Herzlich willkommen
beim CIDAS!

Das Campus-Institut Data Science (CIDAS) ist eine zentrale wissenschaftliche Einrichtung der Georg-August-Universität und zugleich ein Knotenpunkt für die fakultätsübergreifende und campusweite Zusammenarbeit auf dem Gebiet Data Science. Dies betrifft insbesondere die Koordination und Weiterentwicklung der Aktivitäten in Forschung, Lehre und Wissenstransfer im Bereich Data Science. Durch die Interaktionen der verschiedenen Disziplinen Informatik, Mathematik, Statistik und den verschiedenen Anwendungsfächern besteht ein hoher interdisziplinärer Austausch, der durch das CIDAS auch auf internationaler Ebene gefördert wird.

CIDAS lecture series
Dr. Helena Bestová: Plant size distribution on islands
am 01. Dezember 2022 um 14:15 Uhr

Abstract: Plant size distribution relates to many important ecosystem properties such as productivity, biomass, carbon storage capacity. Therefore, it is important to understand what are the determinants of plants size distribution. Indeed, several studies already observed that plant size distribution is associated with island area and isolation. However, in order to move from observations to predictions, we need a mechanistic understanding. A promising venue is to improve a simple yet powerful model of species diversity, the Theory of Island Biogeography (TIB), that has been developed over 60 years ago. TIB and follow-up models uses island area and isolation to define a balance between species immigration and extinction. TIB gives an estimate of species richness on islands, but relies on the assumption that species are functionally equal. This means that every individual has equal chance to arrive and persist on the island, no matter its size or other traits. Nonetheless immigration and extinction should scale with plant size. Therefore, island biogeography models could (and should) be improved by incorporating allometric (size related) components. We applied allometric TIB model to 500 islands and their flora. We used the information on available in GIFT and TRY databases, which are the most extensive collections of data on floras and traits. They enabled ecologists to incorporate plant traits in biodiversity models. However, these databases contain a lot of missing data, even though the missing information can be found in thousands of scatter resources such as Floras, monographs and scientific papers. To harness this information for macroecological research, we are developing machine learning models for natural language processing and an automated trait-mining workflow. This will allow us to mobilize previously neglected information from existing literature and fill gaps in global trait databases, improve our models and facilitate future research.

Ort: Informatik Provisorium, Raum 0.102 in der Goldschmidtstraße. Ein anschließendes Get-Together findet in Raum 16-4 in der Goldschmidtstr. 3-5 statt.