Biome partitioning of the global ocean based on phytoplankton biogeography

Abstract

Biomes are geographical units that can be defined based on biological communities sharing specific environmental and climatic requirements. Contemporary ocean biomes have been constructed based on various approaches. These included the biogeographic patterns of higher trophic level organisms, physical and biogeochemical properties, or bulk biological properties such as chlorophyll-a, but none considered the biogeographic patterns of the first trophic level explicitly, i.e. phytoplankton biogeography. A global description of marine biomes based on phytoplankton and defined in analogy to terrestrial vegetation biomes is still lacking. A bioregionalization based on phytoplankton appears particularly timely, as phytoplankton have a high sensitivity to climatic changes and fuel marine productivity. Here, we partition the global ocean into biomes by using self-organizing maps and hierarchical clustering, drawing on the biogeographic patterns of 536 phytoplankton species predicted from empirical evidence. Our approach reveals eight different biomes at the seasonal scale, and seven at the annual scale. The biomes host characteristic phytoplankton species compositions, and differ in their prevailing environmental conditions. The largest differences in phytoplankton composition are found between a Pacific equatorial biome and other tropical biomes, and between subtropical and high latitude biomes. The Pacific equatorial biome is characterized by species with narrower ecological niches, the tropical and subtropical biomes by cosmopolitan generalists, and the high latitudes by species with a heterogeneous biogeography. The strongest differences between biomes are found along gradients of temperature and macronutrient availability, associated with latitude. We test whether our biomes can be reproduced based on indicator species, or potential co-occurrence networks of species determined from the predicted species distributions that are wide-spread in some but rare in other biomes. We find that our biomes can be reproduced by the 51 species identified, which together form significant species co-occurrences. This suggests that species co-occurrences, rather than individual indicator species drive oceanic biome partitioning at the first trophic level. Our biome partitioning may be especially useful for comparative analyses on the functional implications of phytoplankton organization, and impacts on zoogeographical partitionings. Furthermore, it provides a framework for predicting large-scale changes in phytoplankton community structure due to anthropogenic climate and environmental change.