Open Access

Highlights • Northern and eastern grassland-savanna boundary defined by minimum temperature. • Dynamics of fire, frost and growing season temperatures combine to produce this limit. • Western limit is related to moisture availability. • Modern, high-resolution climate data enables refinement of bioclimatic limits. • Reparameterisation improves global model performance at regional scale. Abstract Understanding the controls of biome distributions is crucial for assessing terrestrial ecosystem functioning and its response to climate change. We analysed to what extent differences in climate factors (minimum temperatures, water availability, and growing season temperatures (degree days above 5 °C (GDD5)) might explain the poorly understood borders between grasslands, savannas and shrublands in eastern South Africa. The results were used to improve bioclimatic limits in the dynamic global vegetation model (DGVM) LPJ-GUESS. The vegetation model was also used to explore the role of fire in the biome borders. Results show no clear differences between the adjacent biomes in water availability. Treeless grasslands primarily occur in areas with minimum temperatures and GDD5 values below that of savannas. The standard fire module in LPJ-GUESS is not able to reproduce observed burned area patterns in the study region, but simulations with prescribed fire return intervals show that a combination of low temperatures and fire can explain the treeless state of the grassland biome. These results confirm earlier hypotheses that a combination of low winter temperatures, causing frost damage to trees, and low growing season temperatures that impede tree sapling growth and recruitment, particularly under re-occurring fires, drive the grassland-savanna border. With these insights implemented, the LPJ-GUESS simulation results substantially improved grass distribution in the grassland biome, but challenges remain concerning the grassland-shrubland boundary, tree-grass competition and prognostic fire modelling.

Highlights • Floating ability facilitates water dispersal. • Hydrochorous seed dispersal is more effective than wind dispersal. • Storage in water induced germination rate. Abstract In many Central European countries Fraxinus pennsylvanica is an invasive species that spreads rapidly in floodplain forests. The aim of this study was to analyse anemochorous and hydrochorous dispersal distances and to compare the findings with dispersal data for the native Fraxinus excelsior. A simulation revealed that wind dispersal distances are similar for both species, reaching to 120–250 m. By contrast, the mean floating time (50% floating samaras) measured in an experiment was 2 days in the case of F. pennsylvanica and 0.5 days for F. excelsior. This high floating ability facilitates water dispersal over several kilometres in both species, but for the invasive species the modelled mean dispersal distance was 3.7 times higher. A germination test of F. pennsylvanica seeds revealed that the rate, onset and speed of germination increase with the duration of the inundation. After a maximum storage time in water of about 15 days the germination rate amounts to 78%, which was higher than the germination rate of seeds without storage in water (53%). We also found that regeneration was enhanced in flooded areas. Hydrochory, therefore, may be viewed as an important factor explaining the successful invasion of F. pennsylvanica in floodplain forests in Central Europe.

Climate forecasts show that in many regions the temporal distribution of precipitation events will become less predictable. Root traits may play key roles in dealing with changes in precipitation predictability, but their functional plastic responses, including transgenerational processes, are scarcely known. We investigated root trait plasticity of Papaver rhoeas with respect to higher versus lower intra-seasonal and inter-seasonal precipitation predictability (i.e., the degree of temporal autocorrelation among precipitation events) during a four-year outdoor multi-generation experiment. We first tested how the simulated predictability regimes affected intra-generational plasticity of root traits and allocation strategies of the ancestors, and investigated the selective forces acting on them. Second, we exposed three descendant generations to the same predictability regime experienced by their mothers or to a different one. We then investigated whether high inter-generational predictability causes root trait differentiation, whether transgenerational root plasticity existed and whether it was affected by the different predictability treatments. We found that the number of secondary roots, root biomass and root allocation strategies of ancestors were affected by changes in precipitation predictability, in line with intra-generational plasticity. Lower predictability induced a root response, possibly reflecting a fast-acquisitive strategy that increases water absorbance from shallow soil layers. Ancestors’ root traits were generally under selection, and the predictability treatments did neither affect the strength nor the direction of selection. Transgenerational effects were detected in root biomass and root weight ratio (RWR). In presence of lower predictability, descendants significantly reduced RWR compared to ancestors, leading to an increase in performance. This points to a change in root allocation in order to maintain or increase the descendants’ fitness. Moreover, transgenerational plasticity existed in maximum rooting depth and root biomass, and the less predictable treatment promoted the lowest coefficient of variation among descendants’ treatments in five out of six root traits. This shows that the level of maternal predictability determines the variation in the descendants’ responses, and suggests that lower phenotypic plasticity evolves in less predictable environments. Overall, our findings show that roots are functional plastic traits that rapidly respond to differences in precipitation predictability, and that the plasticity and adaptation of root traits may crucially determine how climate change will affect plants.

The fossil record of the diverse subfamily Passifloroideae (>750 species and 17 genera) is relatively poor. Despite the distinctiveness of its leaves (glandular and often emarginate), most of the fossils from this group have been described from seeds. Fossil seeds have been recovered from Europe, and North and South America. A lack of information on seed morphology for all the genera and tribes of this subfamily has prevented a tribe-level identification of the fossils and a better understanding of their biogeographic patterns. The Passifloroideae is divided into three tribes: Passifloreae with 10 genera, Paropsieae with six genera and the monotypic Jongkindieae. This study provides new descriptions for 15 species from 5 genera from the mostly Afrotropical tribe Paropsieae based on herbarium material, and introduces an online seed database and a key for 100 species of Passifloroideae compiled from literature and direct observations. Our study shows a low morphological diversity among the seeds of Paropsieae in comparison to a much larger diversity within Passifloreae. Some rare morphologies are only present in Passifloreae and can be used to assign seeds to this tribe. Within the Paropsieae, Androsiphonia has seed that are very distinct from those in the other genera in the tribe and also from the rest of the subfamily. The genus Paropsia exhibits two main morphotypes, while the genera Barteria, Paropsiopsis and Smeathmannia have very similar seeds with a highly conserved morphology. These results suggest that living or fossil Paropsieae cannot be identified confidently based solely on seed characters.

Field work in the Kibira National Park (Burundi), located in the Kivu-Ruwenzori system of the Afromontane Region, revealed the existence of a new species clearly belonging to the Argocoffeopsis-Calycosiphonia clade (Coffeeae, Rubiaceae). The species shows striking heterophylly: the plagiotropous branches have several nodes bearing reduced or even scaly leaves. For the rest, it shares characters with Calycosiphonia and Kupeantha. Therefore, a morphological comparison with the clade is done, as well as molecular phylogenetic analyses. The morphology of the novelty is closer to Kupeantha than to Calycosiphonia, inter alia because the anthers have no transverse septa, in contrast to the multilocellate anthers of Calycosiphonia. However, the molecular data advocate for a position in Calycosiphonia – a result weakening the morphological distinction between Calycosiphonia and Kupeantha. The former genus is no longer restricted to species with transverse septa in the anthers and with placental outgrowths around the seed. The new species is formally described as Calycosiphonia albertina Ntore & Robbr. sp. nov. Nomenclaturally, this placement is also the most conservative option. A taxonomic treatment, illustrations, a geographical distribution map, and a preliminary conservation assessment are provided. The previous inclusion of Calycosiphonia pentamera in Kupeantha based on morphology is here corroborated by molecular analyses.