Climate change is generating extreme climate events, affecting ecosystem integrity and function directly through increases in abiotic stress and disturbance and indirectly through changes in the strength of biotic interactions. As consumers play an essential role in ecosystem functioning and have been shown to be highly sensitive to climate conditions, improved understanding of their role under changing environmental conditions is necessary to accurately anticipate climate change impacts on ecosystem integrity. We evaluated if prolonged heavy rains, a climatic event increasing in severity in many places around the world and coincident increases in coastal flooding duration intensify consumer control of foundational saltmarsh grass structure and quantify the consequences of flooding-consumer interactions on saltmarsh range extent. To achieve this, we analyzed: historic trends in crab grazing; crab numbers and activity in and out of rainy years on the low marsh edge; vegetation retreat from the low marsh edge at a plot-scale in a manipulative exclosure experiment; vegetation retreat at a landscape-scale from drone image analyses; and the vertical erosion in the lowest edge of an Argentinean salt marsh. During flooded periods, crabs congregated in the low marsh, resulting in localized overgrazing of saltmarsh grass and the rapid horizontal retreat of the marsh edge (98.5 cm on average). Saltmarsh edge retreat resulted in a loss of ~4.5% of the total marsh area at the landscape-scale. Inside crab exclusion plots, although grass cover declined slightly during the study period, the marsh edge did not retreat. Synthesis: This study provides experimental evidence that an extreme climate event can destabilize a local consumer-prey interaction, indirectly triggering the range contraction of a critical coastal habitat. This work contributes to a growing body of research demonstrating that consumers can be unleashed, rather than suppressed, by extreme climatic events. Moreover, in cases where consumer fronts form during such events, the result can be not only local (along habitat edges) but also landscape-scale extinction of foundation species and the habitats they biogenically create. Together, this supports the more general call that models of future climate scenarios integrate the indirect effects on ecosystem-regulating food web interactions. --

1-  Are extreme events that cause longer lasting flooding becoming more frequent?

To evaluate historical and contemporary patterns of extreme rainfall events (and their potential association with crab grazing), we obtained daily precipitation values from the Argentinean National Weather Service (Servicio Meteorológico Nacional Argentino) for the Mar del Plata station (37° 56’ S; 57° 35’ W), located 25 km south of our study site. We analyzed monthly total extreme precipitation (mm) for a 16-years period (between November 2002 and December 2018), for which we have crab herbivory data (see next section). For this analysis, we only summed the days with more than 50 mm of rain (extreme precipitation) in the monthly extreme precipitation totals.

2- Can these extreme rainfall events induce crab consumer front formation?

Crab feeding activity was estimated during two consecutive years, one with typical rainfall events (2013: 815 mm annually), and another after a very rainy event (2014: 1255.2 mm annually). The observation area comprised five, 5 m² (0.5 × 10 m) plots in the saltmarsh edge and five plots of the same size in the mudflat perpendicular to the shore that were staked on the corners. Observations of surface crab activity were carried out with binoculars (Bushnell 10×50 at a 10 m distance) every 40 minutes along the tidal cycle (from low tide to high tide). This methodology was repeated for 5 consecutive days in early September each year. Total sampling effort (time spent on surveys) was the same for each plot and day. During these surveys, we recorded the total number of crab individuals feeding on the surface per hour in each habitat.

We supplemented the crab grazing observation surveys with surveys of crab herbivory by summarizing the percentage of live Spartina densiflora leaves with signs of herbivory (i.e. lacking their tips and/or with missing tissue along leaf edges; see Alberti, Cebrian, et al., 2011) per tiller during 2002-2005, 2012, 2017 and 2018 from historical data sets collected by members of our research group and others at this study site. Since leaf production and crab herbivory are seasonally variable (Alberti, Cebrian, et al. 2011; J. Alberti unpubl. Data), we compared three years for which we had herbivory data for the same period (August-September) and that annual precipitation was remarkably different (2003: 1077.7 mm, 15 % above the mean; 2004: 739.7 mm, 21 % below the mean; 2017: 1420.5 mm, 52 % above the mean). We used these data to evaluate if there were differences in crab herbivory in the lowest edge of the salt marsh between years with contrasting annual rainfall.

3- What is the separate and combined effect of extreme precipitation and crabs on vegetation biomass?

In May 2016 we initiated a crab exclusion experiment that lasted two years (until June 2018). Twenty 0.25 m² areas, standardized for plant cover, composition, and density, located in the lowest edge of the saltmarsh and randomly assigned each to one of the following two treatments: (1) control, which were unmanipulated areas delimited with wooden stakes in two opposite corners, and (2) crab exclusion, that consisted in exclosure plots delimited with plastic mesh fences (50 x 50 x 40 cm, 1 cm² opening; N= 10 replicates per treatment). In June 2018, we harvested all the aerial plant biomass, separated it into live and dead material, oven-dried and weighed it.

4- Can crabs cause the retreat of the lowest salt marsh edge and, hence, marsh range contraction?

We also estimated if crab herbivory was intense enough to cause a retreat of the marsh edge at a plot scale. Thus, we measured the distance of the straight line perpendicular to the shore that connected the location of the saltmarsh edge at the beginning of the experiment and at the end, for each exclosure plot given that remained as islands of live plants in a matrix of short dead tillers (along 57 m of the marsh; see Supplementary Figure S1). The saltmarsh edge can easily be defined by an abrupt change in vegetation cover (from > 40 % to 0 %). Positive values indicate saltmarsh plant expansion into the mudflat, while negative values would denote a retreat of the saltmarsh edge.

Given that the edge showed a clear retraction (see Results), and that the line where the vegetation was at the beginning of the experiment was easily identified due to the exposed rhizomes or short dead tillers, we used an RTK drone flying at 35 m height to obtain high resolution images of the saltmarsh edge (1.5 cm pixel^-1, see Supplementary Figure S2) in May 2019. With the resulting orthomosaic (i.e. combination of high resolution images into a single one spanning a much longer spatial extent), we measured marsh edge retraction and estimated relative elevation were the vegetation was at the beginning and were it was after the extreme rainfall event on 50 points along 250 m of the marsh edge (i.e. at the ’landscape scale’). To evaluate if saltmarsh retreat was related to edge elevation (e.g. if lower elevation areas are inundated more and, hence, more exposed to crab herbivory), we analyzed whether edge movement was related to the initial relative intertidal elevation of the vegetation edge (i.e. before the extreme rainfall event). We obtained the relative difference in intertidal elevation before and after the extreme rainfall event from the drone images.

5- How do changes in vegetation due to precipitation-crab interactions modify vertical erosion?

One year after the end of the experiment (June 2019), we estimated the vertical erosion in different points of our experimental area. We quantified the vertical distance between sediment surface and rhizomes in zones formerly covered by S. densiflora using a ruler.  We sampled two different conditions: areas where vegetation was lost, and areas where vegetation remained as isolated patches (see Figure 1 for zones without vegetation). We repeated this measurement in a more erosive zone, adjacent to our experimental area, where we expected higher vertical erosion after vegetation loss (see Supplementary Figure S4). Positive values (i.e. rhizomes exposed above the sediment surface) indicate vertical erosion, while negative values denote that the rhizomes were still covered by surficial sediments and thus belowground.