Heterogeneity around CO2 vents obscures the effects of ocean acidification on shallow reef communities

Abstract

Abstract Studies that use CO2 vents as natural laboratories to investigate the impacts of ocean acidification (OA) typically employ control-impact designs or local-scale gradients in pH or pCO2, where impacted sites are compared to reference sites. While these strategies can accurately represent well-defined and stable vent systems in relatively homogenous environments, it may not adequately encompass the natural variability of heterogeneous coastal environments where many CO2 vents exist. Here, we assess the influence of spatial heterogeneity on the perceived impacts of OA at a vent system well established in the OA literature. Specifically, we use a multi-scale approach to investigate and map the spatial variability in seawater pH and benthic communities surrounding vents at Whakaari-White Island, New Zealand to better understand the scale and complexity of ecological impacts of an acidified environment. We found a network of vents embedded in complex topography throughout the study area, and spatially variable pH and pCO2 levels. The distribution of habitats (i.e. macroalgal forests and turfing algae) was most strongly related to substratum type and sea urchin densities, rather than pH. Epifaunal communities within turfing algae differed with sampling distance from vents, but this pattern was driven by higher abundances of a number of taxa immediately adjacent to vents, where pH and temperature gradients are steep and white bacterial mats are prevalent. Our results contrast with previous studies at White Island that have used a control-impact design and suggested significant impacts of elevated CO2 on benthic communities. Instead, we demonstrate a highly heterogeneous reef where it is difficult to separate effects of reduced pH from spatial variation in reef communities. We urge that future research carefully considers and quantifies the biological and physical complexity of venting environments, and suggest that in dynamic systems, such as White Island, the use of control-impact designs can oversimplify and potentially overestimate the future impacts of OA.