Data for: Predicting the contribution of single trait evolution to rescuing a plant population from demographic impacts of climate change

Evolutionary adaptation can allow a population to persist in the face of a new environmental challenge. With many populations now threatened by environmental change, it is important to understand whether this process of evolutionary rescue is feasible under natural conditions, yet work on this topic has been largely theoretical. We used unique long-term data to parameterize deterministic and stochastic models of the contribution of one trait to evolutionary rescue using field estimates for the subalpine plant Ipomopsis aggregata and hybrids with its close relative I. tenuituba. In the absence of evolution or plasticity, the two studied populations are projected to go locally extinct due to earlier snowmelt under climate change, which imposes drought conditions. Phenotypic selection on specific leaf area (SLA) was estimated in 12 years and multiple populations. Those data on selection and its environmental sensitivity to annual snowmelt timing in the spring were combined with previous data on heritability of the trait, phenotypic plasticity of the trait, and the impact of snowmelt timing on mean absolute fitness. Selection favored low values of SLA (thicker leaves). The evolutionary response to selection on that single trait was insufficient to allow evolutionary rescue by itself, but in combination with phenotypic plasticity it promoted evolutionary rescue in one of the two populations. The number of years until population size would stop declining and begin to rise again was heavily dependent upon stochastic environmental changes in snowmelt timing around the trend line. Our study illustrates how field estimates of quantitative genetic parameters can be used to predict the likelihood of evolutionary rescue. Although a complete set of parameter estimates are generally unavailable, it may also be possible to predict the general likelihood of evolutionary rescue based on published ranges for phenotypic selection and heritability and the extent to which early snowmelt impacts fitness.

Methods
The study sites consisted of three “Poverty Gulch” sites in Gunnsion National Forest and one site “Vera Falls” at the Rocky Mountain Biological Laboratory, all in Gunnison County, CO, USA. Focal plants included two sets of plants. One set (data from 2009-2019) consisted of plants in common gardens at three sites: an I. aggregata site (hereafter “agg”), an I. tenuituba site (hereafter “ten”) and a site at the center of the natural hybrid zone (hereafter “hyb”). The second set consisted of plants growing in situ at two of the same Poverty Gulch sites (“agg” and “hyb”), and an I. aggregata site at Vera Falls (hereafter “VF”; data from 2017-2023). 

The common gardens were started from seed in 2007 and 2008. Measurements of SLA in these gardens began when plants were 2 years old, either 2009 or 2010 depending upon the garden, as they are only small seedlings during their first summer after seed maturation. By 2018, all but 15 of the 4512 plants originally planted had died, with or without blooming, and we stopped following these gardens. Starting in 2017, in situ vegetative plants at the I. aggregata site and the hybrid site whose longest leaf exceeded 25 mm were marked with metal tags to facilitate identification.  

In each year of the study, one leaf from each vegetative plant was collected in the field and transported on ice to the RMBL, 8 km distant. There each leaf was scanned with a flatbed scanner and analyzed using ImageJ to measure leaf area. The leaf was dried at 70 deg C for 2 hours and then weighed to obtain dry mass and calculate SLA as area/dry mass. For plants in the common gardens, SLA was measured on 982 leaves from 383 plants in 2009 – 2014. For in situ plants, SLA was measured on one leaf from each of 877 plants in 2017 – 2022. Fitness was estimated as the binary variable of survival to flowering. Plants that were still alive in 2019 in the common gardens or in 2023 at the end of the study were assumed to survive to flowering.

These data were used to estimate selection differentials on SLA in each of 12 years. We then combined this information with previous information on heritability and the effect of snowmelt date in the spring on mean absolute fitness, measured as the finite rate of population increase, from a previous demographic study. This information was used to parameterize models of evolutionary rescue that we developed. We developed two models that differed in how snowmelt timing changed: a Step-change model and a Gradual environmental change model and analyzed both deterministic and stochastic versions. All analysis and modeling was done in R ver 4.2.2.

进化适应可使种群在面临全新环境挑战时得以存续。当前许多种群正受到环境变化的威胁，因此探究自然条件下进化救援（evolutionary rescue）这一过程是否可行具有重要意义，但目前该领域的研究大多停留在理论层面。本研究采用独特的长期观测数据，以亚高山植物Ipomopsis aggregata及其与近缘物种I. tenuituba的杂交群体为研究对象，通过野外实测参数，针对单一性状在进化救援中的贡献构建确定性与随机性模型并完成参数化。
若不存在进化或表型可塑性（phenotypic plasticity），受气候变化引发的融雪提前所导致的干旱胁迫影响，本研究中的两个目标种群预计将在当地灭绝。针对比叶面积（specific leaf area, SLA）的表型选择，研究团队在12年间、多个种群中完成了参数估算。上述关于选择及其对春季年度融雪时间的环境敏感性的数据，与该性状的遗传力（heritability）、表型可塑性以及融雪时间对平均绝对适合度（mean absolute fitness）影响的既往数据进行了整合。
选择偏好较低的比叶面积值（即叶片更厚）。仅依靠该单一性状的进化响应无法单独实现进化救援，但结合表型可塑性后，该性状可在两个研究种群中的一个内促进进化救援的发生。种群规模由衰退转为增长所需的年限，在很大程度上取决于围绕长期趋势的融雪时间随机性环境波动。本研究阐明了如何通过野外实测的数量遗传参数，预测进化救援发生的可能性。尽管完整的参数估算集通常难以获取，但基于已发表的表型选择与遗传力范围，以及融雪提前对适合度的影响程度，仍可对进化救援发生的总体可能性进行预测。

研究方法
研究样地包括美国科罗拉多州甘尼森县（Gunnison County, CO, USA）内甘尼森国家森林（Gunnison National Forest）中的3处“贫困峡谷（Poverty Gulch）”样地，以及落基山生物实验室（Rocky Mountain Biological Laboratory）内的1处“维拉瀑布（Vera Falls）”样地。目标植株包含两组：第一组（数据采集于2009-2019年）为3处同质种植园（common gardens）中的植株，分别为Ipomopsis aggregata样地（下文简称“agg”）、I. tenuituba样地（下文简称“ten”），以及自然杂交带中心的样地（下文简称“hyb”）；第二组为原生境生长的植株，涵盖2处上述贫困峡谷样地（“agg”与“hyb”），以及维拉瀑布样地的Ipomopsis aggregata种群（下文简称“VF”；数据采集于2017-2023年）。
同质种植园于2007年和2008年由种子播种建立。由于种子成熟后的第一个夏季植株仅为小型幼苗，因此该种植园的比叶面积（SLA）测量始于植株2龄期，具体年份为2009年或2010年，依种植园而定。至2018年，初始种植的4512株植株中仅余15株存活（无论是否开花），研究团队停止对该种植园的监测。自2017年起，研究团队对Ipomopsis aggregata样地与杂交带样地中最长叶片超过25 mm的原生境营养生长期植株，使用金属标牌进行标记以方便识别。
在研究的每一年中，研究人员从每株营养生长期植株上采集一片叶片，置于冰上运至8公里外的落基山生物实验室（RMBL）。随后使用平板扫描仪对每片叶片进行扫描，并通过ImageJ软件分析以测定叶面积。将叶片置于70℃下烘干2小时后称重以获取干质量，并通过叶面积与干质量的比值计算比叶面积（SLA）。
针对同质种植园中的植株，研究团队在2009-2014年间对383株植株的982片叶片进行了比叶面积测量。针对原生境植株，研究团队在2017-2022年间对877株植株各采集一片叶片进行测量。适合度以“是否存活至开花”这一二分类变量进行估算：将2019年仍存活的同质种植园植株，以及研究结束时（2023年）仍存活的原生境植株，均认定为存活至开花。
上述数据被用于估算12年间每一年的比叶面积选择差。随后，研究团队将该数据与既往种群生态学研究中获取的信息进行整合，包括该性状的遗传力、春季融雪日期对以种群有限增长率为指标的平均绝对适合度的影响。利用这些整合后的信息，对本研究构建的进化救援模型进行参数化。
本研究构建了两类融雪时间变化模式不同的模型：阶跃变化模型与渐进式环境变化模型，并分别对其确定性与随机性版本进行了分析。所有分析与建模均在R 4.2.2版本软件中完成。