Biological Station in south-western Virginia, USA (37′21′′297N, 80′33′′249W). In this population, plants grow both under a deciduous forest canopy and in light gaps caused by tree falls. On average, light gaps receive 10 times more radiation than understorey habitats (Galloway & Etterson 2007).
Campanulastrum americanum may grow as an annual or a biennial. Vegetative rosettes must over-winter prior to flowering. Therefore, season of germination determines an individual’s life history schedule. Seeds that germinate in the fall flower the following summer as winter annuals. In contrast, seeds that delay germination until spring remain a rosette for their first growing season and then flower in their second summer as biennials. Maternal light environment and maternal flowering time influence germination season and therefore offspring life history schedule (Galloway 2002; Galloway & Etterson 2007; Galloway & Burgess 2009).
The influence of maternal flowering time on offspring life history schedule is mediated through the reproductive phenology. Flowering begins in mid-summer and, in the study population, continues until the beginning of autumn. Flowers are tightly clustered in inflorescences located at nodes on the main stem and any side branches. Typically, single flowers are open at a node, with flowers in adjacent nodes open at the same time. Flowers are predominately bumblebee pollinated (Galloway, Cirigliano & Gremski 2002; Kilkenny & Galloway 2008) and typically open for 2 days (Evanhoe & Galloway 2002). When fruits mature, four pores open at the top and seeds disperse passively when plants are jostled. Half of the seeds are dispersed within 5 days of fruit maturation, and almost all seeds are dispersed within 3 weeks (Galloway & Burgess 2009). There is no after-ripening period (Baskin & Baskin 1984); therefore, seeds can germinate immediately following dispersal. Because there is a strong correlation between when a flower is open and when the seeds from that flower mature and disperse (Galloway 2002; Galloway & Burgess 2009), maternal flowering time acts through reproductive phenology to influence offspring life history schedule.
Creation of selection lines: artificial selection on flowering time
To evaluate whether evolution of flowering time may alter offspring life history in C. americanum, three generations of artificial selection for early-flowering, late-flowering and a control, selected at random with respect to flowering time, were conducted under greenhouse conditions (see Burgess, Etterson & Galloway 2007 for details). Artificial selection was initiated using 200 field-collected maternal seed families from the study population. One hundred individuals were assigned to each of two base populations to create independent replicate selection lines. When flowering, the earliest flowering 20% (E), the latest flowering 20% (L) and a random control group (20%, C) in each base population were crossed to produce six selection lines, one replicate of each flowering-time treatment in each base population. Each plant was crossed to three other members of its selected group acting either as a male (pollen donor) in two crosses and a female (pollen recipient) in one cross or as a male once and a female twice, resulting in 30 families per selection line. One hundred plants from these crosses, evenly distributed among families, were grown from each selection line to form the next generation. Following three generations of selection using the same procedure, seeds (fourth generation) were grown in the greenhouse to evaluate the response to selection on flowering time. Plants from the early-flowering selection lines flowered on average 25.4 days before plants in the late-flowering selection lines, while the control lines were intermediate (Burgess, Etterson & Galloway 2007). Seeds from this same generation were planted into their home population for the current study.
Early- and late-flowering selection lines under field conditions
Seeds from the six selection lines were germinated under controlled conditions to remove potential differences in the season of germination that may have arisen as a correlated response to selection. Six seeds from each family in the early (E1 and E2, 28 families each), control (C1 28 families and C2 30 families) and late (L1 and L2, 30 families each) selection lines were individually germinated in the growth chamber in a fully randomized design for a total of 1044 seeds. Seeds were planted in plug trays in a 6 : 1 ratio of Pro-Mix BX (Premier Hort., Quebec, Canada) to fritted clay. Trays were placed in a growth chamber with 21 °C day/12 °C night, 12 h days for 42 days. Then, the temperature was reduced to 5 °C for 7 weeks to stimulate flowering.
Plants were transplanted into light-gap and understorey areas of their home population after snowmelt at the end of March. Thirteen blocks were haphazardly located in each light environment. Five plants from each selection line were planted in randomized locations in each block for a total of 65 plants per line per light environment. Blocks were cleared of other vegetation prior to planting, and rosettes were located 25 cm from each other. To minimize transplant shock, plants were watered after planting. Blocks were surrounded by plastic mesh to prevent deer browsing. Post-transplant mortality reduced the number of plants that reached flowering in each selection treatment and light environment (summed over replicate lines) to 117 early-flowering, 122 control and 114 late-flowering in the understorey and 82 early-flowering, 91 control and 76 late-flowering in the light gaps.
Reproductive phenology was scored on all plants. After bolting, plants were checked daily for the date of first flower. Reproductive phenology was scored weekly on a five-node section of the main stem including the first flowering node and the four nodes immediately above it. Each week, the number of flowers, unripe fruit and mature fruit were recorded. Mature fruit were those that had dehisced and could disperse seeds. On many understorey plants, the five-node segment represented the majority of reproductive nodes, whereas it was closer to a third of the main stem nodes on light-gap individuals. The number of open flowers on each plant was also counted weekly throughout the reproductive season to estimate the seasonal patterns of flower deployment. Total flower production was estimated by the sum of these weekly whole plant counts. Individual flowers are typically open 2 or 3 days (Evanhoe & Galloway 2002); therefore, this total is likely an underestimate of the actual flower production. The end of reproduction, which approximates the end of seed dispersal, was recorded as the date on which almost all fruit in the five-node section had dehisced (≤ 5 immature fruits remaining).
Seasonal patterns of flower production and fruit maturation for each plant were summarized with ‘average flower date’ and ‘average dispersal date’, respectively. Average flower date is a weighted average of the dates flowers were produced throughout the season. It was calculated by weighting each census date by the proportion of an individual’s total flower production that was open on that date and summing over all census dates (cf. Nuismer & Cunningham 2005). Similarly, average seed dispersal date, calculated for the five-node segment on each individual, represents a weighted average of the date when an individual’s fruits matured and began dispersing seeds. Average flower and dispersal dates are influenced by the duration, variation and skew of these phenological processes. While not a substitute for examining seasonal patterns, they permit the distribution to be represented by a single number. Finally, the duration of reproduction was estimated by the number of days between the start of flowering and the end of reproduction.
The response to selection on flowering time and correlated responses in reproductive phenology in the contrasting light environments was evaluated using analysis of variance (Proc MIXED, SAS Institute 2009). Selection treatment, line nested within selection treatment, and light environment were included as fixed effects in the analysis, whereas block was treated as a random effect and nested within environment. The interactions between light environment and both selection treatment and selection line within treatment were also included in the model. Following a significant effect of the selection treatment, Tukey’s multiple comparisons of means were used to determine which treatments differed at α = 0.05. To meet the assumptions of normality, average flower date and average dispersal date were ln-transformed prior to analysis.
Analysis of covariance was used to separate stage-specific effects of the selection treatments and environment on phenological traits from the carry-over of effects expressed earlier in ontogeny. In the analysis, the previous phenological stage was included as a covariate, thereby permitting a test of genetic and environmental effects on the current phenological stage. Specifically, date of first flower was the covariate in the analysis of average flower date, average flower date was the covariate for average dispersal date and average dispersal date was the covariate for end of reproduction. All traits and covariates were ln-transformed providing a uniform scale for variables. We inspected the data for each covariate (previous phenological stage) to ensure that there was an overlapping distribution among treatments, despite differences in means. On average, there was an 83% overlap in distribution of the covariates among treatments (73% average flower date, 82% average dispersal date and 95% end of reproduction) indicating that it was appropriate to use the previous phenological stage as a covariate, although it did vary among treatments. Structured equation modelling is a more traditional analytical tool to address stage-specific effects. However, the sequential nature of phenology precludes developing the branched model required for this approach. Duration of reproduction was not included in this analysis as it encompasses all post-flowering traits.
Response to selection on flowering time was expressed under field conditions. Plants in the late-flowering selection lines initiated flowering on average 15 days after those in the early-flowering selection lines. The unselected controls were in between, flowering on average 5 days after the early lines and 10 days before the late lines (Table 1A, Fig. 1a).
|Source||d.f.||Flowering start||Average flower date||Average dispersal date||End of reproduction||Duration of reproduction|
Light environment had a strong influence on flowering time with plants in the understorey flowering on average 8 days after those in light gaps (Table 1A, Fig. 1a). However, there was no difference in the expression of response to selection between light environments (Table 1A). On average, late-flowering lines flowered 15 days and control lines 5 days, after the early-flowering lines regardless of whether plants were grown in light gaps or understorey habitats (Fig. 1a).
The differences between the early- and late-flowering selection lines were also found in later reproductive traits. Average flower date of the late-flowering selection lines was 17 days later, average dispersal date was 16 days later and the end of reproduction was 17 days later than the early-flowering selection lines (Table 1A, Fig. 1b–d). Similarly, average flower date, average dispersal date and the end of reproduction of the control lines were 5 or 6 days later than the early-flowering selection lines and 11 days before the late-flowering selection lines. When the previous phenological stage was included as a covariate in the analyses, none of the differences between the selection treatments remained significant (Table 1B). Therefore, differences in reproductive phenology between the early- and late-flowering selection lines can be attributed to the altered initiation of reproduction.
Reproductive phenology was more similar than flowering date between light-gap and understorey habitats. Although plants grown in the understorey initiated flowering 8 days after those in light gaps, there was no difference in average flower date between the two light environments (Table 1A, Fig. 1b). Average dispersal date showed a similar pattern of no difference in timing between the light environments with the exception of early-flowering selection lines that dispersed their seeds earlier on average in light gap than understorey habitats (Table 1A: light*selection, Fig. 1c). The end of reproduction, in contrast, was slightly later in the light gaps than in the understorey (Table 1A, Fig. 1d). This 6-day delay in the end of reproduction in the light-gap environments was found across the flowering-time treatments.
Environmental effects on reproductive timing were stronger after accounting for earlier phenological traits (Table 1B). For example, after adjusting for differences in the initiation of flowering, the average flower date was 6 days later in light gaps than in the understorey, suggesting that flowers are deployed over a broader window of time in light gaps (Fig. 2a,b). Similarly, after accounting for differences in average dispersal date, reproduction finished more slowly in the light gaps than the understorey (Fig. 2c,d). In contrast, after accounting for differences in average flowering date, average dispersal date was similar between light environments for control and late-flowering selection lines, but for the early-flowering selection lines was later for understorey than light-gap plants.
Although shifts in average flower date among the three selection treatments were similar to that of initiation of flowering, patterns of floral deployment differed somewhat among the selection treatments (Fig. 2a,b). Early-flowering plants had a broad early peak; in the light gap, two-thirds of the season’s flowers were produced in that peak. In contrast, flower production in the control lines grown in light gaps had a discrete peak where 30% of the season’s flowers were open on a single day. Finally, in the late-flowering selection lines, peak flowering included a smaller proportion of the season’s flowers with a greater proportion of flowers produced on plants that had been flowering longer.
There was a greater difference in seed dispersal patterns than seasonal flowering phenology between the light environments. Seasonal patterns of flower deployment were similar between light-gap and understorey habitats with early-flowering selection lines having a broader early peak and all three flowering treatments having a synchronized but smaller later peak (Fig. 2a,b). In contrast, while the differences between the selection lines were comparable across both light environments for seed dispersal, the temporal pattern differed. Mature fruit began dispersing seed over a longer time period in light gaps than in the understorey (Fig. 2c,d). For example, in the understorey early-flowering, control and late-flowering selection lines had 43%, 42% and 31% of their fruit ripen in 1 weekly census interval. In contrast, in the light gaps, the interval with the peak fruit ripening had only 20% of the fruits initiate seed dispersal, regardless of selection line. A condensed seed dispersal period resulted in an earlier end to reproduction in the understorey (Fig. 1d). In total, with a later start and a more rapid finish, the reproductive phenology was more condensed in understorey than light-gap habitats (Table 1A, Fig. 3). Although the duration of reproduction was longer in light gaps, the difference between light environments was greater for the late-flowering selection lines (18 days) than for the early-flowering or control lines (11 and 13 days, respectively; Table 1A and Fig. 3).
Response to artificial selection on flowering date was expressed under natural environmental conditions in C. americanum. Three generations of selection for earlier and later flowering in controlled, greenhouse conditions resulted in a divergence of 25 days in greenhouse-grown plants (Burgess, Etterson & Galloway 2007). When plants from the same generation were planted into the natural habitat of the source population, late-flowering selection lines initiated flowering on average 15 days after the early-flowering selection lines. This difference between selection lines was found regardless of whether plants were grown in understorey or light-gap regions of the natural population, despite a tenfold difference in light availability. In addition, response of flowering time to selection was similar between the independent lines within each treatment. A 15-day difference in flowering time between the selection treatments is substantial in this species where the mean initiation of flowering among years within a light environment varies by less than a week (Galloway 2002; Galloway & Etterson 2009; Lin & Galloway 2010). In contrast, flowering time may be substantially more variable in other taxa (e.g. Ollerton & Diaz 1999; Tarayre et al. 2007; Kudo et al. 2008).
Consistency in the expression of flowering time across environments was not necessarily expected because the expression of quantitative traits is environment-dependent. For example, controlled-environment results from artificial selection for temperature tolerance in Drosophila were found only half of the time in natural conditions (Kristensen, Volker & Hoffmann 2007). For flowering time, where the molecular basis of the trait is fairly well understood, studies have found different QTLs influence trait expression between experimental environments (i.e. day-length Weinig et al. 2002; Blackman et al. 2011) and between controlled and field environments (Weinig et al. 2002; Anderson, Lee & Mitchell-Olds 2011). The altered genetic architecture suggests trait expression will differ across environments. As an example, there was a low correlation between greenhouse and field expression of flowering time in Boechera (average R = 0.22, Anderson, Lee & Mitchell-Olds 2011). In contrast, the relatively consistent expression of selection response in field-grown C. americanum suggests limited genotype-by-environment interaction for genes that determine flowering time under a range of natural conditions.
Selection on flowering time altered the phenology of all subsequent reproductive traits. Correlated responses to the selection on flowering time were found for average flower date, average dispersal date and the end of reproduction. All these traits shifted as a consequence of selection at least as much as flowering date itself, suggesting that they are tightly correlated with the timing of flowering (Falconer & Mackay 1996). The tight correlation was further supported by the similarity among selection treatments when the previous phenological trait was included as a covariate in the analysis. This result indicates that there were no stage-specific effects of the selection treatments on phenology beyond altering flowering time; all differences in reproductive phenology between the selection treatments represent the carry-over of effects expressed earlier in ontogeny. In other species, it is typically not known whether phenological changes associated with a shift in flowering time represent independent responses to environmental conditions or the carry-over of earlier effects (e.g. Lacey et al. 2003; Sherry et al. 2007; Post et al. 2008). A previous study that altered flowering time in C. americanum through environmental manipulation also found reproductive phenology showed strong phenotypic integration (Galloway & Burgess 2009). However, that integration was environmentally based, whereas the strong phenotypic integration in these artificial selection lines likely has both genetic and environmental components. Regardless, the strong integration of reproductive phenology suggests that selection on flowering time in natural populations will alter the timing of all subsequent stages of reproduction.
Selection lines planted in the forest understorey flowered on average 8 days after those in light gaps. This result mirrors previous studies in C. americanum (Galloway & Etterson 2009; Lin & Galloway 2010) and other herbaceous taxa (e.g. Sultan 2001; Steinger, Roy & Stanton 2003), although the reverse, earlier flowering with increased shade is a common shade-avoidance response (reviewed in Smith & Whitelam 1997; also Donohue et al. 2001; Callahan & Pigliucci 2002). However, differences in timing across light environments were not found for the later traits, average flower date and average dispersal date (two of the three selection treatments), despite the tight correlation in reproductive phenology found among selection treatments. Even more so, when the previous phenological stage was accounted for, later phenological traits were reached at a relatively earlier time in the understorey. Therefore, although average flower date was the same between understorey and light-gap habitats, later initiation of flowering for understorey plants resulted in a shorter time interval between phenological stages than plants in light gaps. The same pattern was found without accounting for earlier phenology in the earlier end of reproduction in the understorey relative to light gaps. Over the time span of reproduction, the cumulative effect of these differences in phenology across light environments was a 29% longer reproductive period in light gaps than in the understorey, a pattern comparable to typical shade-avoidance responses (Smith & Whitelam 1997). These differences, in part, are driven by divergence in the size of plants in the two light environments. When plant size was included as a covariate, differences between the light environments were reduced (results not shown), indicating the smaller plants, such as those found in the understorey, have a shorter reproductive period (e.g. Ollerton & Lack 1998).
Although flowering was uniformly delayed in all selection lines in the understorey, later reproductive traits had modest differences in response to the light environment between the flowering time treatments. For example, seed dispersal date was earlier in light gaps than the understorey for the early-flowering selection lines but not the other lines, and late-flowering lines lengthened their reproductive duration in light gaps slightly more than early-flowering or control lines. However, these differences in response to the light environment between the selection lines were modest. Reproductive phenology was most strongly influenced by genetic differences that were expressed across environments and responses to the light environments that were expressed across all selection lines. Similarly, in Brassica rapa, selection due to natural drought resulted in earlier flowering but no difference in plasticity in response to experimental drought (Franks 2011).
Altering reproductive phenology may influence offspring life history schedule in C. americanum. Because fall-germinating seeds grow as annuals, while spring germinants have a biennial life cycle, shifting the reproductive phenology may change life history frequency. Earlier work in this population, simulating an expanded seed dispersal window, found that seeds planted early, at the end of the summer, were more likely to germinate as annuals, while seeds planted later, in mid-autumn, were more likely to germinate as biennials (Galloway & Burgess 2009). The average dispersal date measured here is a summary statistic that represents the dispersal time of the average fruit on the main stem of each plant. The average dispersal date of the late-flowering selection lines was over 2 weeks later than the early-flowering selection lines. Therefore, on average, more annual offspring are expected from the early-flowering selection lines and more biennial offspring from the late-flowering selection lines. The extent to which a 2-week shift in dispersal time would be expected to alter offspring life history depends on the light environment. Germination frequency of annuals in the understorey declines linearly with planting date until the first week in September (Galloway & Burgess 2009). After that, the germination frequency of annuals is small through September and near-zero for later planted seeds. In the current study, by the first week of September, only early-flowering selection lines had a non-negligible proportion of fruit that were dispersing seeds, suggesting that only these lines would be likely to produce annual offspring in the understorey. Although less information is available about the relationship between dispersal time and germination season for seeds dispersed in light gaps, germination of annuals is more common in light gaps than in the understorey (Galloway & Etterson 2007), and more annuals germinate from early planted seeds in these gaps (L. F. Galloway & K. S. Burgess, unpubl. data). Therefore, we again expect a greater frequency of annuals from the early-flowering selection lines that ripened their fruits before the late-flowering lines.
In contrast to the genetic differences in flowering time, our results suggest that light environment is not expected to influence offspring life history schedule. Seed dispersal time was the same across light environments for control, late-flowering and one of the early-flowering selection lines, indicating that the understorey light habitat, while delaying the initiation of reproduction, does not similarly delay seed dispersal. Indeed, estimating life cycle or transgenerational influences of the light environment from flowering date alone would be inaccurate. In total, although annuals are more common in light gaps and biennials in the understorey (Galloway & Etterson 2007), results presented here suggest these differences are due to the germination environment and not to the timing of seed dispersal.
Evolution of flowering time in response to selection, such as the pervasive selection for earlier flowering in temperate populations (Munguía-Rosas et al. 2011), requires genetic variation. In C. americanum, there is substantial genetic variation for flowering time in natural populations. Following three generations of selection where 20% of the population contributed to the next generation, flowering time of the early and late selection lines was separated by over 2 weeks under field conditions. Other studies that have found strong response to selection on flowering time have often started with composite populations representing a diverse genetic pool (e.g. Campbell et al. 2009; Van Dijk 2009). It is noteworthy that in C. americanum, within a single natural population, there was sufficient genetic variation to result in substantial, ecologically relevant, evolutionary change following only three generations of selection. The estimated heritability (0.44; Galloway, Etterson & McGlothlin 2009) and realized heritability (0.31 early-flowering, 0.23 late-flowering; Burgess, Etterson & Galloway 2007) of flowering time in C. americanum are similar to the estimates in other taxa (0.38; Geber & Griffen 2003). This similarity suggests that phenotypic selection, such as that imposed by warmer climates, will likely alter flowering time in C. americanum and a range of other plant taxa.
Evolution of flowering time is likely to alter traits throughout the life cycle, in part by creating transgenerational effects. The extent to which changes in flowering time have cascading effects throughout the life cycle depends on whether correlated responses to selection expressed in subsequent phenological traits maintain, reduce or enhance the changes in flowering time. In addition, expression of life cycle effects also depends on whether plasticity to local conditions, such as the light environment, enhances or reduces the shift in flowering time. Between-generation consequences of flowering time evolution represent indirect genetic effects (Wolf et al. 1998), and evolutionary change in a trait that itself is not under direct selection. The combination of between-generation indirect genetic effects with within-generation genetic correlations results in the potential for selection on a single trait to alter the expression of a suite of other traits. Similar patterns of evolution are equally likely in animals indicating the broad relevance of these ideas; for example, life history trade-offs (Roff 2002) or the influence of maternal provisioning or body size on offspring traits (Mousseau & Fox 1998; Räsänen & Kruuk 2007). In summary, the presence of widespread linkages among traits must be recognized in order to understand the potential shifts in phenotype in response to selection.
We thank C. Ballare, R. Gittman, B. Haggerty, L. Lee, J. Painter, D. Stone, H. Truong and T. Kugler for assistance; Mountain Lake Biological Station for logistical support; the Galloway laboratory for comments on the manuscript; and NSF DEB-0316298, NSF DEB-1020717 and a UVA-FEST award to LFG for financial support.
- 2011) Life-history QTLs and natural selection on flowering time in Boechera stricta, a perennial relative of Arabidopsis. Evolution, 65, 771–787. , & (
- 2009) Parental effects in ecology and evolution: mechanisms, processes, and implications. Philosophical Transactions of the Royal Society B, Biological Sciences, 364, 1169–1177. & (
- 1984) The ecological life-cycle of Campanula americana in Northcentral Kentucky. Bulletin of the Torrey Botanical Club, 111, 329–337. & (
- 2011) Contributions of flowering time genes to sunflower domestication and improvement. Genetics, 187, 271–287. , , , , , , & (
- 1999) Phenological changes reflect climate change in Wisconsin. Proceedings of the National Academy of Sciences USA, 96, 9701–9704. , , & (
- 2006) Climate change - Evolutionary response to rapid climate change. Science, 312, 1477–1478. & (
- 1997) Effects of pollinators, herbivores, and seed predators on flowering phenology. Ecology, 78, 1624–1631. (
- 2007) Artificial selection shifts flowering phenology and other correlated traits in an autotetraploid herb. Heredity, 99, 641–648. , & (
- 2002) Shade-induced plasticity and its ecological significance in wild populations of Arabidopsis thaliana. Ecology, 83, 1965–1980. & (
- 2009) Rapid evolution in crop-weed hybrids under artificial selection for divergent life histories. Evolutionary Applications, 2, 172–186. , , & (
- 2010) Natural selection and genetic constraints on flowering phenology in an invasive plant. International Journal of Plant Sciences, 171, 960–971. & (
- 2003) Artificial selection: a powerful tool for ecologists. Ecology, 84, 1650–1660. (
- 2010) Complex responses to climate drivers in onset of spring flowering across a semi-arid elevation gradient. Journal of Ecology, 98, 1042–1051. , & (
- 2007) Variation in vegetative and flowering phenology in a forest herb caused by environmental heterogeneity. American Journal of Botany, 94, 1570–1576. , & (
- 2005) Niche construction through phenological plasticity: life history dynamics and ecological consequences. New Phytologist, 166, 83–92. (
- 2009) Completing the cycle: maternal effects as the missing link in plant life histories. Philosophical Transactions of the Royal Society B, Biological Sciences, 364, 1059–1074. (
- 2001) Adaptive divergence in plasticity in natural populations of Impatiens capensis and its consequences for performance in novel habitats. Evolution, 55, 692–702. , , , & (
- 2005a) Environmental and genetic influences on the germination of Arabidopsis thaliana in the field. Evolution, 59, 740–757. , , , , , & (
- 2005b) The evolutionary ecology of seed germination of Arabidopsis thaliana: variable natural selection on germination timing. Evolution, 59, 758–770. , , , , , & (
- 2011) The quantitative-genetic and QTL architecture of trait integration and modularity in Brassica rapa across simulated seasonal settings. Heredity, 106, 661–677. & (
- 2009) Timing of flowering: opposed selection on different fitness components and trait covariation. The American Naturalist, 173, 819–830. & (
- 2007) Time after time: flowering phenology and biotic interactions. Trends in Ecology & Evolution, 22, 432–439. , , , , & (
- 2004) Evolutionary potential of Chamaecrista fasciculata in relation to climate change. 1. Clinal patterns of selection along an environmental gradient in the Great Plains. Evolution, 58, 1446–1458. (
- 2002) Floral longevity in Campanula americana (Campanulaceae): a comparison of morphological and functional gender phases. American Journal of Botany, 89, 587–591. & (
- 1996) Introduction to Quantitative Genetics, 4th edn. Longman, UK. & (
- 2002) Rapid changes in flowering time in British plants. Science, 296, 1689–1691. & (
- 2010) Toward a synthetic understanding of the role of phenology in ecology and evolution. Philosophical Transactions of the Royal Society B, Biological Sciences, 365, 3101–3112. & (
- 2011) Plasticity and evolution in drought avoidance and escape in the annual plant Brassica rapa. New Phytologist, 190, 249–257. (
- 2007) Rapid evolution of flowering time by annual plant in response to a climate fluctuation. Proceedings of the National Academy of Sciences, 104, 1278–1282. , & (
- 1964) Cytotaxonomic studies in the genus Campanula. Wentia, 11, 1–104. (
- 2002) The effect of maternal phenology on offspring life history in the herbaceous plant Campanula americana. Journal of Ecology, 90, 851–858. (
- 2009) Manipulation of flowering time: phenological integration and maternal effects. Ecology, 90, 2139–2148. & (
- 2002) The contribution of display size and dichogamy to potential geitonogamy in Campanula americana. International Journal of Plant Sciences, 163, 133–139. , & (
- 2005) Population differentiation and hybrid success in Campanula americana: geography and genome size. Journal of Evolutionary Biology, 18, 81–89. & (
- 2007) Transgenerational plasticity is adaptive in the wild. Science, 318, 1134–1136. & (
- 2009) Plasticity to canopy shade in a monocarpic herb: within and between generation effects. New Phytologist, 182, 1003–1012. & (
- 2003) Outcrossing rates and inbreeding depression in the herbaceous autotetraploid Campanula americana. Heredity, 90, 308–315. , & (
- 2009) The contribution of direct and maternal genetic effects to life history evolution. New Phytologist, 183, 826–838. , & (
- 2003) Inheritance and natural selection on functional traits. International Journal of Plant Sciences, 164, S21–S42. & (
- 2007) Reproductive limits of a late-flowering high-mountain Mediterranean plant along an elevational climate gradient. New Phytologist, 173, 367–382. , & (
- 2011) Selection on flowering time in Mediterranean high-mountain plants under global warming. Evolutionary Ecology, 25, 777–794. , , & (
- 2009) Long-term temporal changes of plant phenology in the Western Mediterranean. Global Change Biology, 15, 1930–1948. & (
- 2010) Impact of climate change on plant phenology in Mediterranean ecosystems. Global Change Biology, 16, 1082–1106. & (
- 2011) Response of individual components of reproductive phenology to growing season length in a monocarpic herb. Journal of Ecology, 99, 242–253. & (
- 2009) Beyond gradual warming: extreme weather events alter flower phenology of European grassland and heath species. Global Change Biology, 15, 837–849. , , & (
- 2000) Direction selection on initial flowering date in Phlox drummondii (Polemoniaceae). American Journal of Botany, 87, 382–391. & (
- 2008) Reproductive success in varying light environments: direct and indirect effects of light on plants and pollinators. Oecologia, 155, 247–255. & (
- 2007) Can artificially selected phenotypes influence a component of field fitness? Thermal selection and fly performance under thermal extremes. Proceeding of the Royal Society B: Biological Sciences, 274, 771–778. , & (
- 2008) Linkages between phenology, pollination, photosynthesis, and reproduction in deciduous forest understory plants. Ecology, 89, 321–331. , & (
- 1996) Parental effects in Plantago lanceolata. I. A growth chamber experiment to examine pre- and postzygotic temperature effects. Evolution, 50, 865–878. (
- 1983) Effect of parental flowering and dispersal times on offspring fate in Daucus carota (Apiaceae). Oecologia, 60, 274–278. & (
- 2003) Multigenerational effects of flowering and fruiting phenology in Plantago lanceolata. Ecology, 84, 2462–2475. , , , & (
- 2010) Environmental context determines within- and between- generation consequences of herbivory. Oecologia, 163, 911–920. & (
- 1998) Maternal Effects as Adaptations. Oxford University Press, Oxford, UK. & (
- 2011) Meta-analysis of phenotypic selection on flowering phenology suggests that early flowering plants are favoured. Ecology Letters, 14, 511–521. , , & (
- 2012) Environmental control of reproductive phenology and the effect of pollen supplementation on resource allocation in the cleistogamous weed, Ruellia nudiflora (Acanthaceae). Annals of Botany, 109, 343–350. , , & (
- 2005) Selection for phenotypic divergence between diploid and autotetraploid Heuchera grossulariifolia. Evolution, 59, 1928–1935. & (
- 1999) Evidence for stabilising selection acting on flowering time in Arum maculatum (Araceae): the influence of phylogeny on adaptation. Oecologia, 119, 340–348. & (
- 1998) Relationships between flowering phenology, plant size and reproductive success in Lotus corniculatus (Fabaceae). Plant Ecology, 139, 35–47. & (
- 2008) Predispersal seed herbivores, not pollinators, exert selection on floral traits via female fitness. Ecology, 89, 1802–1810. & (
- 2007) Influences of species, latitudes and methodologies on estimates of phenological response to global warming. Global Change Biology, 13, 1860–1872. (
- 2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature, 421, 37–42. & (
- 2002) Changed plant and animal life cycles from 1952 to 2000 in the Mediterranean region. Global Change Biology, 8, 531–544. , & (
- 2000) Temporal variation in the female components of reproductive success over the extended flowering season of a Mediterranean perennial herb. Oikos, 89, 485–492. & (
- 2003) Phenotypic integration: studying the ecology and evolution of complex phenotypes. Ecology Letters, 6, 265–272. (
- 2000) Herbivory and natural selection on flowering phenology in wild sunflower, Helianthus annuus. Oecologia, 122, 72–82. (
- 2008) Phenological sequences reveal aggregate life history response to climatic warming. Ecology, 89, 363–370. , , & (
- 2011) Effects of experimental shifts in flowering phenology on plant-pollinator interactions. Ecology Letters, 14, 69–74. & (
- 2007) Maternal effects and evolution at ecological time-scales. Functional Ecology, 21, 408–421. & (
- 2002) Life History Evolution. Sinauer Associates, Sunderland, Massachusetts, USA. (
- 2009) Pollinator-mediated selection on floral display and flowering time in the perennial herb Arabidopsis lyrata. Evolution, 63, 1292–1300. & (
- SAS Institute. (2009) SAS/STAT Version 9.2. SAS Institute, Cary, North Carolina, USA.
- 2007) Divergence of reproductive phenology under climate warming. Proceedings of the National Academy of Sciences USA, 104, 198–202. , , , , , , & (
- 1997) The shade avoidance syndrome: multiple responses mediated by multiple phytochromes. Plant Cell Environment, 20, 840–844. & (
- 2003) Evolution in stressful environments II: adaptive value and costs of plasticity in response to low light in Sinapis arvensis. Journal of Evolutionary Biology, 16, 313–323. , & (
- 2001) Phenotypic plasticity for fitness components in Polygonum species of contrasting ecological breadth. Ecology, 82, 328–343. (
- 2007) Flowering phenology of Ulex europaeus: ecological consequences of variation within and among populations. Evolutionary Ecology, 21, 395–409. , , , & (
- 2009) Evolutionary change in flowering phenology in the iteroparous herb Beta vulgaris ssp maritima: a search for the underlying mechanisms. Journal of Experimental Botany, 60, 3143–3155. (
- 2011) Climate change and plant regeneration from seeds. Global Change Biology, 17, 2145–2161. , , , & (
- 2002) Novel loci control variation in reproductive timing in Arabidopsis thaliana in natural environments. Genetics, 162, 1875–1884. , , , , , , , & (
- 1998) Evolutionary consequences of indirect genetic effects. Trends in Ecology and Evolution, 13, 64–69. , , , & (