Life-history expressions of annual plants in unpredictable environmnets: from theoretical models to empirical tests

Abstract

Survival of annual plants in unpredictably varying environments may be possible through life-history adaptations, i.e. seed dormancy, or through phenotypic variation in life-history traits. In this thesis we studied these evolutionary and environmental life-history expressions along a steep rainfall gradient in Israel, spanning from arid habitats with highly variable and unpredictable rainfall to less variable and more predictable mesic Mediterranean habitats. We studied three winter annual species abundant along the gradient, and differing in dormancy from low to high: Bromus fasciculatus, Biscutella didyma and Hymenocarpos circinnatus. The major objective of this thesis was to perform original empirical tests for optimal germination strategies that were frequently theoretically modeled, but rarely empirically proved. Namely, we tested the pioneering model that predicts the germination fractions are constant over time and proportional to increasing amounts and predictability of rainfall (Cohen 1966). We studied the effects of average annual rainfall separately from maternal effects, which has not been done before. Our results show germination fractions of Biscutella and Bromus varying in a clinal fashion along the gradient, supporting the Cohen (1966) model. The trend for Biscutella was consistent during three consecutive years indicating a strong genetic determinant of germination. This genetic basis still needs to be established, and we advocate careful addressing of maternal effects too. Maternal effects have been considered in optimal germination strategy only recently, when predictive germination was modeled with respect to competition and environmental variability (Tielbörger & Valleriani 2005). For the first time, we tested this model in real field populations, namely the prediction that maternal fecundity is positively related to rainfall amounts and offspring germination is negatively related to maternal fecundity. We document maternal effects in more humid and more competitive environment (the Mediterranean), and suggest their short-term advantage in controlling sibling competition. With this pioneering approach to test maternal effects on predictive germination, we encourage more rigorous empirical studies to find out whether the observed patterns are universal. In a large-scale field experiment we tested the predicted optimal germination strategies (Cohen 1966, 1967) under naturally varying climatic conditions. Simultaneously, we evaluated the importance of separating climatic from other environmental variables. We reciprocally transplanted conspecific seeds of the species from and to the four sites along the gradient and expected higher germination fractions of seeds from wetter origins and at wetter sites. By sowing seeds on both local and standard soil, we separated climatic from edaphic and neighbor effects. We found strong substrate effects but relatively weak climate effects. This indicates that local environmental conditions may override effects of climate and should be critically addressed in future studies testing for local adaptation to current climate and for the potential of species to adapt to changing climate. Finally, adaptations to local climate and pre-adaptations to changing climate may reflect in life-history traits beyond germination. While exploring phenotypic variation in several morphological and reproductive traits of Biscutella didyma, we found clinal trends in the trait expression along the gradient, indicating two strongly opposing selective forces at the two extremes of the aridity gradient, resulting in contrasting strategies within the species. These clinal trends in the studied traits were consistent with germination trends of Biscutella in our parallel studies and indicate ecotypic differentiation/local adaptation of the climatic populations. Regarding survival under variable climate and potential climate changes, our observed trends in seed dormancy and other life history traits encourage future research on within-population genetic variation related to adaptive traits.

Pflanzen müssen in den einzelnen Stadien ihres Lebenszykluses Anpassungen an die jeweils herrschenden Umweltbedingungen entwickeln, um langfristig überlebensfähige Populationen aufbauen zu können. Diese Anpassungen, z.B. Keimungsstrategien, Dormanz und Samengrößen, müssen inbesondere in Lebensräumen mit räumlich und zeitlich unvorhersehbar variierenden Umweltbedingungen sehr komplex sein. Mit Hilfe der Modellierung wurden schon früh (Cohen 1966) optimale Keimungstrategien für Lebensräume entwickelt, die sich in der Vorhersehbarkeit der Umweltbedingungen unterscheiden (predictable vs unpredictable). Die Ergebnisse dieser Modelle sind in der ökologischen Wissenschaft allgemein akzeptiert, ohne dass eine empirische Überprüfung vorgenommen wurde. Ziel dieser Arbeit ist es, am Beispiel von drei winterannuellen Pflanzenarten (Bromus fasciculatus, Biscutella didyma and Hymenocarpos circinnatus) und entlang eines Gradienten der Regenfallvariation zu testen, ob die sich aus den Modellierungen ergebenden optimalen Keimungsstrategien empirisch nachweisbar sind. Die Ergebnisse zeigen, dass diese Anpassungen an die Vorhersagbarkeit der Umweltbedingungen existieren. Die Ausprägung der Anpassungen sind jedoch artspezifisch und variieren orientiert an den Umweltbedingungen des Vorjahres. Dies bedeutet, dass in Lebensräumen mit jährlich stark wechselnden Umweltbedingungen, die Umweltbedingungen eines gegebenen Jahres zu Veränderungen in der generellen Anpassungsstartegie der Pflanzen führen.