Neighbor effects on population growth rate differ among populations due to variation in demographic rate sensitivities in Sedum lanceolatum

Population growth rates will respond to an environmental driver only if the driver impacts demographic rate(s) and the population is sensitive to impacted demographic rate(s). If populations vary in the sensitivity of population growth rate to demographic rates, the effect of an environmental driver on population growth rate could vary across populations, even if the effect of the driver on demographic rates does not vary across populations. Here, we use five years of demographic data of a common alpine plant, including data from a neighbor removal experiment and a climate warming experiment, to quantify the relative contribution of neighbor effects on demographic rates vs. sensitivity of population growth rate to demographic rates to across-population variation in neighbor effects on population growth rate. We find neighbor effects on population growth rate vary significantly across populations, and this effect is driven primarily by variation in sensitivity of population growth rate to demographic rates across populations. Further, results from our climate warming experiment suggests variation in sensitivity across populations is partly driven by temperature differences. Our results highlight the importance of considering changes in the sensitivity of population growth rate to demographic rates across space, and show how changes in sensitivity can contribute to spatial variation in the effects of a driver on population growth rate.

Methods
Data collection on variation in demography across elevations

To quantify variation in demography across elevations in S. lanceolatum, we conducted demographic censuses in two low-elevation and two high-elevation populations over four annual transitions. Hereafter, we call the data collected at these four populations, described in this section, the “demography” data. These four populations differ systematically in temperature, with higher temperatures at the two low-elevation populations. In 2011, during peak fruiting in late summer and early fall, we marked and mapped 1341 individuals across the four populations, measuring size (sum of volumes, height x area, across all rosettes), and counting fruits. We returned annually during peak fruiting until 2015 to score survival, measure size, and count fruits. Altogether, we collected data on 6705 individual x transitions, including data on very small plants and seedlings. We used these data to estimate five size-dependent vital rates: annual survival, mean size after one year of growth, variance in size after one year of growth, probability of fruiting, and number of fruits given a plant fruited. Note that while vegetative reproduction is common, occurring when individual rosettes take root after separating from primary stems, connections between parent and vegetative offspring are apparent above ground. Thus, any vegetative reproduction is subsumed into our estimates of growth.

To quantify seedlings per fruit (i.e., the number of seedlings present each year per fruits in the year prior), we first estimated the number of seeds per fruit at each elevation and then conducted a seed addition experiment at each population to estimate the population-specific number of seedlings per seed. To estimate seeds per fruit, we collected all fruits from 71 plants across our four populations in 2010, 2011, and 2012. For a subset of these plants (n = 25), we weighed the total seed mass and counted the total seed number, using these data to obtain the average seed mass for each elevation. In all, we counted 1073 seeds. We then used the relationship between seed mass and seed number to create a regression to estimate the population-specific number of seeds per fruit from the 72 plants (by weighing seed mass from individual fruits). To estimate the population-specific number of seedlings per seed, we conducted seed addition experiments at each population. At each population, we established five paired 10 cm x 10 cm plots (10 plots total) that were proximate to one another and 30 cm from any marked plants; one of the pair was randomly assigned to a seed addition treatment and the other to an unmanipulated control treatment. In 2014, we added 50 S. lanceolatum seeds to the seed addition treatments. In 2015, we counted the number of seedlings in all plots, calculating seedlings per seed as: (seedling number in seed addition plots – seedling number in control plots)/50. This procedure corrects our seed addition experiment for any background germination due to germination from a seedbank or natural dispersal into the plot. If seedling number was higher in control than in seed addition plots (occurs in 2/17 pairs), we estimated seedlings per seed as zero. Due to limited data, these values were averaged across plots and elevations rather than population specific values. Finally, we multiplied the elevation-specific seedlings per seed by the elevation-specific number of seeds per fruit to estimate seedlings per fruit for each population.

Data collection on neighbor effects across elevations

To quantify neighbor effects in S. lanceolatum, we conducted a neighbor removal experiment at each of the same four populations from 2011-2015. We call the data collected in this experiment the “removal experiment” data. In 2011, we marked and mapped 719 individuals that were >= 30 cm apart, and arrayed roughly equally across small, medium, and large size classes and across populations. We randomly assigned each individual to either a neighbor removal or an unmanipulated control treatment; small, medium, and large plants were roughly equally distributed among treatments. Annually, starting in 2011, we removed all above-ground biomass from a 15 cm radius around neighbor removal treatment plants, taking care not to disturb the soil. In 2011, during peak fruiting in late summer and early fall, we measured size and counted fruits of all these individuals. Until 2015, we returned annually during peak fruiting to score survival, measure size, and count fruits, replacing any individuals that died with newly marked individuals. Altogether, we collected data on 3595 individual x transitions, which we used to estimate the same five size-dependent vital rates as we did for our demography data. Note that we did not collect separate estimates of seedlings per fruit for neighbor removal treatments.