Temperature, phenology, and plant defenses predict seed production near northern range limit

Background/Question/Methods

Organisms have been assumed to be locally adapted. Yet an increasing number of studies are finding greater fitness in genotypes from foreign populations. One explanation is that populations have become maladapted due to global change. This may be due to a mismatch with adaptations needed for new climate conditions or indirectly due to climate mediated changes in biotic interactions such as herbivory. Here I address local adaptation close to the northern range-limit of Oenothera biennis by asking: (1) What geographic/environmental traits predict O. biennis fitness? (2) Do herbivores decrease O. biennis performance? (3) What plant traits predict plant fitness, and which are most important? I assess the performance of genetic variation from across the entire range of O. biennis by growing 147 populations in a common garden close to the northern range limit. I estimate seed production as a proxy of plant fitness, and measure herbivory, phenology, leaf morphology and chemical defense traits. Using generalized linear models (GLMs) I assess where the most fit genotypes are located and how far away geographically and climatically these peaks are from the common garden. Using stepwise AIC, GLMs, and machine learning I then explore which plant traits best predict increased seed production.

Results/Conclusions

On average genotypes from further south perform better in the common garden. Specifically, seed production peaked in genotypes from 2.1° south of the common garden representing a 1.2°C mean annual temperature change. Increased herbivory did not strongly predict decreased seed production. However, most measured traits predicted seed number either strongly or moderately. There was strong evidence that intermediate flowering date, earlier bolt date, intermediate trichome number and lower SLA predicted greater seed number and moderate evidence that increased fruit chemical defenses predicted greater seed number. When assessing the relative importance of all traits in predicting seed number, flowering date, bolt date, and fruit total phenolics were most important both when assessed through multiple regression and gradient forest analyses. These result present evidence that local maladaptation may be explained by changes in temperature. Plant phenology appears to be providing adaptation to increased temperature, likely through earlier flowering and bolt date. Increased fruit chemical defenses may provide adaptations against herbivores we did not directly measure. Alternatively, fruit phenolics may provide abiotic adaptation or correlate with unmeasured traits that increase seed production during increased temperature. Overall, these results suggest climate change may be eroding away local adaptation to both biotic and abiotic factors.