Back

Consistent predictors of microbial community composition across scales in grasslands reveal low context-dependency

Radujkovic, D.; Vicca, S.; van Rooyen, M.; Wilfahrt, P.; Brown, L.; Jentsch, A.; Reinhart, K. O.; Brown, C.; De Gruyter, J.; Jurasinski, G.; Askarizadeh, D.; Bartha, S.; Beck, R.; Blenkinsopp, T.; Cahill, J.; Campetella, G.; Canullo, R.; Chelli, S.; Enrico, L.; Fraser, L.; Hao, X.; Henry, H. A. L.; Hohn, M.; Jouri, M. H.; Koch, M.; Lawrence Lodge, R.; Li, F. Y.; Lord, J. M.; Milligan, P.; Minggagud, H.; Palmer, T.; Schröder, B.; Szabo, G.; Zhang, T.; Zimmermann, Z.; Verbruggen, E.

2021-11-30 ecology
10.1101/2021.11.29.470306 bioRxiv
Show abstract

Environmental circumstances shaping soil microbial communities have been studied extensively, but due to disparate study designs it has been difficult to resolve whether a globally consistent set of predictors exists, or context-dependency prevails. Here, we used a network of 18 grassland sites (11 sampled across regional plant productivity gradients) to examine i) if the same abiotic or biotic factors predict both large- and regional-scale patterns in bacterial and fungal community composition, and ii) if microbial community composition differs consistently with regional plant productivity (low vs high) across different sites. We found that there is high congruence between predictors of microbial community composition across spatial scales; bacteria were predominantly associated with soil properties and fungi with plant community composition. Moreover, there was a microbial community signal that clearly distinguished high and low productivity soils that was shared across worldwide distributed grasslands suggesting that microbial assemblages vary predictably depending on grassland productivity.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.