Introduction

In seasonal environments, timing of events such as emergence, growth, reproduction and seasonal senescence is often temporally and developmentally correlated (Rathcke and Lacey 1985, Kelly 1992, O’Neil 1997, Aizen 2003, Sola and Ehrlén 2007, Keenan and Richardson 2015). Such correlations imply that selection might only rarely be exerted independently on the timing of a single event. Instead, selection on a focal phenological trait is often a combination of direct selection, and indirect selection acting via other phenological traits (Rathcke and Lacey 1985, Kelly 1992, Ehrlén 2015, Galloway et al. 2018). Selection could also target the relative timing of phenological traits. Such selection has been found for development time to maturity and reproduction in both plants and animals (Kingsolver and Pfennig 2004). Lastly, the strength or direction of selection on a focal trait might often depend on the level of other phenological traits (correlational selection; Lande and Arnold 1983, Phillips and Arnold 1989, Kelly 1992). Taken together, this implies that to accurately estimate selection acting on a focal phenological trait, it is necessary to account for indirect selection via temporally correlated traits, as well as for correlational selection.
In temperate plants, the timing of reproduction is a particularly important life-history trait, as it influences interactions with the biotic and abiotic environment and is often strongly linked to fitness (e.g. Elzinga et al. 2007, Ehrlén 2015, Austen et al. 2017). Phenotypic selection has been found to favour early flowering in many temperate plant species (Harder and Johnson 2009, Munguía-Rosas et al. 2011), and the consistency of this pattern has raised questions as to why plants do not appear to respond to this selection (Austen et al. 2017). One suggested explanation for the observed pattern is that selection on flowering time is exerted indirectly via correlated life-history traits and that this indirect selection is not accounted for in analyses (cf. Rathcke and Lacey 1985, Austen et al. 2017). For example, several studies have documented correlations between flowering time and vegetative phenology (e.g. Kelly 1992, Diggle 1999, Yao and Mehlenbacher 2000, Sola and Ehrlén 2007, Brachi et al. 2012). If vegetative and reproductive spring phenology are correlated, evolution of flowering time is influenced also by selection on timing of vegetative phenology, and selection will act simultaneously on both traits. In these cases, we expect direct and indirect selection on both traits, as well as on their relationship. Yet, very few studies have simultaneously examined selection on vegetative and reproductive phenology, and quantified direct, indirect and correlational selection (but see Kelly 1992).
In this study, we explored the relationship between vegetative spring phenology and flowering phenology in a natural population of the perennial understorey herb Lathyrus vernus in Sweden, and used multiple regressions to estimate phenotypic selection on these two traits. Previous studies with this system have found that selection favours earlier flowering in most years (Ehrlén and Münzbergová 2009, Ehrlén and Valdés 2020), but also that flowering time is correlated with start of shoot growth and leaf development (Sola and Ehrlén 2007). It is thus possible that some of the observed selection on flowering time is exerted indirectly, via selection on vegetative phenology. We expected that most of the among-year differences in the distributions of phenological traits in L. vernus are driven by climatic conditions (Ehrlén and Valdés 2020), while within-year differences among individuals in these traits are caused by other factors, including genetic differences. We also expected the causes of selection to both depend on the density and frequency of phenological trait values within years (soft selection, e.g. in terms of among-individual competition for pollinators, and avoidance of seed-predation and vertebrate grazing), and to be density-independent (hard selection, e.g. in terms of effects of weather conditions) (cf. De Lisle and Svensson 2017). To assess selection on vegetative and reproductive spring phenology, we monitored leaf-out day and first flowering day, and recorded individual fitness in terms of seed production, during three years. We addressed four specific questions: 1) How closely are vegetative and reproductive phenology correlated? 2) Is there phenotypic selection acting on vegetative and reproductive phenology? 3) If so, to what extent is selection on vegetative and reproductive phenology direct vs. indirect? 4) Is there selection on the relationship between vegetative and reproductive phenology, in terms of selection for time of development between leaf-out and flowering or correlational selection?