Fig 1.
Methods of this paper divided into three major phases.
(A) Data collation involved acquisition of habitat data (green box), boundaries of marine protected areas (MPAs) (blue box), and observed and projected data (red and orange boxes, respectively) on sea-surface temperature (SST). In the selection and calculation of metrics of thermal stress (B), we derived metrics of chronic and acute stress from observed and projected datasets and combined them to define thermal-stress regimes. Regimes were delineated based on upper and lower terciles labelled as “high” (highest 33% of values, dark red or orange) and “low” (lowest 33% of values, light red or orange), respectively. The incorporation of warming disturbances into conservation planning (C) consisted of setting conservation objectives for each thermal-stress regime, evaluating their achievement in existing MPAs, and identifying priority areas that would achieve unmet objectives. Arrows in gray indicate the flow of information and lighter boxes linked by dashed lines depict types of data or analyses involved in each step.
Fig 2.
The study area and the chronic stress metric.
(A) Sectors (northern, central, southern), reef cells (n = 428), and the existing MPA boundaries along the Brazilian coast. MPAs are classified according to their main management categories: no-take areas and multiple-use areas. Letters a-e with stars denote approximate locations of reef cells selected to depict temperature variability (see Fig 6). (B) Decadal SST trends describe observed chronic stress for each reef cell from NOAA satellite data. (C) Decadal SST trends describe projected chronic stress for each reef cell, downscaled from PCM1 general circulation model output.
Fig 3.
Conceptual illustration of the effects of acute thermal stress on coral-reef ecosystem state described by a logistic function.
Empty circles indicate values used to fit the model. The form of this function assumes that the time that coral reefs spend with reduced ecosystem function (capacity to grow, repair, and reproduce), tc, is short at low DHW values because we expect that corals would recover quickly (within one year). This is followed by a steeper increase in tc; when widespread mortality begins (DHW reaches 8°C-weeks), the time that corals would spend recovering increases rapidly as bleaching-level events intensify above this level. When almost the entire community is extirpated over large spatial scales (above about 16°C-weeks), we expect to have small increments of tc with increasing DHW for the ecosystem as a whole because only stress-tolerant species that can withstand greater acute disturbance are present. The function then reaches an upper bound (i.e., in the formulae—asymptotic value—equal to 20 years) which is the maximum time required to regenerate a fully functional ecosystem after bleaching causes massive mortality and extirpates all organisms.
Fig 4.
The nine thermal-stress regimes defined within our study area (A), defined by combinations of high and low values for observed (Ob) and projected (Pr) chronic (Ch) and acute (Ac) stress. The rationale for management of each regime is summarised in (B). (C) Conservation objectives (dark green vertical lines) for each thermal-stress regime and their coverage by MPAs (green bars). Objectives prescribe the percentage of the total extent of reef cells in the regime (100%, 50%, or 30%) requiring management and the type of management required (no-take—solid green lines; multiple-use areas—dashed green lines). Horizontal bars indicate the percentage of each thermal-stress regime covered by the two types of MPAs: no-take MPAs are indicated by green bars; multiple-use MPAs are indicated by light green bars. The green checkmark symbol indicates that the conservation objective has been fully achieved in both extent and management type; the red “x” indicates that the conservation objective has not been attained. Objectives were formulated for explictiness in the design of MPAs to account for resilience to warming impacts, considering supporting evidence in the literature (see S2 Fig for further details about rationales to protect all regimes).
Fig 5.
Decision tree for using information on chronic and acute stress derived from observed and projected data to formulate quantitative conservation objectives for warming disturbances.
Ch = chronic stress, Ac = acute stress. Percentage values inside boxes in the bottom of the figure are prescribed (but indicative here) coverages by no-take and multiple-use MPAs.
Fig 6.
Annual maximum SST and Degree Heating Weeks (DHWs) for five reef cells within different thermal-stress regimes.
Approximate locations of the five cells are shown as a-e in Fig 2A. Observed data (satellite NOAA) are shown by black solid lines (SST values) and filled bars (DHWs) while projections (GCM PCM1 output) are shown by gray solid lines (SST values) and filled bars (DHWs). Warming trends (in °C per decade) are shown for observed (ɷO) and projected (ɷP) time series. The thermal-stress regime allocated to each reef cell is indicated in the top right of each graph, and defined in Fig 3.
Fig 7.
The acute thermal stress metric.
Accumulated time for which the ecosystem is under reduced ecosystem function from acute stress events for all reef cells according to observed (A) and projected (B) time series. Times are derived from the logistic function used to relate intensity of acute stress events to recovery time (Fig 3) and summed through each time-series and presented as years per decade. Panels for reefs in the northern, central, and southern sectors of the study area correspond to insets in Fig 2A.
Fig 8.
Distribution of thermal-stress regimes across study area.
(A) Reef cells to which the nine regimes were allocated. Empty (black outlined) cells are unclassified because they have middle-tercile values for at least one of the four variables used to classify regimes. Labels for thermal regimes match those in Fig 4. Views for reefs in the northern, central, and southern sectors correspond to insets in Fig 2A. (B) Areal coverage of each thermal-stress regime as a percentage of the total of all reef cells allocated to regimes (n = 101). Each regime was encapsulated within a single sector.
Fig 9.
Spatial prioritization of coral reefs in Brazil based on our conservation objectives for incorporating warming impacts.
Maps show best solution and selection frequency Marxan outputs when selection of reef cells coinciding with existing MPAs was mandatory (A and B, respectively) or optional (C and D, respectively). Views for reefs in the northern, central, and southern sectors of the study area correspond to insets in Fig 2A.