Modifications in the timing of life-history events can alter the biotic and abiotic environments experienced during an organism's lifetime. In plants, germination timing plays a critical role in relation to seasonal environmental conditions, pollinator availability, competitive dynamics, etc. Individuals can compensate for a change in timing of germination by modifying their growth and flowering time. However, biotic and abiotic factors may affect these compensatory responses. In this study we assess how biotic and abiotic differences due to planting date influence the timing of flowering, survival, and reproduction. To do this we manipulated germination timing of the annual California Goldfields (Lasthenia californica DC. ex Lindl., Asteraceae) in a serpentine grassland in northern California, by seeding three times during the growing season (November, January, and March). Neighbor removal plots were compared with control plots to examine influences of close neighbors and seasonal priority effects (due to early individuals pre-empting resources) on flowering time and duration, growth, and reproductive success (survival to reproduction and inflorescence production). Both planting date and neighbor removal treatments significantly impacted flowering time and duration, growth, and reproduction in this species. Later planting dates did delay flowering time, but this delay was minimal as flowering time was constrained within set biotic and abiotic boundaries. In addition, we find that a mixture of planting times and levels of neighbor removal can extend the duration of flowering on the landscape. Inflorescence production and survival declined with later planting dates, but neighbor removal counteracted this decline. We find that L. californica exhibits graded growth allocation, as well as plasticity in flowering time in response to planting date. This study has implications for the timing of restoration projects, as planting time influences both the timing of flowering as well as the overall reproductive success of planted individuals. Our results suggest that practitioners should aim to plant earlier in the season, but that neighbor removal may counteract some of the costs of late planting.
Growth and reproductive timing are critical for individual success. This is especially true for annual plants, as an individual's lifetime reproductive output is dependent on the conditions experienced over a single growing season (Cohen 1976; Schmitt 1983). In annual species, germination timing dictates the biotic and abiotic conditions experienced throughout the lifetime of an individual, and therefore shapes an individual's fitness. Plants go through an initial juvenile phase of vegetative growth to acquire sufficient carbon reserves before flowering (Simpson and Dean 2002). Therefore, germination timing also influences reproductive timing, with implications for population dynamics and community interactions.
Germination timing may be optimized in different ways during the growing season. Earlier germinating individuals benefit from a longer growing season and early access to resources (Lortie and Turkington 2002; Wainwright et al. 2012). Later germinating individuals may benefit from avoiding unfavorable conditions early in the season (e.g., frost) (Petrů et al. 2006; Donohue et al. 2010; Mercer et al. 2011), but may experience harsh end of season conditions, such as early frosts or, in Mediterranean-type climates, drought and heat extremes. Therefore, a population may benefit from staggered timing of germination across the season, which may also result in both inter- and intraspecific temporal resource partitioning (Dyer et al. 2008; Orrock and Christopher 2010; Leverett et al. 2018).
Plant reproductive timing is initiated as a response to abiotic factors in the environment, and flowering can be triggered by temperature, moisture and/or photoperiod cues, interacting with plant size and internal resource states (Rathcke and Lacey 1985). With sufficient resource availability, an annual plant will initiate reproductive growth after the abiotic cue(s) triggering flowering are experienced. Because nearby plants modify the temperature, moisture, and light in the surrounding environment, as well as available resources (Rathcke and Lacey 1985; Simpson and Dean 2002), inter- and intraspecific plant-plant interactions (e.g., competition and facilitation) also play an important role in the timing of reproduction and can have strong effects on the evolution of life history timing (Ellner 1987; Metcalf et al. 2015; Leverett 2017).
In grasslands of Mediterranean-type climates, season duration is relatively unpredictable. In northern California, precipitation is highly variable, and expected to become more so as the climate changes (Swain et al. 2018). Most annual plant germination in California grasslands occurs after rainfall events of at least 15 mm (Heady 1977). Rainfall in these systems may come as early as October or as late as March, and the onset of the summer dry season is variable. For this reason, species growth and reproductive timing are often dependent on precipitation timing, sometimes in complex and dynamic ways (e.g., Pearson et al. 2021). In locations like this, with unpredictable season duration, annual plants are expected to exhibit a graded allocation strategy, in which both vegetative and reproductive growth occur simultaneously during an intermediate period between purely vegetative and purely reproductive growth (King and Roughgarden 1982). This graded allocation strategy allows for bet-hedging and is predicted to yield optimal reproductive allocation when season duration is unpredictable (Wong and Ackerly 2005). Based on this theory, we expect that plants in Mediterranean-type grasslands will exhibit a graded allocation strategy, and a capacity to respond adaptively to variation in germination or planting dates.
In environments with physical stresses that limit plant growth, such as systems with ultra-mafic soils (e.g., serpentine), the importance of competition vs. facilitation in neighboring plants can vary. Competition may be strong between species because resources are scarce (Tilman 1988), but facilitation may also play a role if neighboring vegetation can improve survival (Bertness and Callaway 1994). Because endemics to low-fertility soils are often poor competitors, these systems are generally thought to be refuges from competition (Tansley 1917; Sharitz and McCormick 1973; Kruckeberg 1984). Facilitation is predicted to be stronger in harsh environments, but the balance of competition and facilitation may vary over time, as seen in more productive systems (Leverett 2017). Serpentine soil systems are characterized by high variation in their soils, plant density and productivity, and therefore the influence of species interactions may also vary across microsites (Moore and Elmendorf 2011).
The timing and duration of flowering is also important to pollinators. Insects that rely on nectar and pollen may be especially dependent on the timing of early season flowering species as these are the first floral resources available. Bees and other pollinators require pollen and nectar resources throughout their flight and nesting seasons, and longer flowering seasons support robust bee populations (Russo et al. 2013). The timing of plant and insect life history events have both been shifting in response to climate change (Elzinga et al. 2007), and a longer duration of flowering may support both plant and insect populations as timing shifts occur, by buffering asynchrony in the timing of interacting species (Olliff-Yang et al. 2020). Because of this, we are interested in factors that influence flowering duration.
In this study we examine the effects of differential seeding time on plant growth and flowering time, with the goal of informing restoration management practices. We explore how differences in biotic environments and abiotic conditions after germination due to different planting dates and presence vs. absence of neighbors influence the timing and duration of flowering and reproductive success in a widespread annual plant used commonly in restoration practices. Specifically, we address: 1) Does delayed planting time delay individual plant flowering time? 2) How does planting time influence population flowering duration? and 3) How does germination timing influence survival and reproduction with and without neighboring plant removal?
Methods
Study Site and Species
The study species, California Goldfields (Lasthenia californica DC. ex Lindl. [Asteraceae] sensu Chan et al. 2002), is an annual forb that is widespread in northern California. This species is commonly used in restoration projects, which occur throughout the winter and early spring. As an obligate outcrossing species (Ornduff 1966), L. californica is dependent on pollinators for successful fruit set. This species also germinates quickly after exposure to cool wet conditions, which is ideal for establishing different germination cohorts. Populations of L. californica are found in a variety of different environments and habitats, including serpentine and non-serpentine grasslands.
This study occurred in a serpentine grassland at the UC Davis McLaughlin Natural Reserve in California (Lake Co., CA; 38.86007°, –122.40806°, 632 m elevation; Fig. 1A). Serpentine soils are formed from the metamorphosis of ultra-mafic mantle crust, yielding substrate that is high in heavy metals, low in essential nutrients, and low in calcium-to-magnesium (Ca2+/Mg2+) ratios (Safford et al. 2005). Site location was chosen based on the proximity to the source seed collection, and where the focal species, L. californica, is known to do well (C. Koheler, UC Davis Donald and Sylvia McLaughlin Natural Reserve, personal communication). This experiment was conducted during the growth season of L. californica during the 2016 water year, from November 2015 through June 2016. This species typically germinates with the first rains, grows vegetatively during the late fall and winter, and reproduces in the early spring. During this study the average daily air temperature was 12°C (minimum temperature –3.9°C, maximum temperature 32.8°C) and cumulative precipitation received was 268 mm (Western Regional Climate Center, Knoxville Creek, 38.861944°, –122.417222°, 670 m elevation [<1 km from the study site]), with a mean annual temperature of 15.9°C in 2016 (Wang et al. 2016). At this site, the average mean annual temperature was 15.2°C and mean annual cumulative precipitation was 765 mm for the 1981–2010 time period (generated with ClimateWNA v4.62, based on methodology in Wang et al. (2016)), so the conditions during the year of this study were relatively typical, but drier than the long-term mean.
Fig. 1.
Study site and experimental set up. A) Photo of study site on 15 April 2016. The south aspect of the serpentine meadow is in the foreground of the photo, and the picture was taken facing to the west. The seed collection site (“Goatgrass meadow”) can be seen in the center of the photo, and the UC NRS field station is located in the hills in the background. Photo by R. L. Olliff-Yang. B) Diagram of a single experimental block, with an example plot layout. Treatments consisted of two competition levels across three planting dates. The ibutton measuring soil temperature for each block was located in the center (asterisk). This experiment consisted of 18 of these blocks in a randomized design.
Seed was collected for this study in 2011 and 2015 from many (>100) individuals in an adjacent meadow. Seeds were stored together in paper bags at room temperature in a low humidity room until they were planted. All planting dates coincided with a rainfall event (Fig. 2).
Fig. 2.
Timing of biotic and abiotic events. A) Growth and flowering time of Lasthenia californica in experimental plots. Solid lines indicate ambient plots (A), and dashed lines indicate neighbor removal plots (NR). Circle points indicate planting dates for each cohort, plus signs note average date of maximum height reached, and triangles indicate average start of flowering. B) Growing degree day (GDD) accumulation for each cohort (solid grey lines) and precipitation events during the experiment (black dot-dash line). Large circle points at the beginning of each accumulation curve indicate planting dates in November, January, and March. In all panels colors indicate planting time (dark grey = November 1 2015, medium grey = January 1 2016, and light grey = March 1 2016). Black vertical lines connecting graphs show where flowering start corresponds with growing degree day accumulation. Note lower accumulation of degree days in both later planted cohorts.
Treatments
To control for spatial variation in soil properties, treatments were set up in blocks. We set up 18 randomized blocks, each with six 30 × 30 cm plots arranged in a grid – one plot for each combination of planting date (3) and neighbor removal treatments (2), as described below and diagrammed in Figure 1B. Each plot had a 10 cm buffer between itself and the next plot and the edge of the block. Blocks were placed in the meadow using stratified random sampling, with nine plots placed randomly on the north facing slope, and nine on the south facing slope of the meadow to evenly account for aspect variation. Thermochron ibuttons (iButtonLink, LLC, Whitewater, WI) were deployed in November, 5 cm below the soil surface in the center of each block, and used to calculate differences in thermal degree-days experienced by plants. These were set to record temperature every hour, and allowed us to examine relative temperature differences between blocks.
In each block two plots were seeded on each planting date: November 1, 2015, January 1, 2016, and March 1, 2016. In each plot 20 seeds were planted, with 8 and 12 seeds from the 2011 and 2015 seed collections, respectively. During planting, small holes (ca. 2 mm) were made with sewing pins and seeds were placed into the near-surface depressions, using tweezers to guide each seed in, pappus side up. Seeds were planted individually in a 4 × 5 grid with 5 cm distance between each other. A total of 2160 seeds were planted across all 108 plots (6 plots × 18 blocks). The location of each planted seed was marked with a pin, with a large colorful head and pins were left in place throughout the study to identify planted seedlings. All three seeding cohorts were planted just before a rain event to stimulate germination.
In each block, one of the two plots per planting date was assigned to be an ambient plot (A), and the other a neighbor removal plot (NR) in which all other plants were clipped at the beginning of the season and every two weeks thereafter until flowering was complete (Fig. 1). Clipping at the soil surface likely did not eliminate all interactions with neighbors, but minimized these as much as possible without disturbing the soil. No plots needed to be clipped before the November planting date, as the meadow had burned in August 2015 and no germination had yet occurred.
Experimental plants were surveyed every two weeks. At each survey, vertical plant height (to tip of tallest leaf) was measured, and number of inflorescences was recorded. Survival to reproduction was calculated as the percent of germinated individuals that made it to flowering. During survey dates the inflorescence counts were categorized as “buds”, “flowering”, “recently flowering”, and “fruiting”. The “recently flowering” inflorescences were beginning to senesce (wilting and drying, with darker golden flowers and spent anthers), but could be assumed to be flowering during the week prior to the survey. Therefore, these inflorescences were given a flowering date of one week prior for the flowering time analyses. It is possible that some inflorescences were missed, going from bud to fruit in the period between surveys, but including both fresh and week-old inflorescences seemed to capture flowering in most individuals. Dates of flower counts were recorded as day of year, counting from January 1. For each individual, start date was calculated as the date of first flower recorded, peak date as the date with most inflorescences in flower, and end date as the last date of inflorescences in flower.
Plants in the Asteraceae produce inflorescences with few to many individual flowers on a head (Keil 2021). In L. californica, each flower has the potential to produce one seed, although this is dependent on successful pollination. The number of seeds per inflorescence is therefore dependent on both number of flowers produced (which can range from 3 – 50+) and on successful pollination. In this study, the number of inflorescences was counted to estimate a proxy of reproduction and infer possible differences in fitness between treatments. We observed that individuals that were able to produce many inflorescences also often produced more flowers per inflorescence. Individuals with more inflorescences also had longer overall flowering duration, and therefore an increased likelihood of successful pollination events. Hence, we believe that counting inflorescences provides a conservative measure of treatment differences in total fecundity (i.e., true differences would be even larger).
There was some difficulty with seedling identification as L. californica seedlings can look very similar to other species (e.g., Plantago erecta E.Morris before leaf hairs emerge). Although seedlings were marked by pins, if a similar species germinated very close to the pin, seedlings could have been confused. At the time of data collection, unclear species identification was noted. In addition, other L. californica individuals were present in ambient plots, and in some cases it was unclear which individuals were growing from the planted marked seed. All plants with uncertain identification or origin were removed from analyses.
Treatment Effects on Growth Rate
The effects of planting date, neighbor removal, and their interaction on the time from planting date to flowering were assessed using linear mixed models, using plot nested within block as a random factor (n = 18 blocks). Seed collection year (2011 or 2015) was also included as a fixed factor in the models to account for variation due to age of seeds. Residuals were examined to verify that model assumptions were met. Importance and significance of the fixed effects (seeding date, neighbor removal, and their interaction) in the models were determined using likelihood-ratio tests in R (version 3.31, R Core Team, R Foundation for Statistical Computing, Vienna, Austria).
Treatment Effects on Flowering Time and Duration
The effects of planting date, neighbor removal, and their interaction on flowering date (the calendar day of year of flowering in 2016), were assessed using linear mixed models, using plot nested within block as a random factor (n = 18 blocks), and seed collection year was included as a fixed factor to control for seed age effects. Differences in the duration of flowering in all treatment combinations (planting date × neighbor removal) were also tested. Residuals were examined to verify that model assumptions were met. Importance and significance of the fixed effects (seeding date, neighbor removal, and their interaction) in the models were determined using likelihood-ratio tests in R.
The effect of all treatments on the overall flowering timing and duration on the landscape was also examined by looking at the differences in start and end flowering dates and assessing the flowering time overlap between treatment plots. To examine the possibilities for heterogeneous treatments on the landscape to extend overall flowering duration, we examined duration and overlap of the flowering between planting and neighbor removal treatments. Following methods in Olliff-Yang and Ackerly (2020) for examining flowering overlap, we subtracted the average end date of early treatment combinations from the end dates of late treatment combinations. Similarly, early flowering start dates were subtracted from late flowering start dates. We then compared these numbers to examine season duration with and without treatment influences.
Treatment Effects on Inflorescence Number and Survival
The effects of seeding date and neighbor removal on inflorescence production were also assessed using linear mixed models, using plot nested within block as a random factor, and seed collection year was again included as a fixed factor to account for seed age. Data for inflorescence production was analyzed using a Poisson distribution generalized mixed model to fit this positive, integer-valued response variable. Due to low survival in the March cohort, reproduction differences were only tested in November and January cohorts. Fixed effect importance in the models were determined via examination of F-values from the model output. All analyses were performed in R and mixed models were fit using the lme4 package (Bates et al. 2015).
Results
Treatment Effects on Growth Rate
Planting date influenced growth rate and time to flowering. Initiation of first flower occurred faster in plots with later planting dates, as the number of days from planting to first flower was significantly reduced (Likelihood-ratio, χ2(2,484) = 316.2, P < 0.0001, Figs. 2, 3), with approximately 52 fewer days from germination to flowering in the November vs. March planting cohorts. Neighbor removal did not influence the number of days from planting to first flower (Fig. 3A). The initiation of flowering seemed to require a minimum height threshold around 3 cm, as both January and March cohorts began to flower around this height, while November cohorts began flowering at a slightly taller height on average (Fig. 3B). Maximum height was reached at a similar time despite different planting dates (Fig. 2A).
Fig. 3.
Flowering phenology, duration, and season extension. Planting cohorts are indicated by month abbreviations (Nov 1 = November 1 2015, Jan 1 = January 1 2016, and Mar 1 = March 1 2016), and treatment is shown with color (neighbor removal [NR] = dark grey, and ambient [A] = light grey). In all panels whiskers indicate standard error. A) Mean number of days to first flower compared across planting cohorts and neighbor removal treatments, (Planting date significant [F2, 62 = 2032, P < 0.0001]) B) Mean height at first flower compared across planting cohorts and neighbor removal treatments, C) Mean flowering duration (in # of days) of individuals in each treatment and planting cohort. D) Mean start, peak, and end dates of flowering in each planting cohort and treatment. Bar ends indicate start (left end) and end (right end) dates. Middle points are peak dates. Points indicate means and are staggered vertically for visibility, with start point the lowest, and end point the highest. Whiskers show standard error.
Treatment Effects on Flowering Time and Duration
Planting date and competition removal both influenced the timing of flowering in the season. Later planting dates yielded later start, peak, and end dates of flowering in the year (Table 1), with individuals planted January 1 flowering a mean of six days later, and March 1 flowering a mean of 14 days later than individuals planted in November. While neighbor removal plots exhibited similar flowering start dates, removing neighbors shifted flowering peak and end dates later (Table 1). Both later planting time and presence of neighbors shortened flowering duration (Table 1, Fig. 3C).
Table 1.
Planting Time and Neighbor Removal Influences on Flowering Date. Models of Start, Peak and End dates of flowering in 2016, as well as the Duration of flowering (calculated as the number of days flowering). Plot nested within block is included as a random effect in all models. The P-values indicate significance of including each fixed effect in the model (α < 0.05), as determined by Likelihood-ratio Tests comparing the full model against a model with the fixed effect removed. Fixed factors include: PD = planting date (levels: November 2015, January 2016, March 2016), NR = neighbor removal (levels: ambient, neighbor removal), and CY = seed collection year (levels: 2011, 2015).
Flowering time of plots with different treatments was complementary and planting both early and late cohorts led to longer flowering on the landscape overall, due to asynchronous flowering time between plots. Having staggered germination timing increased the duration of flowering on the landscape by approximately 11 days in ambient plots, due to different flowering time between cohorts (Fig. 3D). Flowering duration was further increased when competition removal plots were also considered, with an increase of up to 26 days of flowering on the landscape with the staggered timing between both later planting and competition removal plots (Fig. 3D).
Treatment Effects on Survival and Inflorescence Number
The percentage survival of germinated individuals to flowering was different between planting cohorts, with higher survival overall in earlier cohorts (Fig. 4A). Reproductive output (in November and January) was affected by both planting date and neighbor removal treatments. Later planting reduced the number of inflorescences produced by an average of 0.5 inflorescences per plant in ambient plots (Poisson, F(1,9798) = 46.56, P < 0.001); although low survival in March cohort meant we were unable to test, there does not appear to be a difference between reproductive output in January vs. March cohorts in ambient plots for plants that survived to flower (Fig. 4B, light grey [ambient] bars). In neighbor removal plots, planting date appears to have only reduced inflorescence number in the March cohort, by an average of one inflorescence per plant (Fig. 4B). Neighbor removal increased reproduction (Poisson, F(1,9888) = 69.4, P < 0.001, Fig. 4B). When combined into overall inflorescences per seed planted (which incorporates germination percentage, survival to flowering and inflorescence production) the results reveal a decrease in overall inflorescence production with each planting date, and an increase with neighbor removal (Fig. 4C). Assuming that each inflorescence has at least one flower and one successful pollination event leading to a mature seed, the mean of ∼1 inflorescence per planted seed in the ‘control treatment’ (early cohort with neighbors present) suggests a stable or growing population at this site, based on the demography in this study year.
Fig. 4.
Planting date and neighbor removal effect on survival and reproduction. A) percent of germinated individuals that survived to flowering, B) total inflorescence numbers per plant, and C) mean inflorescences per seed planted. In all panels light grey bars show ambient plots, and dark grey bars are neighbor removal plots. Whiskers show standard error.
Although the topographic treatment was not replicated, aspect did influence survival and reproduction in the treatments. Plots on the north-facing slope exhibited higher overall reproduction (Fig. 5), as survival and inflorescence production were lower in general on the south aspect. Neighbor removal seemed to ameliorate some of the effects of late planting on fecundity, except for the latest planting date (March 1) on the south-facing slope.
Fig. 5.
Effect of neighbor removal and planting time on overall survival and inflorescence production depends on aspect. Overall inflorescence production per seed on north-facing (left panel) and south-facing (right panel) slopes. Light grey bars show ambient plots, and dark grey bars are neighbor removal plots. Whiskers show standard error
Discussion
Planting time and neighbor removal influence flowering time, flowering duration, and inflorescence number in L. californica. This study gives evidence that flowering in this species responds to planting date, with delayed peak and end flowering dates and a shorter duration of flowering in later germinating individuals. Flowering time was strongly tied to a window in spring, and months of difference in planting time only led to slight differences in the calendar date of flowering time. Additionally, later planting dates decreased the survival and reproduction of individuals, resulting in lower overall planting success of later cohorts. The longer duration of flowering observed in early geminating and neighbor removal plots was due to an increase in the number of inflorescences produced by individual plants. We recommend earlier plantings with density reduction treatments for the most successful individuals in restoration contexts and suggest that extremely late planting should be avoided unless there is a goal to extend flowering at the end of the season.
Patterns observed in days to flower and date of maximum height reached revealed nuances of growth allocation timing. Despite the extreme differences in germination time, flowering start time was not shifted as dramatically as might be expected, indicating that that abiotic constraints at the end of the season may limit the capacity for flowering time shifts. Additionally, plants compensated for the germination delay by speeding up development, and maximum height was reached on similar dates in all planting cohorts (Fig. 2). Later planting dates also shortened the number of days between germination and first flower (Fig. 3A). A similar phenomenon has been noted in frog development with later egg hatch cohorts exhibiting faster development in the absence of priority effects (Murillo-Rincón et al. 2017). The shortened time to reproduction reveals the tradeoff between vegetative growth and reproduction at the end of the season, likely due to increasing water limitation as summer approaches. Neighbor removal seemed to ameliorate this tradeoff slightly, as plants were able to begin flowering slightly later and produce more inflorescences in these plots (Figs. 3A and 4B), indicating that competitive effects are stronger than any facilitative interactions among neighboring plants. However, despite faster growth rates, there seemed to be a height threshold at the time of first flower, and flowering in the March cohort was delayed until individuals reached a height of approximately 3 cm (Fig. 3B). This suggests that the switch to flowering is not entirely dependent on environmental cues, and this height may reflect a size threshold where sufficient carbon accumulation for reproduction has been reached.
These data show that L. californica exhibits a graded allocation growth strategy, and a capacity to respond to variation in germination or planting dates. In fact, over 75% of flowering individuals grew vegetatively after initiation of reproductive growth (464 of 588 [78.9%] individuals grew in height after bud initiation). Additionally, as flowering start occurred at a similar time in all cohorts despite drastically different germination dates, this species exhibits plasticity in flowering time, switching from vegetative growth to flowering when the right conditions occur, as long as sufficient plant size has been achieved.
The season length in California can vary dramatically each year due to rainfall patterns and climate oscillation patterns (e.g., El Nino Southern Oscillation), and differences in year-to-year rainfall are becoming more extreme and variable (Swain et al. 2018). The shallow and infertile soils in serpentine grasslands typically hold little water, and seasonal flowering time in these systems is often earlier due to faster soil dry-down rates (Schmitt 1983; Rajakaruna and Bohm 1999; Rossington et al. 2018). These factors make the season length unpredictable, with plant senescence at the end of the season tied to the soil dry-down. Therefore, a graded-allocation strategy is a logical response to natural selection, as it allows for bet-hedging when season lengths are unpredictable (Wong and Ackerly 2005).
Delayed planting greatly reduced the number of inflorescences produced, but neighbor removal ameliorated this effect. The decrease in reproduction occurred between November and later cohorts in ambient plots, but did not decline until the March cohort in neighbor removal plots. This pattern indicates that competition removal may ameliorate some loss of fitness due to late planting. Early growth was best for reproductive output, and therefore early planting time should be prioritized in restoration projects. This difference in total inflorescence production may indicate fitness differences, but full reproductive fitness will depend on total seed set. Lasthenia californica is obligately outcrossing (Ornduff 1966), and fluctuations in the presence of pollinators throughout the season might further influence reproductive success. It should also be noted that there was less germination overall in the March planting cohort, but seeds planted late may survive in the soil and yield successful individuals in the next growing season, and populations may vary in seed dormancy (e.g., congener L. fremontii (Torr. ex A.Gray) Greene [Torres-Martínez et al. 2017]). Tracking cohorts over multiple years is needed to determine any long-term impacts of planting time.
Flowering duration was shortened by later planting time, but neighbor removal influenced this pattern (significant PD × NR interaction, F(2,106) = 3.7, P = 0.03). Neighbor removal lengthened individual plant flowering duration in the November and January, but not the March, cohorts (Fig. 3C). This interaction between planting time and neighbor removal treatments suggests that flowering at this location may be more constrained by abiotic factors at the end of the season (high temperature and low moisture) than at the beginning of the season. Flowering duration was cut short due to constraints at the end of the season regardless of neighbor environment. This likely reflects the stress of elevated temperatures and drought as summer approaches in Mediterranean-type systems, which is the main constraint for survival and reproduction of annual plants in these regions (Larcher 2000). This finding is consistent with congener L. gracilis (DC.) Greene, as individuals flower for longer periods of time in higher moisture conditions (Cox and Olliff-Yang 2021).
Neighbor removal delayed peak and end flowering, an effect that was most pronounced in the March cohort (Fig. 3D). Although reduced competition resulted in larger plant sizes overall, this did not result in an earlier switch to flowering, as found in other systems (Rathcke and Lacey 1985). This supports the prediction that selection should favor individuals that accelerate flowering in the face of increasingly scarce resources (Callahan and Pigliucci 2002), although only in peak and end flowering. The fact that flowering start date was not different with neighbor removal, indicates that neighbor removal indirectly contributes to the later peak and end dates, likely due to increased plant size and higher numbers of inflorescences produced per plant in plots with fewer competitors (Figs. 3B and 4B).
The study meadow burned in August 2015, only a few months before this experiment was conducted. Light availability is unlikely to have been changed much, as this is a low-productivity system with no canopy cover, but any soil nutrient changes from the fire might have an effect. Concentrations of calcium, potassium, and phosphorus are enriched in ash due to their resistance to volatilization (Pellegrini and Jackson 2020), and availability of these nutrients often increases in the soil after a fire. The low soil Ca2+/Mg2+ ratio often limits growth in serpentine soils, and fire can increase Ca2+/Mg2+ in these systems (DeSiervo et al. 2015). These nutrient changes may affect growth and plant-plant interactions, as species take advantage of the nutrient pulse. Fire has been shown to enhance native species richness on serpentine soils in this area (Harrison et al. 2003). While fire is likely to have some influence on the growth responses in this study, fire effects are often less pronounced on serpentine than more productive soil types (Safford and Harrison 2004).
Theory predicts that longer flowering duration can potentially ameliorate impacts of phenological mismatch on pollination mutualisms with climate change (Olliff-Yang et al. 2020). Therefore, based on our results, staggered reproductive timing in one location due to seeding time and competition reduction may also be a valuable conservation technique. Variable seeding time extended the flowering time in L. californica by an average of 11 days in ambient plots. Competition reduction in neighbor removal treatments yielded an average extension of 13 days. The extension from competition removal occurred mainly due to a lengthened end of season flowering duration, as individuals continued to produce inflorescences. Combining the two techniques in this location resulted in an average overall flowering time extension of 26 days. Therefore, these techniques may aid in increasing the overlap between flowering and pollinator presence, extending pollen and nectar resources for mutualistic flower visitors and enhancing reproductive assurance for the plants. However, the tradeoff of reduced survival and reproductive effort of later planted individuals should be taken into consideration.
During the study we observed additional heterogeneity in flowering time and reproduction due to aspect. The blocks were evenly split between a north-facing and south-facing slope, and abiotic differences likely led to additional extension in the duration of the flowering season (e.g., Olliff-Yang and Ackerly 2020). Plots on the north-facing slope exhibited later and longer flowering on average. Aspect also influenced inflorescence production (Fig. 5), interacting with the effect of planting time. Survival and inflorescence production were lower in general on the south aspect, and competition removal did not seem to ameliorate the effects of the March planting time. Although the topographic treatment was not replicated, this observation suggests that the success of later planting will depend on the abiotic conditions of a site. There may be a longer window of time to plant in cooler and wetter habitats, such as on North-facing slopes, and planting in hotter drier parts of the landscape should be prioritized earlier in the season.
In this study we see that both planting date and plant-plant interactions can influence the timing and success of reproduction. While there is a capacity to respond to late planting, as L. californica exhibited both plasticity in flowering time and a graded allocation growth strategy, later germination timing reduced fecundity and shortened flowering duration of individuals. This study has implications for the timing of grassland management and restoration practices, as planting time and density changes (e.g., due to grazing) may influence both the timing of flowering and overall fitness of individuals. Restoration practices in California annual grasslands include planting seed from October through March. These results suggest that later planting may result in an unsuccessful growth season. Reduced competition can ameliorate some of the negative impacts of a late start, but earlier planting dates should be prioritized.
Acknowledgments
Funding for this work was provided by a Mildred E. Mathias Research grant (UC NRS), The California Botanical Society, a Myrtle Wolf Scholarship (East Bay CNPS), a Natalie Hopkins Award (CNPS), and the UC Berkeley Department of Integrative Biology. Additional support was provided by the National Science Foundation Graduate Research Fellowship Grant (#1049702) to R. L. O.-Y. We would like to thank the staff at McLaughlin Reserve, especially C. Koehler, for support with logistics, site selection, Lasthenia californica seed and field support. Special thanks to A. C. Yang who assisted with the field work, discovered innovative ways to plant tiny seeds, found solutions for more comfortably sitting for hours counting and measuring thousands of plants on rocky soils, and provided much needed moral support. We are also grateful to T. E. Dawson, L. Larsen, W. Sousa, S. N. Sheth, M. F. Oldfather, P. Papper, M. M. Kling, E. E. Beller, C. E. Willing and S. Pierre for their advice and discussion during the conception and writing process. Thank you also to two anonymous reviewers, and to editors J. Yost, J. B. Whittall, and A. Hove whose comments helped improve this manuscript.
