Ranking ecological contingencies from high-order factorial data demonstrate tidy control of biodiversity from facilitation cascades in estuaries on the South Island of New Zealand

Community ecology has been described as a ‘mess’ because ecological processes vary in space, time, and across species traits, resulting in myriads of ecological contingencies and low scientific predictability. Here we aimed to identify and rank ecological contingencies and improve predictability, using fully crossed high-order factorial mensurative and manipulative experiments across axes of spatiotemporal variability that may influence the strength of facilitation cascades on the South Island of New Zealand. Facilitation cascades, arising from chains of positive interactions, are prevalent in intertidal sedimentary estuaries, where biogenic habitat-formers, like bivalves and attached seaweed, provide hard substrates, food, hiding places, and reduce environmental stress for small animals. Specifically, we measured facilitation of >65,000 small mobile invertebrates across eight archetypical contingencies, i.e., within and between seasons, latitudes, sites with different distances to the open ocean, vertical intertidal elevations, wider habitat matrix (bare sediment vs. seagrass beds), and between small-scale habitat-forming species (the endemic cockle Austrovenus and attached cosmopolitan seaweeds, Ulva and Gracilaria spp.) and their sizes. Overall, our multifactorial tests revealed that most higher-order interactions (three-way or more) were not important ecologically, and that many important lower-order interactions (two-way) were ‘simple’, demonstrating that facilitation can increase when and where the baseline biodiversity is higher. Furthermore, most of the main test factors were significant and ecologically important, suggesting that facilitation of animals, generally and across other factors, were strongest on large and morphologically complex seaweeds, at near-ocean sites and deeper intertidal elevations, and in warmer summer months. Our case study highlights a relatively tidy – not messy – control of biodiversity of intertidal epifauna, and that high-order factorial sampling can help unravel and rank co-occurring spatiotemporal drivers to better understand ecological contingencies. Finally, our results may also inform management of habitat-forming species to preserve estuarine biodiversity and maintain their secondary production.

Methods
This dataset was collected through high-order crossed-factorial field surveys and field experiments conducted in estuaries accross the South Island of New Zealand to measure biodiversity responses of small mobile invertebrates associated with co-occurring cockles and seaweeds. Individual cockles (Austrovenus stutchburyi) were collected with or without (controls) attached seaweeds (Ulva or Gracilaria). Labeled samples were transported chilled and stored frozen at -20 ºC. In the laboratory, samples were thawed and rinsed with freshwater in a 250 µm sieve. Detached inhabitants were transferred into a counting chamber (Bogorov 1927) and identified to lowest taxonomic level possible, and counted under a stereomicroscope with up to 40x magnification. All samples were stored in vials with 70% ethanol solution and labeled accordingly for preservation. The cockles and separated seaweed were weighed after oven-drying at 55 ºC for 72 h. Small and large Gracilaria were ≤0.6 and >0.6 gDW, whereas small and large Ulva (with a lower DW:WW ratio) were ≤0.35 and >0.35 gDW, respectively. Comprehensive data recording was maintained throughout the data processing workflow by systematically logging all relevant information in a dedicated spreadsheet. This included (1) background data such as sample ID, spatial, temporal, and habitat factors, (2) dry weights of cockles and seaweeds, and (3) individual counts of identified invertebrates. 

Note: A small subset of this dataset contains data from an already published dataset in Dryad: https://doi.org/10.5061/dryad.dr7sqvb5t. This subset is part of the authors' previously published work in Ecosphere. The current dataset associated with the Ecography paper is significantly larger, containing extensive data collected beyond the scope of the previous study.