Demographic processes that ensure the recovery and resilience of marine populations are critical as climate change sends an increasing proportion on a trajectory of decline. Yet for some populations, recovery potential remains high. We conducted annual monitoring over 9-years (2012–2020) to assess the recovery of coral populations belonging to genus Pocillopora. These populations experienced a catastrophic collapse following a severe typhoon in 2009. From the start of the monitoring period, high initial recruitment led to the establishment of a juvenile population that rapidly transitioned to sexually mature adults, which dominated the population within six years after the disturbance. As a result, coral cover increased from 1.1% to 20.2% during this time. To identify key demographic drivers of recovery and population growth rates (λ), we applied kernel resampled Integral Projection Models (IPMs), constructing eight successive models to examine annual change. IPMs were able to capture reproductive traits as key demographic drivers over the initial 3 years, whilst individual growth was a continuous key demographic driver throughout the entire monitoring period. IPMs further detected a pulse of reproductive output subsequent to two further Category 5 typhoon events during the monitoring period, exemplifying key mechanisms of resilience for coral populations impacted by disturbance. Despite rapid recovery, (i.e., increased coral cover, individual colony growth, low mortality), IPMs estimated predominantly negative values of λ, indicating a declining population. Indeed, whilst λ translates to a change in the number of individuals, the recovery of coral populations can also be driven by an increase in the size of coral surviving colonies. Our results illustrate that accumulating long-term data of historical dynamics and applying IPMs to extract demographic drivers are crucial for future predictions that are based on comprehensive and robust understandings of ecological change. 

Data collection

Orchid Island (22°03′N, 121°32′E) is a 45 km² volcanic, tropical island 64 km off the coast of Taiwan, encircled by a narrow fringing reef (5–10 m depth), leading to a dramatic drop-off. Such reef topography is sensitive to typhoons that are both frequent and intense in the region (Ribas-Deulofeu et al., 2021). In 2009, the island was severely affected by Typhoon Morakot (Hall et al., 2013), the deadliest typhoon to hit Taiwan in recorded history, which caused a ~66% decline in mean live coral cover (~60% to ~20%) along reefs in southern Taiwan (Kuo et al., 2011).
          Three years after this major disturbance in 2012, three parallel 20 m transects were established at ~8 m depth spaced ~2.5 m apart at a site to the southwest of the island (named Green Grassland; 22°00'N 121°34'E). Usually, this reef site is relatively sheltered from both the waves generated by the winter north-easterly monsoon and summer south-westerly winds. However, on this occasion was proven susceptible to the typhoon in 2009, impacted by a west to south-westerly swell. To the side of each transect, 50 cm × 50 cm permanent quadrats (n = 11) were haphazardly positioned to assess demographic changes over time (Appendix S1: Figure S1 and Figure S2). A total of 33 quadrats were made permanent by placing markers (iron pegs) at each corner with tags attached indicating the designated identification number, making the quadrat easier to detect for future monitoring. With this method, only four quadrats were lost in subsequent surveys, whereas the position of the other quadrats (n = 29) remained. However, the precise location may slightly vary due to the ever-changing dynamics of the reef. For this reason, when quadrats were placed at each marked position, a wide-scale community picture was taken ca. 2 m above the substrate, in order to correct any error in positioning during the analysis, if necessary (Appendix S1: Figure S1a). The quadrat was then photographed at a higher-resolution ca. 1 m above the substrate in order to capture the overall benthic composition (Appendix S1: Figure S1b). The quadrat was then divided into four sections, which were individually captured in order to attain high-resolution images of individual colonies (Appendix S1: Figure S1c). In each of these sections, smaller sized individuals (ca. < 5 cm in length) were photographed with close-up images and scale. This protocol was repeated annually between 2012 and 2020. From photographs of the 29 permanent quadrats, every colony (n = 336) observed was first measured for its two-dimensional (2D) colony size (projected area) using Image J software (Schneider, Rasband and Eliceiri, 2012). The 2D colony size was used for the analysis of annual change in coral cover. Coral cover (%) was determined by measuring the sum of 2D projected areas of coral colonies relative to the area of all quadrats combined. All quadrats were originally occupied by Pocillopora, but from 2016 to the end of the monitoring period, 1 quadrat was empty of Pocillopora colonies with the cover calculation still taking this area into consideration. In the rare case of a slight overlap of colonies (n = 4), the 2D projected area could be easily deduced for the unseen part of colonies.

Measurement of demographic vital rates

In this study, we focused on locally dominant genus Pocillopora spp. The relative contribution of species to the Pocillopora complex was genetically examined by randomly sampling colonies at the site and barcoding mtORF region after extraction of genomic DNA (Johnston, Forsman and Toonen, 2018). Out of 31 sampled Pocillopora colonies, 17 were P. verrucosa and 14 were P. meandrina (Appendix S1: Table S1 and S2). Besides the two dominant Pocillopora species, there were at least two other Pocillopora species present; Pocillopora eydouxi and Pocillopora sp. These species are broadcast-spawners, with the exception of Pocillopora sp., which is a brooder (Mulla et al. 2021). Due to the difficulty in identifying species morphologically in the field, especially at the early life stages, we treated species as a Pocillopora complex (Pocillopora populations).

            As corals are 3D structures, colony size (surface area) was used for Pocillopora colonies in the IPMs, which allowed us to build higher-resolution models. 3D surface area (cm²) was allometric and estimated from 2D projections using a pre-established relationship. Detailed information on the 2D to 3D conversion can be found in Appendix Figure S3. We extracted information on colony growth, survival and recruitment of Pocillopora populations over the 9-year period using size-thresholds in 3D to distinguish visible recruits (0.4–10 cm²; n = 154), juveniles (10.1–100 cm²; n = 369) and adults (> 100.1 cm²; n = 532). These threshold for visible recruits was determined from the size range of newly appearing individuals from each year from the second year of monitoring. The threshold for juveniles was determined by the maximum size of visible recruits and the minimum size of sexually mature individuals (described in more detail below). These thresholds differentiate sexually immature (visible recruits/juveniles) to mature (adults) individuals, used for ecological interpretation.
            To identify size-specific relationships of demographic traits associated with reproduction, two nubbins (~5 cm in branch length) were collected from 40 colonies of varying size of P. verrucosa (probably including P. meandrina: 68.2–685.8 cm² in 3D size) during the reproductive season (April, 2017) at neighbouring Green Island (Lin and Nozawa, 2017). In addition, a further 20 nubbins (of the same size) were collected (68.2–364.7 cm² in 3D size) to determine the minimum size of sexual maturity at the same time. Nubbins were fixed in a 10% formalin-seawater solution and examined using standard histological methods. Tissue of nubbins were decalcified and dehydrated in an alcohol series using a tissue processor (Thermo Scientific, Excelsior ES, USA) and embedded in paraffin wax (Thermo Scientific Histoplast PE, USA). Samples were then cut with a microtome (Thermo Scientific, Finesse 325) at 6 µm thick intervals. Xylene was used to deparaffin samples and tissue sections were mounted on glass slides, stained with hematoxylin and eosin using a staining machine (Shandon Varistain, Thermo Scientific, USA) and then preserved with Organol/Limonene mounting medium and a glass cover. Sections were examined under a BX51 light microscope (Olympus, Japan). For each nubbin, 2 polyps were haphazardly chosen and the number of oocytes per polyp was determined by observing the entire section of each polyp (a total of 4 polyps per colony). The probability of a colony being reproductively active was determined by the presence or absence of oocytes over colony size.