**1. Introduction**

Warming sea surface temperatures caused by climate change threaten coral reefs globally [1]. Increased water temperatures cause coral bleaching (reviewed in Reference [2]) which can cause total or partial mortality for colonies if the corals are unable to recover (reviewed in Reference [3]). Coral mortality leads to reef degradation as the reef loses structural complexity and is overgrown by algae, often leading to an algae-dominated phase shift [4]. Reef degradation directly causes the loss of reef-related ecosystem services, such as seafood production, shoreline protection, habitat provision, materials for medicines, and nitrogen fixation, among others [5].

Significant ecological declines driven by anthropogenic stressors are occurring on coral reefs around the world [6]. In 2000, an estimated 11% of all coral reefs had already been lost with an additional 16% damaged beyond the point of being functional ecosystems [7]. From 1985–2012 the Great Barrier Reef experienced a 50.7% decrease in coral cover [6] and the coral cover in the entire Indo-Pacific is 20% less than historical levels from 100 years ago [8]. Hawaiian reefs, however, have one of the lowest threat ratings in the Pacific (less than 30% threatened) [9]. From 1999–2012 mean Hawaiian coral cover and diversity remained stable statewide, including within Kane'ohe Bay [ ¯ 10]. Reefs within Kane'ohe ¯ Bay have repeatedly shown resilience by recovering from natural and anthropogenic disturbances such as bleaching events [11]. Increasingly frequent bleaching events threaten the longevity of coral reef ecosystems [12] and whether or not corals can become adaptive or resistant to bleaching is contested in current literature [12]. However, corals in Kane'ohe Bay have shown resilience through acclimatization ¯ to increased temperatures [13]. In this study resilience is defined as 'the ability of an ecosystem to recuperate its structure and functions after a perturbation' [14].

#### *Kane'ohe Bay, Hawai'i ¯*

Coral reefs in Kane'ohe Bay, located on the northeast side of O'ahu, Hawai'i (21 ¯ ◦4 N and 157◦8 W), have some of the highest levels of coral cover (54–68% compared to statewide average of 24.1%) across the Hawaiian islands [10,11,15]. Due to the unique geographic properties of Kane'ohe Bay, ¯ these reefs experience elevated summer water temperatures (1–2 ◦C), which offshore reefs will not be subjected to for several years [16].

Kane'ohe Bay represents one of the few recorded examples of a phase shift reversal, in which ¯ the reefs were coral-dominant then algal-dominant and have returned to coral-dominated reefs all within a 40-year period [17]. From 1960–1970 the human population in Kane'ohe doubled, leading ¯ to effluent municipal and military sewage to be discharged in the bay, causing eutrophication and a subsequent decline in coral cover and diversity [18]. Following the release of effluent sewage into the bay, the algae *Dictyosphaeria cavernosa*, stimulated by increased nutrient availability, spread widely, causing a phase shift from coral-dominated to algae-dominated [19,20]. Following the 1979 sewage diversion, coral cover in the bay more than doubled in just four years [21] as nutrient levels decreased [19].

The first documented coral bleaching event in Kane'ohe occurred in 1996, in which the total coral ¯ mortality was < 1% [22]. A second, more severe bleaching event occurred in 2014 [16]. While nearly half of all corals in the southern region of the bay were pale or bleached immediately following a 2014 bleaching, there was only 1% total coral mortality three months later [23]. In 2015, another widespread bleaching event affected the Kane'ohe Bay reefs, however a 15% decrease in bleaching compared ¯ to the 2014 event suggested some corals may be acclimatizing to increased temperatures, although higher levels of mortality were observed [11]. Kane'ohe Bay has retained high coral cover despite ¯ Hawaiian offshore water temperatures increasing by 1.15 ◦C over the past 60 years [11]. Corals within the bay also show increased resistance to acidification and warming waters compared to other corals in O'ahu [24]. The historical resilience of corals in Kane'ohe Bay and the consistently high coral cover ¯ while many reefs around the globe are in decline led to the following research question: How has coral cover and community composition changed in response to 18 years of warming temperatures and two major bleaching events in a well-studied coral reef ecosystem?

#### **2. Materials and Methods**

#### *2.1. Study Site: Kane'ohe Bay, Hawai'i ¯*

The study site was a 600-meter section of the Malauka'a fringing reef (21.44300899◦ N, 157.80636◦ W to 21.43853104◦ N, 157.806541◦ W) in the south-west of Kane'ohe Bay, which was initially surveyed ¯ in 2000 [25]. Similar to other reefs in the bay, *Porites compressa* and *Montipora capitata* are the dominant reef-building corals. The northern section of the reef is approximately 125 meters offshore of Kealohi Point at He'eia State park. The southern 200 meters of the study site is adjacent to the Paepae o He'eia (traditional Hawaiian fishpond) where there is ongoing estuarine restoration focusing on sociocultural benefits [26]. The southern end of the reef is subjected to freshwater stream and pond output from He'eia stream and a triple mak¯ ah¯ a (sluice gate) within Paepae o He'eia [ ¯ 27]. The selected reef suffered bleaching and low mortality (<5%) during the 2014/2015 bleaching event [11].

#### *2.2. Comparative Study Setup*

## 2.2.1. Benthic Survey

Coral cover and benthic community composition were measured through a quali-quantitative comparison using a modified version of the point intercept transect (PIT) (as described by Reference [28]) in the initial survey (2000) and follow up survey (2018). The PIT method identifies benthic cover every 50 cm along a transect [29]. During the 2000 study [25], benthic cover was recorded every meter and thus repeated as such in the 2018 study. Coral species, algae species, crustose coralline algae, turf, sand, and rubble were recorded along each transect. Crustose coralline algae and turf were pooled together into 'non-coral substrate' and sand and rubble were pooled together into 'mixed sand' as they were not separated from one another in the 2000 survey. Additionally, transects from the 2000 study continued until the edge of the reef platform was reached, causing transects to consist of varying lengths (5–34 m) dependent on the width of the reef. The locations of transect sites (n = 60) during the 2000 survey were resurveyed in 2018 using a Garmin GPSMAP 78s; 3 m accuracy (Garmin Ltd., Olathe, KS, USA). Transects were spaced 10 meters apart to survey the 600-meter portion of the fringing reef (Figure 1). Both surveys were conducted with one snorkeling observer identifying all species in situ. Two community descriptors, cover and community composition, are used to empirically describe resilience to environmental stressors present at the site [14].

**Figure 1.** Map of Malauka'a fringing reef with transects overlaid within Kane'ohe Bay, O'ahu. Note the ¯ variation in transect length due to reef width. Photo Credit: Digital Globe.

## 2.2.2. Seawater Temperature

Daily mean seawater temperatures (◦C) for 2000 to 2018 in Kane'ohe Bay were calculated from ¯ PacIOOS Moku o Lo'e weather station (http://www.pacioos.hawaii.edu/weather/obs-mokuoloe/).

## 2.2.3. Statistical Analysis

A two-tailed t-test was used to determine changes in daily average temperatures between 2000 and 2018 within RStudio IDE Version 1.1.456 (RStudio, Inc., Boston, MA, USA) [30]. A non-metric multidimensional scaling (NMDS) ordination plot using Bray–Curtis distance was created to visualize the 2000 and 2018 benthic communities within ggplot in RStudio [30]. A matched pair Wilcoxon signed-rank analysis was used to compare changes in individual species and groups (i.e., corals, algae, and mixed sand) between years (2000 vs. 2018) within transects using JMP Pro 13 (*JMP* ®, Version 13, SAS Institute Inc., Cary, NC, USA) [31]. A permutational multivariate ANOVA (PERMANOVA) and a permutational test of multivariate dispersion (PERMDISP) were ran to determine if overall species composition changed between 2000 and 2018 using PERMANOVA+ [32] in PRIMER 7 Version 7.0.13 (PRIMER-e (Quest Research Limited) Auckland, New Zealand) [33]. The data for the PERMANOVA and PERMDISP was square root transformed before calculating a Bray–Curtis similarity matrix. The PERMANOVA was ran with two factors- fixed factor 'year' (2 levels, 999 unique permutations) and random 'transect' (6 levels, with transects pooled into 6 groups of 10 based on location, 998 unique permutations) nested in 'year.'
