Bleaching response of coral species in the context of assemblage response

Abstract

Caribbean coral reefs are declining due to a mosaic of local and global stresses, including climate change-induced thermal stress. Species and assemblage responses differ due to factors that are not easily identifiable or quantifiable. We calculated a novel species-specific metric of coral bleaching response, taxon-α and -β, which relates the response of a species to that of its assemblages for 16 species over 18 assemblages. By contextualizing species responses within the response of their assemblages, the effects of environmental factors are removed and intrinsic differences among taxa are revealed. Most corals experience either a saturation response, overly-sensitive to weak stress (α > 0) but under-responsive compared to assemblage bleaching (β < 1), or a threshold response, insensitive to weak stress (α < 0) but over-responsive compared to assemblage bleaching (β > 1). This metric may help reveal key factors of bleaching susceptibility and identify species as targets for conservation.

Introduction

Anomalously warm water disrupts mutualistic endosymbioses between photosynthetic dinoflagellates (genus Symbiodinium) and coral hosts through the bleaching response, eliminating the primary mode of energy acquisition for corals (Muscatine 1990) and leading to the degradation of reef ecosystems globally (Wilkinson 2008). Bleaching responses differ across individuals, taxa, and events (e.g., Wooldridge 2014) through an unknown combination of factors intrinsic to the coral-Symbiodinium holobiont (e.g., hosting Symbiodinium phylotypes with differential thermotolerance; e.g., Barshis et al. 2013) or extrinsic to the holobiont (e.g., differential thermal anomalies across reefs; e.g., Ban et al. 2014) with the added complexity of coral acclimation or adaptation to thermal stress (e.g., Guest et al. 2012; Howells et al. 2013).

To isolate the effect of factors intrinsic to the holobiont, we propose a novel metric of bleaching response (taxon-α and -β) that relates species-specific susceptibility to that of the assemblages where they were surveyed. This metric is contextualized by the responses of the assemblage of corals; if all coral species are equally susceptible the response of each species would mirror its assemblage and react to bleaching stimulus and history of environmental conditions at that site identically. We argue that deviation from the bleaching response of the assemblage should mostly reflect the effects of intrinsic factors on differential bleaching. Parameters α and β were originally derived to model the effect of external and internal factors on differential bleaching susceptibility among corals (Swain et al. 2016a). Here we present taxon-α and -β as a species-specific metric of bleaching response calculated using 125 standardized records for 16 taxa at 18 Caribbean sites (Swain et al. 2016a) which experienced thermally induced bleaching events separated by space and time, and were therefore exposed to dissimilar stress events and environmental histories. Using these data, we ask: (1) is the bleaching response of a coral species or assemblage a reflection of the thermal-stress anomalies they are exposed to; and (2) does the bleaching response of species differ from that of the assemblages where they are found (i.e., do species over-react or under-react to similar thermal stress and environmental history)?

Methods

Data collection

Standardized records of bleaching and mortality for Caribbean corals were culled from Swain et al. (2016a). These included 371 bleaching response records (B_jk, or bleaching response of taxon j at site k), that are a composite response metric of bleaching and associated mortality that occurred following a thermal event (but not distal effects such as disease or repeated bleaching) standardized across survey methods and bleaching criteria (Swain et al. 2016a), for 39 taxa across 55 Caribbean sites from 1987 to 2006 (Electronic supplementary material, ESM, Table S1). Since bleached corals may either die or recover, and observation timing may bias the data, both bleaching (greater in early surveys) and associated mortality (greater in later surveys) were used to calculate B_jk and provide a more accurate assessment. Thermal anomaly records, reported as degree heating weeks (DHW), for matching dates and locations were culled from the U.S. National Oceanic and Atmospheric Administration coral reef watch satellite monitoring database (http://coralreefwatch.noaa.gov). DHW is the product of degrees (°C) above the highest monthly mean sea surface temperature and its duration in weeks,and bleaching is considered likely at DHW > 4 and mass bleaching and mortality at DHW > 8 (http://coralreefwatch.noaa.gov). We used maximum DHW within 6 months prior to survey to capture peak thermal stress. for 42 sites from 1998 to 2006, which included 39 taxa in 287 records.

Data analysis

Coral assemblage-level bleaching response (site-BRI or SBRI_k), was calculated as the mean of individual bleaching and mortality responses (B_jk) for all taxa j at a given site k (Swain et al. 2016a) as:

$${\text{SBRI}}_k=\frac{1}{J_k}{\sum }_{j=1}^{J_k}{B}_{jk}$$ (1)

where J_k is the total number of taxa in site k.SBRI_k was calculated for coral assemblages with ≥ 3 taxa (363 B_jk records of 39 coral taxa at 49 sites from 1987–2006).

The effect of thermal stress on bleaching response was assessed as the relationship between taxon-specific bleaching responses (B_jk) and thermal anomaly (6 month maximum DHW) at sites with ≥ 3 taxa (279 B_jk records of 39 coral taxa at 36 sites from 1998–2006).

To compare bleaching response of taxon j with that of its coral assemblage without the influence of target taxon j, the response of j was removed from the mean bleaching response of the assemblage (site-B_jk or sB_jk) (Swain et al. 2016a) as:

$${\text{sB}}_{jk}=\frac{1}{J_k-1}{\sum }_{i=1,i\ne j}^{J_k}{B}_{ik}$$ (2)

where J_k is the total number of taxa in site k. Calculation of sB_jk was restricted to sites containing ≥ 3 taxa with variable bleaching responses (363 B_jk records of 39 taxa at 49 sites from 1987-2006).

Sensitivity of individual taxon-specific bleaching responses, B_jk to that of the entire assemblage can be modeled (Swain et al. 2016a) as:

$$B_{jk}=u_{jk}+{\beta }_{j}s{B}_{jk}+{\alpha }_j,$$ (3)

where β_j is the taxon-specific bleaching sensitivity (taxon-β), u_jk is variability in estimation of B_jk due to error of measurements with Var [u_jk] = Var [δB_jk], sB_jk is described in Eq. 2, and coefficient α_j is a taxon-specific parameter describing species-specific sensitivity to weak stress (e.g., seasonal Symbiodinium density variation; Fitt et al. 2000) in the absence of assemblage response (taxon-α). Taxon-β can be directly measured as the linear regression coefficient in the regression of B_jk on sB_jk as:

$${\beta }_j=\frac{co{v}_k\left(B_{jk},s{B}_{jk}\right)}{va{r}_k\left(s{B}_{jk}\right)}={\rho }_k\left(B_{jk},s{B}_{jk}\right)\frac{{\sigma }_k\left(B_{jk}\right)}{{\sigma }_k\left(s{B}_{jk}\right)}$$ (4)

where Cov and Var are covariance and variance operators, ρ is the correlation coefficient, and σ is standard deviation, all estimated for a given taxon (j = constant). Taxon-β ∼1 indicates a rate of change in bleaching response of taxon j that is similar to the assemblages where it is found, taxon-β < 1 indicates an under-response relative to the rate of change of its assemblages, and taxon-β > 1 indicates an over-response relative to the rate of change of assemblages. Taxon-β was calculated for taxa surveyed at ≥ 3 sites in assemblages with ≥ 3 taxa (125 B_jk records of 16 taxa at 18 sites from 1987–2006).

Results and discussion

Is bleaching response of a coral species or assemblage a reflection of the thermal-stress anomalies they are exposed to?

Thirty-six Caribbean coral assemblages (sites with ≥ 3 taxa; 7.75 ± 4.55 [mean ± SD] taxa per site, 39 total unique taxa) experienced thermal anomalies ranging 0–15.7 DHW from 1998–2006 (Fig.1; ESM Table S1). Mean bleaching response of all taxa surveyed in a site (site-BRI, or SBRI_k; Eq. 1) varied 19-fold among assemblages (SBRI_k mean 23.0% ± 16.34, range 2.75–52.9%; ESM Table S1; Fig.1). Bleaching response variability across sites has been used to identify factors affecting that response (e.g., water flow, thermal history, solar irradiance, and acute thermal stress; McClanahan et al. 2005; Howells et al. 2013; Marcelino et al. 2013; Swain et al. 2016a,b).

Fig. 1.

Response of coral assemblages (SBRI_k) at 36 Caribbean sites (numerals correspond to site numbers in Electronic supplementary material Table S1) to thermal anomalies (6 months max degree heating weeks: DHW). Site bleaching response increases with thermal anomalies (r² = 0.22, p < 0.01; a), across the Caribbean at DHW of >8 (b), 4–8 (c) and < 4 (d). Shading of site markers corresponds to the gradient of SBRI_k values, with lightest markers representing highest bleaching response values. Maps constructed using Mapbox and OpenStreetMap

Although there was a significant positive correlation between DHW and SBRI_k, DHW was a poor predictor of SBRI_k (linear regression: r² = 0.21, p = 0.005, n = 36 sites; Fig. 1a), which is consistent with other studies of Caribbean corals (e.g., Yee et al. 2008). Similarly, thermal anomalies significantly increased bleaching response of individual taxa, but DHW was a poor predictor of individual bleaching response (r² = 0.10, p < 0.001, n = 287, 39 taxa; ESM Fig. S1). These results suggest that differential bleaching among assemblages is only partially explainable by exposure to different thermal anomalies, as indicated by other studies (e.g., McClanahan et al. 2015). We argue that, as a measure of assemblage bleaching response at a site, SBRI_k should reflect the intensity and duration of thermal anomalies, thermal history of the site, chronic and acute environmental stresses that were not directly observed, and capacity of the holobiont to resist thermal stress.

Does the bleaching response of species differ from that of the assemblages where they are found?

Taxon-β, which is the regression coefficient relating bleaching response of a species (B_jk) to the response of its assemblages (sB_jk) (Eq. 4), was determined for 16 species (limited to taxa surveyed in ≥ 3 sites, and sites with ≥ 3 taxa; Fig. 2; Table 1). Relative to the bleaching response of its assemblages, species: (1) over-responded, i.e., increased bleaching response at a greater rate than their assemblages (e.g., Acropora palmata, taxon-β > 1); (2) under-responded, i.e., increased bleaching response at a lesser rate than their assemblages (e.g., Diploria labyrinthiformis, taxon-β < 1); or (3) responded similarly (e.g., Siderastrea siderea, taxon-β ∼ 1) (Fig. 2; Table 1). Because taxon-β accounts for the bleaching response of a taxon in relation to its assemblages, it removes the effects of environment and history on the response of species and reveals the intrinsic (physiological) differences between taxa.

Fig. 2.

Taxon-β for Caribbean corals. Responses of corals (B_jk) relative to their assemblages (sB_jk) for 16 coral species (a) and exemplars of species with taxon–β > 1 indicating over-response (b), taxon–β < 1 indicating under-response (c), and taxon–β ∼1 indicating similar response (d)

Table 1. Taxon-β and taxon-α values and associated errors, sample sizes, and significance for 16 coral species.

Taxon	Taxon β	Taxon α	Std error	Correlation coeffficient (r2)	No of sites	Mean taxa/site	p-value
Montastraea cavernosa	0.15	10.25	0.38	0.02	9	12.7	0.708
Porites porites	0.15	23.85	0.39	0.02	9	13.2	0.709
Madracis myriaster	0.22	-1.25	0.33	0.13	5	15.4	0.56
Porites astreoides	0.29	14.76	0.43	0.04	13	10.8	0.509
Madracis decactis	0.37	-5.75	0.12	0.72	6	15.5	0.033
Meandrina meandrites	0.48	9.94	0.50	0.15	7	14.7	0.386
Agaricia agaricites	0.70	33.54	1.41	0.04	8	12.4	0.635
Diploria labyrinthiformis	0.79	-6.37	0.24	0.58	10	11.9	0.011
Orbicella faveolata	1.06	7.56	1.31	0.12	7	13.3	0.455
Colpophyllia natans	1.12	-10.24	0.34	0.61	9	12.6	0.013
Orbicella annularis	1.16	-6.42	0.60	0.25	13	10.0	0.08
Siderastrea siderea	1.18	-10.00	0.21	0.82	9	12.9	0.0007
Diploria strigosa	1.28	-13.62	0.47	0.51	9	12.6	0.031
Favia fragum	1.46	10.15	0.58	0.76	4	9.0	0.127
Diploria clivosa	2.38	-37.04	0.55	0.95	3	17.3	0.145
Acropora palmata	2.53	-30.29	0.98	0.77	4	6.0	0.124

Although taxon-β quantifies relative differences in the rate of bleaching response increase, taxon-α further contextualizes the response of the target taxon by indicating its species-specific sensitivity to weak stress in the absence of assemblage response (Table 1). Considered together, the 16 Caribbean species fall under two main categories of response with a trade-off between taxon-α and taxon-β: (1) corals that are overly-sensitive to weak stress in the absence of assemblage bleaching (taxon-α > 0) but under-responsive to increasing thermal stress compared to assemblage bleaching response (taxon-β < 1, Fig. 3a quadrant II; Fig. 3b); or (2) corals that are insensitive to weak stress in the absence of assemblage bleaching (taxon-α < 0) but over-responsive to increasing thermal stress compared to assemblage bleaching response (taxon-β > 1; Fig. 3a, quadrant IV). Additionally, a few species were either overly-sensitive and over-responsive relative to their assemblages (taxon-α > 0 and taxon-β > 1; Fig. 3a quadrant I) or insensitive and under-responsive relative to their assemblages (taxon-α < 0 and taxon-β < 1; Fig. 3a quadrant III).

Fig. 3.

Relationship between taxon-α and taxon-β for Caribbean corals (r² = 0.56, p < 0.001; a), with an exemplar species (b) demonstrating a relatively high α (9.94) but low β (0.48) values typical of quadrant II (a)

Bleaching responses of corals with a trade-off between taxon-α and taxon-β proceed on opposing trajectories (Fig. 3a quadrants II and IV). Corals within the high taxon-α and low taxon-β category (Fig. 3a quadrant II, Fig. 3b) demonstrate a saturation bleaching response where corals over-respond compared to the baseline assemblage response, but under-respond as the assemblage response increases. Corals within the low taxon-α and high taxon-β category (Fig. 3a quadrant IV) demonstrate a threshold bleaching response where corals under-respond compared to the baseline assemblage response, but over-respond as the assemblage response increases. Survival of a mild or short thermal-stress event would be of greater concern for corals with a saturation bleaching response, while survival of a severe or long thermal-stress event would be of greater concern for corals with a threshold bleaching response. Corals within the high taxon-α and high taxon-β category (Fig. 3a quadrant I) would be considered as taxa of highest concern (due to consistent over-response), while those within the low taxon-α and taxon-β category (Fig. 3a quadrant III) would be considered as taxa of least concern (due to consistent under-response)..

Limitations and general considerations

Accuracy of taxon-α and -β (standard error, p-value of regression, and y-intercept; Table 1) depends on the number of sites where a taxon is surveyed and on accuracy with which assemblage bleaching response is determined. The latter, in turn, depends on (1) the number of taxa surveyed at each site (which was the primary factor determining accuracy in our dataset), (2) how representative those taxon responses are of assemblage response, and (3) other factors including differences in survey methods and bleaching criteria which have been accounted for by the standardization of individual B_jk records (Swain et al. 2016a). A low p-value of the linear regression corresponds to a smaller standard error of the estimation of taxon-β (Table 1). Although some of the regressions are significant (p < 0.05), many are non-significant due, in part, to small site sample sizes. The accuracy of taxon-α and -β should increase as more surveys of bleaching response become available, particularly with inclusion of sites with a large number of taxa presenting a wide range of bleaching responses.

By relating the response of a coral to its assemblage, extrinsic factors in the bleaching response are removed and the effect of individual intrinsic factors can be assessed through analysis of variance of taxon-α and -β. Individual intrinsic factors could be identified experimentally by their effect on the bleaching response, and by ranking their importance as bleaching determinants by the portion of taxon-α and -β variance that they could explain. If the number and total effect of intrinsic factors are known, the approximate upper bound on the effect of the remaining intrinsic factor could be inferred. However, individual intrinsic factors may not act independently (e.g., the effect of Symbiodinium thermotolerance may be influenced by skeletal light scattering; Swain et al. 2016b) and extrinsic factors such as thermal history may lead to acclimation or adaptation of corals (Guest et al. 2012; Pratchett et al. 2013).

The intrinsic capacity of the holobiont to resist stress (taxon-β) and its species-specific stress sensitivity in the absence of assemblage response (taxon-α) provide tools for comparative assessments of the connection between species traits and bleaching responses to inform conservation and management efforts. Estimates of variance in bleaching response among corals are critical for assessing the resilience of individual reefs and coral species to climate change (Maynard et al. 2015). These new metrics can be expanded and updated as additional bleaching records become available and can target locations in time or space. Changes in bleaching response over time could be evidence of adaptation to climate change (Guest et al. 2012; Pratchett et al. 2013) and taxon-α and -β could be used to identify suitable candidate species for interventive conservation efforts, such as selective breeding or assisted migration (van Oppen et al. 2015).

Supplementary Material

338_2017_1550_MOESM1_ESM

Figure S1. Response of individual corals (B_jk) to thermal anomalies (6 mo max DHW).