Co‑occurrence structure of late Ediacaran communities and influence of emerging ecosystem engineers

Understanding the roles of habitat filtering, dispersal limitations, and biotic interactions in shaping the organization of animal communities is a central research goal in ecology. Attempts to extend these approaches into deep time have the potential to illuminate the role of these processes over key intervals in evolutionary history. The Ediacaran marks one such interval, recording the first macroscopic benthic communities and a step-wise intensification in animal ecosystem engineering. Here, we use taxonomic co‑occurrence analysis to evaluate how community structure shifted through the late Ediacaran and the role of different community assembly processes in driving these changes. We find that community structure shifted significantly throughout the Ediacaran, with the most dramatic shift occurring at the White Sea‑Nama boundary (~550 Ma) characterized by a split between older, more enigmatic taxonomic groups (the ‘Ediacara-type’ fauna) and more recognizable (‘Cambrian-type’) metazoans. While ecosystem engineering via bioturbation is implicated in this shift, dispersal limitations also played a role in separating biota types. We hypothesize that bioturbation acted as a local habitat filter in the late Ediacaran, selecting against genera adapted to microbial mat ecosystems. Ecosystem engineering regime shifts in the Ediacaran may thus have had a large impact on the development of subsequent metazoan communities.

Methods
Data collection began with a download from the Paleobiology Database (July 2021) of all macrofossil occurrences dated to the Ediacaran period (635-538ma). We searched the primary literature for additional fossil collections and incorporated these data into the initial occurrence data set. This final data set encompassed 1540 fossil specimens from 173 references. Collected data included genus, species, reported collection identification, stratigraphic unit, formation, member, lithological descriptions, modern latitude and longitude of collection, ichnogenera present, algae present, estimated water depth (shallow/deep at coarsest resolution), assemblage zone, and reference information for each fossil specimen. Assemblage zone assignments (the finest temporal units available to the Ediacaran period at present) were manually reviewed for each formation and member from a review of the chronostratigraphic literature, using radiometric dates, stratigraphic sequence, biostratigraphic correlations to assign each fossil occurrence to the Avalon, White Sea, or Nama. 

Genus and species assignments were also manually reviewed for each collection and updated to the latest consensus reported in the primary literature. Generalized biota type was assigned based on morphological and ecological characteristics (eg. feeding mode, body plan) as defined by the primary literature for each genus or member of the same morphogroup. Supporting references are included with Pseudofossils, algae, and microfossils were removed from the data set analyses. While the PDBD includes estimated paleo latitude and longitude, these were recalculated using the software GPlates and Wright et al. (2013) continental reconstruction models after vetting and correcting (where necessary) reported latitude and longitude for existing data. Paleo lat/long were estimated for all data at the midpoint of each assemblage zone each collection was identified to. Fossil occurrence data was assembled into paleocommunities by lumping collections within an estimated 5km radius (at the time of original deposition) of the same formation and member. Paleocommunity lumping was performed in R and these operations have been included in the R scripts included with this supplemental.