Description: As a measure against the adaptive potential of enemies and for reduction of metabolic costs, defense in multicellular organisms is often regulated. The regulation usually involves molecular perception of enemy presence or activity, followed by activation or induction – either local or systemic - of defense-related proteins. Animals and vascular plants are known since more than a century to defend themselves facultatively against pathogens and grazers. Macroalgal defense, in contrast, has until recently mainly been regarded as “constitutive” in the sense of “permanent” or “unregulated”. Indeed, many macroalgae appear to be chemically defended at constantly high levels and this is possibly one of the reasons why the first evidence of enemy-aroused resistance in a macroalga was only detected a few years ago.

Description: The genus Ulva is has broadly negative connotations because of its ability to form harmful “green tides” and the problems it causes with precise species identification, due to its morphological plasticity. During recent years, tides of unattached Ulva compressa U. Linneus 1753 with an atypical sheet-like morphology were for the first time observed in the German Baltic. Here we report that this nuisance alga is conspecific with the type strain of U. mutabilis Föyn 1958 from Faro in Portugal, an important model organism to study morphogenesis, morphogenetics and mutualistic interactions. Different approaches were used to examine conspecificity: (1) Comparisons on vegetative and reproductive features of cultured material of Ulva mutabilis and German Ulva compressa resulted in congruent results proving that a certain morphogenetic mutation pattern is shared. Spontaneous mutations of “slender-like” thalli are appearing whilst the common form exhibits a “leaf-like” wildtype morphology. (2) Interbreeding experiments of gametes of Ulva compressa and Ulva mutabilis were successful and showed a fertile first-generation offspring exhibiting the typical wildtype morphology similar to the phenotype of the parental generation. (3) Phylogenetic and species delimitation analyses were carried out on 128 tufA sequences of Ulva compressa specimens sampled in 2014–2016 in Germany and on tufA sequences of two clones of the strains Ulva mutabilis (sl-G[mt+]) and Ulva mutabilis (wt-[mt-]) to identify Molecular Operational Taxonomic Units (MOTUs). The Generalized Mixed Yule-Coalescent (GMYC) method comprises one major MOTU containing all included sequences of Ulva compressa and Ulva mutabilis, while reference sequences included in the analysis clustered outside this MOTU. This highly supports the monophyly of Ulva compressa and Ulva mutabilis, which can be treated as the same species. As a consequence, U. mutabilis is also a suitable model for future studies of green tides and their molecular and morphogenetic basis in the Baltic Sea.

Description: Fouling is a stressor that might determine the fate of seaweeds, but reports of algal adaptation to epibiosis are scarce. Previous comparisons have shown resistance to epibionts can be higher in non-native than in resident seaweed species, but we do not know whether it is an intrinsic trait of the non-natives or it has been acquired during the invasion process. We here compared native and non-native populations of the same algal species to elucidate this question. Resistance against two groups of epiphytes was assessed in living thalli and in artificial substrata coated with surface extracts, both gained from four Asian (native) and four European (non-native) populations of the red alga Gracilaria vermiculophylla. Two diatom species and two filamentous macroalgae were used as micro- and macro-epiphytes, and one of each type was collected in Asia, while the other came from Europe. Laboratory assays were done in both distributional ranges of G. vermiculophylla and in different seasons. We used a fully crossed design with the factors (i) ‘Origin of Gracilaria’, (ii) ‘Origin of epiphytes’, (iii) ‘Season’ and (iv) ‘Solvent used for extraction’. Both groups of epiphytes, regardless of their origin, attached less to living thalli and to surface extracts from non-native G. vermiculophylla. Fewer diatoms attached to hexane-based extracts, while fewer Ceramium filaments settled on extracts gained with dichloromethane. Our results show for the first time that non-native individuals of a seaweed are better defended against epiphytes than native conspecifics. Furthermore, we found evidence that at least a part of the defence is based on extractable secondary metabolites. We suggest that an enhanced defence against epiphytes after introduction is one reason for G. vermiculophylla’s invasion success. Our observation may also apply to other basibiont–epibiont interactions and could be a key feature of seaweed bioinvasions.

Description: High-quality data is necessary for modern machine learning. However, the acquisition of such data is difficult due to noisy and ambiguous annotations of humans. The aggregation of such annotations to determine the label of an image leads to a lower data quality. We propose a data-centric image classification benchmark with ten real-world datasets and multiple annotations per image to allow researchers to investigate and quantify the impact of such data quality issues. With the benchmark we can study the impact of annotation costs and (semi-)supervised methods on the data quality for image classification by applying a novel methodology to a range of different algorithms and diverse datasets. Our benchmark uses a two-phase approach via a data label improvement method in the first phase and a fixed evaluation model in the second phase. Thereby, we give a measure for the relation between the input labeling effort and the performance of (semi-)supervised algorithms to enable a deeper insight into how labels should be created for effective model training. Across thousands of experiments, we show that one annotation is not enough and that the inclusion of multiple annotations allows for a better approximation of the real underlying class distribution. We identify that hard labels can not capture the ambiguity of the data and this might lead to the common issue of overconfident models. Based on the presented datasets, benchmarked methods, and analysis, we create multiple research opportunities for the future directed at the improvement of label noise estimation approaches, data annotation schemes, realistic (semi-)supervised learning, or more reliable image collection.