Search

Aim: The use of neural networks (NNs) is spreading to all areas of life, and Ecology is no exception. However, the data-hungry nature of NNs can leave out many small, valuable datasets. Here we show how to apply transfer learning to rescue small datasets that can be invaluable in understanding patterns of species co-occurrence. Location: Semiarid plant communities in Spain and México. Time period: 2016–2022. Major taxa studied: Angiosperms. Methods: Based on a large sample of plant species co-occurrence in vegetation patches in a semi-arid area of eastern Spain, we fit a generative artificial intelligence (AI) model that correctly reproduces which species live with which in these patches. Subsequently, we train the same type of model on two communities for which we only have smaller datasets (another semi-arid community in eastern Spain, and a tropical community in Mexico). Results: When we transfer the knowledge learnt from the large dataset directly to the other two, the predictions improve for the community more similar to our reference one. As for the more dissimilar community, improving the accuracy of the transfer requires a further tuning of the model to the local data. In particular, the knowledge transferred relates primarily to species frequency and, to a lesser extent, to their phylogenetic relationships, which are known to be determinants of species interaction patterns. Main conclusions: This AI-based approach can be performed for communities similar or not so similar to the reference community, opening the door to systematic transfer learning for accurate predictions on small datasets. Interestingly, this transfer operates by matching unrelated species between the origin and target datasets, implying that arbitrary datasets can then be transferred to, or even combined in order to augment each other, irrespective of the species involved, potentially allowing such models to be applied to a wide range of plant communities in different climates.

Tumour-infiltrating FoxP3+ regulatory T cells have been identified as both positive and negative prognostic factors in colorectal cancer (CRC) and rectal cancer (RC). In this study we investigated whether immune phenotypes, defined by CD8+ cytotoxic T cell density, may influence the prognostic association of FoxP3+ T cell densities in RC. Tissue microarrays from 154 rectal cancer resections were immunohistochemically double stained for CD8 and FoxP3. CD8+ and FoxP3+ cell densities were measured in the stromal and intraepithelial compartment. Stromal FoxP3+ cell densities were not associated with 10-year overall survival (OS). In the “immune-desert” phenotype, defined by very low stromal CD8+ cell density, a high density of stromal FoxP3+ T cells displayed a tendency towards an association with decreased 10-year OS (p = 0.179). In “inflamed” tumours, defined by high intraepithelial CD8+ T cell infiltration, the opposite was the case and high stromal FoxP3+ T cell densities were a positive prognostic factor (p = 0.048). Additionally, patients with an increased FoxP3/CD8 cell density ratio demonstrated a strong trend towards decreased 10-year OS (p = 0.066). These contrasting findings suggest functional heterogeneity within the group of FoxP3+ T cells. They are consistent with experimental studies which reported suppressive and non-suppressive populations of FoxP3+ T cells in CRC. Furthermore, our study demonstrates that CD8 immunohistochemistry may act as an instrument to identify tumours infiltrated by possibly functionally differing FoxP3+ T cell subtypes.