Study area
This study was conducted between October 2019 and March 2020 in Shimoni, Kenya, focusing on the patchy coral reefs on the coastal side of Shimoni (Fig. 1).
The area is characterized by a semidiurnal tidal regime, with a tidal range between 0.1 m (LAT - Lowest Astronomical Tide: Elevation of the lowest tide expected to occur at site under average meteorological conditions (NOAA 2020)) to 4.1 m (HAT - Highest Astronomical Tide: Elevation of the highest tide expected to occur at a site under average meteorological conditions (NOAA 2020)) (KPA, 2020). The climate is influenced by two alternating seasons: the southern and northern monsoons. From November to March the NE monsoon is the prevailing wind and is associated with the southern hemisphere summer, or dry season. From April to May comes the rainy season. From June to October there is a complete reversal in wind direction and the SE monsoon wind prevails, associated with the southern hemisphere winter (Richmond 2002). According to the National Oceanic and Atmospheric Administration (NOAA), during the course of this study the daily sea surface temperature varied on average between 26.9–30°C in the study region (SI Appendix, Fig. S1) (NOAA 2022).
An intertidal (4°38'58.7"S 39°23'09.8"E) and a subtidal (4°39'01.5"S 39°23'13.0"E) zone were studied. HOBO loggers were placed between February and March 2020 in both zones to register temperature (SI Appendix, Fig. S2). The intertidal zone registered a high daily temperature variability with a daily average of 30.3°C, minimum average temperatures during dark hours registering 28°C and maximum values during light hours reaching 32.5°C. The intertidal reef was patchy and characterized by a low number of coral species, where most specimens developed wide, short and thick colonies. Among these coral colonies it was common to find coral rubble, macroalgae and sea urchins. Additionally, small reef fishes (e.g. families Labridae, Pomacentridae and Ostraciidae) were present with a high incidence of juveniles. The most abundant coral species were massive corals, namely Porites spp., Pavona spp., Dipsastraea spp., Favites spp. and Galaxea spp., but branching forms were also present, namely Pocillopora spp. and Stylophora spp. During spring low tides, some corals would experience air exposure for maximum 1 h.
The subtidal zone registered more stable temperature regimes and a daily average temperature of 30°C, with average minimum temperatures during dark hours registering 28°C and maximum average temperatures during light hours reaching 31°C. The subtidal reef had a higher abundance of coral species, with seagrass meadows interspersed between coral patches. Sea urchins were also common in this area and there were more fish species (other than the families mentioned above, e.g. Acanthuridae, Chaetodontidae, Diodontidae, Pomacanthidae, Scarinae, Scorpaenidae, Serranidae, Siganidae, Tetratodontidae and Zanclidae) and a higher proportion of adult individuals. Other than the aforementioned coral species, in the subtidal zone it was also possible to find more branching species such as Acropora spp. and Seriatopora spp. as well as other massive species belonging to the family Faviidae. The reef-building hydrozoan Millepora spp. and the feather hydroid Aglaophenia spp. were also commonly found. See Knoester et al. (2022) for additional details on this subtidal reef.
Experimental design
Experimental units
P. damicornis coral fragments were cultured in small coral nurseries that contained eight fragments each. Nurseries consisted of PVC pipes mounted onto 1.5-L glass bottles that were anchored into a round concrete disk of 40 cm in diameter (Fig. 2) Four plastic PVC pipes were welded in a cross shape: two measuring 22 cm and two measuring 26 cm, connected to each other and to the bottle by T-joints. One of the joints was screwed on the opening of the bottle and tightly secured with tie wraps. Per PVC pipe, two coral fragments were hung in a slipknot made with a fishing line with 12 cm between the two fragments.
The nurseries were filled with coral fragments (mean length ± SD = 4.2 cm ± 0.7, n = 480) collected from either the intertidal or the subtidal zone. Corals from the intertidal zone were collected from depths between − 1.42 to -0.15 m and corals from the subtidal zone between 0.66 to 1.67 m, all depth values with reference to MLLW (Mean Lower Low Water: Average elevation of the lowest tide recorded at a site under average meteorological conditions (NOAA, 2020)). No more than three fragments were collected from the same colony to ensure that the cultures were genetically diverse and to avoid over-harvesting of donor colonies. In total, 480 coral fragments were collected and 60 nurseries were constructed and deployed, half in the subtidal zone and the other half in the intertidal zone, with adjacent structures separated by 1 m. Thus, four different treatments were prepared (n = 15 each), of which two control treatments: intertidal corals cultured in the intertidal zone (intint) and subtidal corals cultured in subtidal zone (subsub) and two transplant treatments: intertidal corals cultured in the subtidal zone (intsub) and subtidal corals cultured in the intertidal zone (subint).
Acute thermal stress response and recovery
Coral fragments of six nurseries for each of the four treatments were exposed to thermal stress to test their resistance. The remaining nine nurseries for each treatment were kept as control without receiving thermal stress. An ex-situ setup was prepared at the Pilli Pipa dive base at Fire Fly Eco Retreat in Shimoni, Kenya. Six light-blue plastic basins were filled with approximately 65 L of natural seawater and each was equipped with a Dophin WP3000 Wave Pump to maintain water movement and a Sunsun 300-Watt submersible heater to control the temperature. The basins were covered by a small mesh mosquito net so they would not be affected by debris falling from nearby trees, direct sunlight or wind. For additional protection against rainfall, an overlaying canvas was placed above the basins. Water temperature was monitored with a DIGIFLEX digital fish aquarium thermometer.
Corals had been in culture in the nurseries for approximately one month before being thermally stressed. The thermal stress procedure began by setting the water temperature in the basins at ambient temperature in the nursery area (27–28°C). The PVC crosses with coral fragments were unscrewed from the in situ bottle nursery and transported to the ex situ basins so that each basin contained one PVC cross with eight fragments. A series of incubations was done sequentially (i.e. four series of six incubations within a month) and per incubation all corals shared the same culture environment, but half were originated from the intertidal zone and the other half from the subtidal zone. After 24 h of acclimatization to the water in the basins, the temperature was increased at a rate of 0.5–1°C per hour until it reached 32°C. The coral fragments were kept at this temperature for five days. Pictures of fragments were taken daily using a Nikon Coolpix W300, approximately at 1600 hr, to track the progress of fragment brightness increment induced by thermal stress.
After the five-day period of exposure to 32°C, the heaters were turned off so the water could cool down for 24 h. The PVC crosses with coral fragments were then transported back to the in situ culture settings so they could start to recover. Pictures of the recovering coral fragments were taken every other day for a period of two weeks. Between consecutive incubations, the basins were cleaned and water was changed.
Control treatments were run to understand whether the conditions of the ex situ setup alone were affecting coral coloration. Two additional PVC crosses were filled with corals (n = 1 for subtidal and intertidal corals), collected separately from the ones used in the experimental design. One cross was filled with corals from the intertidal zone and another with corals from the subtidal zone. Each cross was placed in a seawater filled basin where the water was kept at ambient temperature for five days. The coloration of these coral fragments was studied by following the same procedure as described above.
Measurements and data preparation
Coral fragment brightness
All pictures used to track brightness during the acute thermal stress trials were taken with a coral health chart from the University of Queensland as colour reference (Siebeck et al. 2006). This scale was used to standardize the pictures to account for natural variations in light. ImageJ software version 1.52 (Ferreira and Rasband 2012) was used to quantify brightness using the pixel intensity of each picture (Labno 2014). Pixel intensity is measured in bits per pixel (bpp) and describes brightness from 0 (black) to 255 (white), being averaged across red, green and blue channels. (Ferreira and Rasband 2012). This method is based on the relation between coral colour and chlorophyll content, where a healthy coral with darker coloration is expected to have higher chlorophyll density than a stressed coral with brighter colours (Winters et al. 2009). Hence, fragment brightness was used as indicator for the coral bleaching response to the applied thermal stress.
Coral growth and live tissue cover
Additional pictures of each coral fragment under in situ conditions were taken in October 2019, December 2019 and March 2020 to measure growth rates and the percentage of live tissue cover. Growth rates were calculated for the October – December and December – March periods to account for the environmental differences due to seasonal temperature increases. Maximum length (l), maximum width (w1) and perpendicular width (w2) were determined for all coral fragments in the beginning (t0) and end (t1), of both periods using a ruler and ImageJ. With this data, Ecological Volume (EV) in cm3 – the total volume of coral fragment and water between branches – was calculated for t0 and t1. From these EV values, Specific Growth Rate (SGR) was obtained for each fragment (Knoester et al. 2019):
$$EV ={{\pi }\left(\frac{{w}_{1}+{w}_{2}}{4}\right)}^{2}\times l$$
$$SGR/d={ln}\left(\frac{E{V}_{{t}_{1}}}{E{V}_{{t}_{0}}}\right) ∕ \left({t}_{1}-{t}_{0}\right)$$
Live tissue cover was assessed for the October 2019 – March 2020 period, since no seasonal differences were verified for these measurements. Live tissue cover was assessed as a measure of coral fragment condition and was calculated by dividing the percentage of living tissue visually estimated at the end of the experiment by the percentage of live tissue estimated in the beginning.
Data Analysis
Data on each coral fragment for brightness, growth and live tissue cover were averaged per nursery tree so that each nursery was considered an independent replicate to avoid pseudo-replication. All data was analysed with RStudio software version 2023.9.1.494 (Posit team 2023). In all statistical analyses, the assumptions of normality and homogeneity of variances were confirmed by plotting the residuals against fitted values in QQ plots and scatterplots, respectively.
The coral fragments in 10 structures (5 for both intsub and subsub treatments) did not survive the first thermal stress trials and had to be replaced with new corals. Data from the original structures was not used and data from the replacement corals was used for growth and live tissue cover analysis for December 2019 to March 2020 only.
Coral brightness differences
Differences in coral brightness caused by the different culture settings alone were assessed through the brightness values registered in the different treatments after one month in culture and prior to any temperature increase by performing a One-way ANOVA followed by a Tukey’s post-hoc.
To assess for differences on how coral brightness changed over time, Linear Mixed Models obtained from the lnme package (Pinheiro et al. 2023) were used. Two separate models were prepared, one to analyse the response to acute thermal stress and another for recovery. In both models the significance of the fixed effects was assessed with an ANOVA and no post-hoc analysis was required. Furthermore, in both models data was transformed with the natural logarithm to deal with positive skews. Coral fragment brightness was considered the response variable. For the analysis on response to thermal stress, the fixed effects of the model were considered the interaction between treatment (intint, intsub, subsub and subint) and time (number of days in thermal stress 1–5). The random effects were considered the interaction between nursery and time, as the brightness values associated with each nursery were expected to vary differently with time. To assess for differences in the control treatments, an additional Linear Mixed Model was applied, following the same method as described above, but by using origin (intertidal and subtidal) interacting with time for the fixed effects portion of the model.
For the recovery from acute thermal stress analysis, the differences in response were assessed between the different origins of the corals (intertidal and subtidal) and not according with the different treatments. Due to time constraints, it was not possible to gather a balanced dataset for recovery on all coral treatments and there was an underrepresentation of all treatments for the entire 13-day recovery period. The differences in recovery from acute thermal stress with time were assessed as described above. The considered fixed effects were the interaction between origin and time (number of days in recovery 1, 3, 5, 7, 9, 11, 13). The random effects were the same as for the thermal stress response model.
Coral growth and live tissue cover
Growth rate and live tissue cover were analysed separately for fragments that did and did not receive the heat shock treatment. For fragments not receiving the heat shock (i.e., unstressed), differences in growth rate in the different treatments for the periods October – December 2019 (57 days) and December 2019 – March 2020 (92 days) were assessed by performing a One-way ANOVA followed by a Tukey’s post hoc for each period. The growth rate of fragments that had been thermally stressed were analysed for the period December 2019 – March 2020 (i.e., after the thermal stress) and potential differences were assessed as mentioned above. Growth data in stressed fragments was transformed with the square root since it was negatively skewed.
Differences in live tissue cover were assessed with the same tests for the full period October 2019 – March 2020. Data in both measurements was transformed with arcsin since it was negatively skewed.