Mechanical vulnerability explains size-dependent mortality of reef corals

Abstract

Understanding life history and demographic variation among species within communities is a central ecological goal. Mortality schedules are especially important in ecosystems where disturbance plays a major role in structuring communities, such as coral reefs. Here, we test whether a trait-based, mechanistic model of mechanical vulnerability in corals can explain mortality schedules. Specifically, we ask whether species that become increasingly vulnerable to hydrodynamic dislodgment as they grow have bathtub-shaped mortality curves, whereas species that remain mechanically stable have decreasing mortality rates with size, as predicted by classical life history theory for reef corals. We find that size-dependent mortality is highly consistent between species with the same growth form and that the shape of size-dependent mortality for each growth form can be explained by mechanical vulnerability. Our findings highlight the feasibility of predicting assemblage-scale mortality patterns on coral reefs with trait-based approaches.

Keywords: Biomechanics; colonial; demography; disturbance; life history; mortality; reef coral.

Figure 1

Alternative models of coral mortality. Past studies have shown that whole-colony mortality rates tend to decrease monotonically with size [indicated by the arrow at (a)]. Monotonic decrease is hypothesised to occur because mortality agents act independently at the module (polyp) scale and therefore have a greater chance of killing smaller colonies and reducing the size of larger colonies [illustrated by the leftward arrow at (b)]. Storm wave mortality processes also cause whole-colony dislodgement and death, which can lead to higher mortality rates in large colonies for some species [upward arrow at (c)], potentially generating bathtub relationships of mortality rate with size.

Figure 2

Mortality rate as a function of colony size. Panels show the best-fit GLM for mortality (solid lines), along with the observations (grey points: 1 indicates mortality, 0 indicates survival; points have been randomly offset from 1 and 0 to help visualise densities with size). Where the growth form model was the best-fitting model, the data have been pooled across species (a–d). Arborescent species are plotted separately (with the same silhouette) as model selection favoured different mortality parameters (e, f). Dashed lines (sometimes concealed by the GLM fit) show best-fit GAMs for each species or growth form respectively. Yearly mortality rates (M) are given for each panel.

Figure 3

Relationships between colony shape factor (CSF) and colony size for the five study growth forms: arborescent (circles; n = 73), tabular (triangles; n = 76), corymbose (pluses; n = 78), digitate (crosses; n = 68) and massive (diamonds; n = 86). Lines show the best-fit multiple regression model, which was subsequently used to estimate CSF for the tagged colonies.

Figure 4

Mean annual mortality of the largest 20% of colonies for each species plotted against their mean estimated mechanical vulnerability (CSF). Abbreviations represent the first letter of the genus name and first two letters of the species name (see Methods for full species names). The dashed line shows the relationship between the variables when the two arborescent species (bottom right) are removed (Table 2b).