Predicting seagrass decline due to cumulative stressors

Highlights

• Easy-to-use program to predict cumulative light and temperature stress on seagrass.

• Software predictions made from a new process-based model of tropical seagrass decline.

• Model suggests net carbon loss rate controls shoot density decline rate in seagrass.

• Model calibrated to data via two posterior-computation methods for Bayesian inference.

• New generalisable cumulative stress index forecasted by model, including uncertainty.

Abstract

Seagrass ecosystems are increasingly subjected to multiple interacting stressors, making the consequent trajectories difficult to predict. Here, we present a new process-based model of seagrass decline in response to cumulative light and temperature stress. The model is calibrated to laboratory datasets for Great Barrier Reef seagrasses using Bayesian inference. Our model, which is fit to both physiological and morphological data, supports the hypothesis that physiological carbon loss rate controls the shoot density decline rate of seagrasses. The model predicts the time to complete shoot loss, and a new, generalisable, cumulative stress index that indicates the potential seagrass shoot density decline based on the time period of cumulative stress. All model predictions include uncertainty estimates based on uncertainty in the model fit to the data. The calibrated model is packaged into a computer program that can forecast the potential declines of seagrasses due to cumulative light and temperature stress.

Introduction

Seagrasses are critical habitats for conservation (Unsworth et al., 2018) since they form the foundation of many temperate and tropical shallow water ecosystems worldwide (Hughes et al., 2009). Globally, seagrass loss is accelerating (Waycott et al., 2009), and the cumulative impacts of multiple stressors (Grech et al., 2011; Brown et al., 2014; Ontoria et al., 2019) play an important role in determining current and future seagrass ecosystem state (Unsworth et al., 2015; Griffiths et al., 2020). For the effects of single stressors on seagrass ecosystems, threshold values can be identified (Lee et al., 2007) which are easily communicable to managers, including minimum light requirements (Erftemeijer and Lewis, 2006; Collier et al., 2016; Chartrand et al., 2016), upper temperature limits (Pedersen et al., 2016; Adams et al., 2017; Collier et al., 2017), upper wave energy limits (Uhrin and Turner, 2018), and timescales of change (Adams et al., 2015; Lambert et al., 2019) and decline (O'Brien et al., 2018). These individual thresholds can be directly used to suggest when seagrass is at risk of substantial decline, but may not capture the potential declines caused by synergistic interactions between stressors that individually do not surpass a threshold. Predicting seagrass decline due to cumulative stressors requires knowledge of how the timescales of loss for seagrass quantitatively depend on these stressors, although this relationship may be complicated for multiple interacting stressors. The problem of synergistic interactions can be addressed by developing a mathematical model that takes into account the nonlinear interactions between stressors and outputs the cumulative impact of these stressors on seagrass decline. It is not feasible to take into account every possible stressor, so a first step is to limit consideration to a few (important) stressors.

The most critical stressor for seagrass is light limitation (Duarte, 1991; Koch, 2001): light affects seagrass growth and decline via modification of photosynthesis rate, a mechanism which operates at the physiological scale (Waycott et al., 2005; McMahon et al., 2013). The rate of photosynthesis controls the rate of carbon gain by the plant, and carbon gain is typically considered the rate-limiting step for plant biomass accumulation (Poorter et al., 2013). Temperature is another important environmental factor controlling seagrass distribution via physiological processes (Collier and Waycott, 2014) that will become increasingly critical as oceans warm (Jordà et al., 2012). In seagrasses, temperature modifies the balance of net carbon exchange via its effects on both photosynthesis (carbon gain) and respiration (carbon loss) (Staehr and Borum, 2011; Collier et al., 2017). Several other carbon loss processes also affect the balance of carbon exchange in seagrasses, including leaf loss, root and rhizome decay, and dissolved organic carbon exudation, although it is unclear if the rates of these loss processes also depend on light or temperature (Kaldy, 2012). Regardless of the exact pathways by which light and temperature alter the balance of carbon exchange, there is recent experimental evidence to suggest that the stress response of seagrass to multiple drivers is governed by the degree of imbalance in its internal carbon budget (Moreno-Marín et al., 2018). Stress responses at the morphological scale would be seen by reductions in biomass or shoot density; and these two metrics of seagrass health are positively-correlated (Vieira et al., 2018). A quantitative link between physiological processes (net carbon exchange) and morphological responses (rate of change in seagrass biomass or shoot density) is commonly assumed in dynamic models of seagrass growth (e.g. Baird et al., 2016), although this quantitative link has not yet been clearly established.

If the interactions between cumulative stressors and the effects of these interactions on seagrass morphology are quantified, reporting these effects requires the development of easily-communicable indicators or metrics that assess the potential stress on seagrass. There are several indicators and metrics that have been proposed for seagrass stress, based primarily on statistical analysis of seagrass responses across many systems (Roca et al., 2016). However, to our knowledge a metric of potential stress to seagrass which integrates cumulative effects of multiple stressors has not yet been proposed.

Metrics of cumulative stress can potentially be output from ecological models that specifically account for the interactions of multiple stressors. Any output metrics of models that are directly relevant to environmental management should also ideally include uncertainty estimates, to ensure that model forecast performance is honestly represented (Dietze et al., 2018). Approaches which merge statistical and mechanistic modelling can estimate uncertainty in predictions but, until recently, such approaches have been a fairly rare occurrence for modelling aquatic systems (Robson, 2014).

In this paper, we address all of the above issues by introducing a new process-based model for seagrass decline that (1) takes into account the cumulative impacts of light and temperature, (2) quantitatively connects these stressors’ effects on seagrass physiology through to predictions of morphological decline responses, and (3) outputs a new, generalisable, metric of cumulative stress that includes robust uncertainty estimates. Laboratory datasets for physiological (i.e. net photosynthesis) (Collier et al., 2018) and morphological (i.e. shoot density) (Collier et al., 2016) responses of three tropical seagrass species to different light and temperature conditions were used to inform the model. The model exploits both datasets to investigate how seagrass physiological and morphological responses may be connected, specifically by investigating the hypothesis that the net carbon loss rate is the primary control of shoot density decline rate.

To account for uncertainty in the model-data fit, the model was calibrated to the aforementioned datasets using Bayesian inference methods; these methods are advantageous because they account for uncertainty propagation (Girolami, 2008). Specifically, for model-data calibration we utilised and compared Sequential Monte Carlo sampling (Doucet et al., 2000) with doubly stochastic variational inference (Titsias and Lázaro-Gredilla, 2014), both of which are posterior-computation methods for Bayesian inference.

Our seagrass decline model outputs a new metric of cumulative stress (cumulative stress index, or CSI), which broadly indicates the expected decline based on a particular combination of environmental stressors and the time period over which the stress occurs. The CSI estimates stress to the seagrass as a percentage. For example, a CSI of 0% indicates no stress (i.e. no expected decline), a CSI of 50% indicates that half of the seagrass shoots are expected to be lost, and a CSI of 100% indicates a prediction of complete shoot loss. This intuitive metric for cumulative stress is therefore conceptually similar to degree-heating-weeks for corals (Gleeson and Strong, 1995; Kayanne, 2017), except that CSI is potentially more generalisable because of its non-dimensional units (% stress) and because it can represent the effects of stressors unrelated to temperature.

The end product of this research is an easy-to-use computer program: the user chooses the seagrass species of interest and inputs specific details about the local light and temperature environment, and the program predicts cumulative stress on the seagrass, including uncertainty bounds in predictions. The program is of particular benefit when the stress on seagrass predicted by individual light or temperature values alone is insufficient to indicate that seagrass may decline due to the cumulative effects of these stressors.