High-resolution stable isotope variations in a coral Porites lutea from the Wenno Island, Micronesia: Implications of low-latitude ocean environments

3.1 Coral sampling and subsampling

The skeletal core of modern living coral (Porites lutea) was collected near Wenno Island in tropical Micronesia, at 7°27'13.4”N, 151°54'6.5”E, on 30 June 2007 (Fig. 1). The coral sample was taken from about 3 m water depth using KIOST (Korea Institute of Ocean Science and Technology)’ self-produced drill instrument. After collecting samples, 3-mm-thick slabs were cut at the laboratory and soft X-radiographs were obtained (Fig. 2). The initial radiographs were used to identify annual density bands and to determine subsample for stable isotope analysis.

Fig. 2.

X-radiographic image of a coral skeleton (Porites lutea). Well-developed annual growth bands reveal alternating high-density (dark color, corresponds to summer) and low-density (light and winter time) intervals. One year is represented by a low- and high-density band pair. Thin white lines depict sampling transects for isotope analysis. 350 samples were taken coral top to 426.5 mm interval only.

Based on visual observation of the X-ray images, the rate of coral growth seems not constant and growth axes are variable. Therefore, a powder subsample was taken along the three sampling axis lines I, II, and III (Fig. 2) and then the respective overlap of ~1 year between two lines yield a continuous coral isotope record. A total of 350 samples were collected from the coral top to the coral length of 426.5 mm with intervals of 0.8-2.4 mm. The sampling resolution corresponds to 13-20 samples/year.

3.2 Oxygen and carbon isotope analysis and in situ measurement of SST and SSS

Stable oxygen (δ¹⁸O_c) and carbon isotope (δ¹³C_c) measurement of the coral skeleton were conducted using an automated Finnigan-MAT 251 gas ratio mass spectrometer at Ohio State University (OSU) of USA. Powdered samples were analyzed by Mass Spectrometry using a carbonate preparation system that reacts samples with 100% phosphoric acid at 70℃ for 10 min. Data were calibrated using an in-house calcite standard, and the international standard NBS-19 was run with each batch of samples to check the accuracy and precision of both δ¹³C and δ¹⁸O ratios. The analytical precision was reported as 0.02‰ VPDB for δ¹³C and 0.03‰ VPDB for δ¹⁸O, respectively.

For the comparison, we selected the previously published open data from the National Oceanic and Atmospheric Administration (NOAA) describing the neighboring area of Aulong (7°16’23’’N, 134°18’07’’E) and Drum Can Islands (7°17’44’’N, 134°24’24’’E) (Morimoto et al., 2002), and the temperate climate region Kikai Island (28°16′ to 28°22′ N and 129°55' to 130°02'E) (Moromoto et al., 2004). Also NOAA Optimum Interpolation (OI) SST data from nearby coral sampling point were collected from the NOAA physical science laboratory to comparison between δ¹⁸O_c and SST variability for 22 years. Detailed explanation about the isotopic similarities will be discussed in following Section 4.3.

The Korea Institute of Ocean Science and Technology (KIOST) has been collecting in situ SST and SSS data using conductivity, temperature, and depth (CTD) equipment (Seabird-25) for regular observation and comparison with other physical data. The CTD equipment had been deployed about 21 m meter water depth every week as a part of the KIOST’s program of monitoring ocean acidification. The collected in situ SST and SSS data were used to compare with oxygen and carbon isotope data of the studied coral. Instrumental data are available since September 2006.

3.3 Determination of coral chronology

Usually corals exhibit annual bands within their skeleton, which representing alternation of low-density (light color) and high-density (dark). Even though band width (growth rate) may be differed by various environmental factors such as temperature, one year is represented by a low- and high-density pair as shown Fig. 2. In this study, coral chronology of studied coral was determined by setting the maximum δ¹⁸O_c value in a given year equal to mid-February (consistent with the minimum SST), and minimum value equal to mid-September (consistent with the maximum SST). Therefore, we linearly interpolated between two fixed points for age determination. The resulting values were interpolated linearly to 15 to 20 values per year, providing a monthly resolution.

The strong δ¹⁸O_c cyclic variation was observed, and it was compared with annual band to determine the chronology of the coral, supporting consistent density banding pattern of low-density bend during winter and high-density band during summer. Regarding the high resolution sampling and seasonal growth rate, we assigned 22 year records from coral top to 426.5 mm depth (Fig. 2). To improve the data quality and enhance analytical accuracy, 37 samples were duplicated or triplicated and the averaged values were finally used.

4.1 Isotope excursion, amplitude, and time-series variation

Oxygen (δ¹⁸O_c) and carbon (δ¹³C_c) isotope ratios of coral skeleton show cyclic seasonal variations, covering a growth history of 22 cycles from the bottom at the core depth of 42.65 cm to the top (Fig. 3). Also, the δ¹⁸O_c data is characterized by a large-scale variation with an amplitude of about 1‰. The highest ratios of δ¹⁸O_c (-5.22‰) occur at 178.98 mm, 271.57 mm, and 285.04 mm depths from the coral top; corresponding to February 1999, 1994, and 1993, respectively (Fig. 3). Also, the lowest ratio (-6.13‰) of δ¹⁸O_c occurs at 372.78 mm depth and the next lowest occur at 51.4 mm and 58.814 mm depths from the top, corresponding to early September 1987 and 2004, respectively. The general difference of δ¹⁸O_c between the highest and lowest values within a cycle is approximately 0.9‰, corresponding to about 3.5℃ SST difference when the δ¹⁸O_c was converted into a temperature scale based on previous research (Epstein et al., 1953; Jouzel et al., 1994). On the other hand, instrumental SST change is approximately 1.5℃ through a year, this means that the variation of the δ¹⁸O_c is influenced by both SST and seawater δ¹⁸O relating to SSS (Gagan et al., 2000). In fact, in situ monitored SSS clearly shows seasonal variations with an amplitude of about 0.8 psu (Fig. 4); this corresponds to 0.27‰ of seawater δ¹⁸O when using a relationship of Fairbanks et al. (1997).

Fig. 3.

Stable oxygen and carbon isotope variation during the period from June 1985 to July 2007 with original coral depth (cm) in upper panel. Twenty-two cyclic variations are shown with high amplitudes. The whole depth was represented corresponding calendar year (lower panel).

Fig. 4.

Variations of instrumental SST and SSS data from Sep 2006 to Sep 2008 (a) and relationship between stable oxygen isotope and SST (b) and SSS (c). Since the coral samples have annual band and growth rate, each subsamples taken within a year were calculated corresponding depth. Black and open circles are instrumental data of seawater temperature and salinity by CTD, respectively.

The δ¹³C_c ratio revealed high amplitude-cyclic variation that was similar to δ¹⁸O_c excursion (Fig. 3). It seems that there was approximately 1‰ difference, and depleted δ¹³C_c ratios were observed in 1994, 2005 and 2001 when δ¹⁸O_c had reached the maximum values. However, minima and maxima of δ¹³C_c ratios preceded those of δ¹⁸O_c ratios by about a month. This time lag between δ¹⁸O_c and δ¹³C_c ratio was also observed in Guam coral sample (Asami et al., 2004). Since the δ¹³C_c ratio is related with metabolic effect and kinetic effects that are directly linked with dissolved inorganic carbon (DIC) thus, the rate of coral growth and calcification in ambient water (Swart et al., 1996), the time lag between δ¹⁸O_c and δ¹³C_c might be arisen by the difference in utility of solar radiation, and nutrient availability.

As shown in Fig. 3, the δ¹⁸O_c ratio increased when the δ¹³C_c ratio was lower. The highest δ¹⁸O_c in a year occurs approximately every February and the lowest δ¹³C_c does not necessarily occur in a particular period. It indicates that the δ¹³C_c variation is related to several factors such as environmental and physiological effects: nutrients, solar radiation, δ¹³C of dissolved inorganic carbon (e.g., Swart et al., 1996). Therefore, the tendency differences in both isotopic excursions may be derived from the independent controlling factors, and these factors might explain the time lag between δ¹⁸O_c and δ¹³C_c.

As discussed, δ¹⁸O_c is usually controlled by temperature and δ¹⁸O of the ambient water. The 2-year monitoring showed that the annual lowest and highest SST reached about 28.2℃ in February and 30.2℃ in September, respectively, with a small seasonal amplitude of about 2℃. The monitored SSS data varied between 34.3 and 33.5 psu, with lower salinity occurring commonly in September-October. Weekly records of these instrumental SST and SSS were compared with δ¹⁸O_c as illustrated in Fig. 4. Based on the comparison between instrumental SST and SSS data and the δ¹⁸O_c ratio in the last two years, a strong negative correlation (R² = 0.77, n = 17) in SST, a slightly lower positive correlation in SSS (R² = 0.35, n = 17) were observed. This suggests that the coral δ¹⁸O_c composition is closely associated with ambient water SST and SSS (Figs. 4b and 4c).

The relatively strong negative relationship between SST and δ¹⁸O_c and weak positive relationship between SSS and δ¹⁸O_c were observed in this study (Figs. 4b and 4c). This numerical relationship among SST, SSS and δ¹⁸O_c of studied coral supports the general relationship among SST, SSS and oxygen isotope ratios in ambient sea water environments. Therefore, δ¹⁸O_c of analyzed coral data directly implies a seasonal variation of SST of the study area. Also the differences of heavier and lighter δ¹⁸O_c corresponding to lower and higher SST with slightly lower SST differences. A weak relationship between SSS and δ¹⁸O_c may have resulted from the offset SST variations.

The relationship between coral δ¹⁸O_c and measured in situ SSTs revels a SST to coral isotope values as follow; SST (℃) = (-2.91×δ¹⁸O_c) + 12.258 (R² = 0.77) (Fig. 4b). This formula does not show big differences to those determined by Gagan et al. (1994) for a Great Reef Porites, and by Boiseau et al. (1998) for a Moores Porites. Using this equation, we reconstructed SST (SST_con.) for the 22 years, and compared with NOAA SST again (Figs. 5a and 5b). The NOAA SST and SST_con. exhibits strong resemblance, suggesting the applicability of coral thermometry in the study area. Relatively good correlation between in situ temperature from study area and NOAA SST for a year may supports proxy of δ¹⁸O_c (Fig. 5c).

Fig. 5.

Comparison of δ18Ocwith converted temperature SSTano(av.) (a), and with NOAA SST (b). A year comparison between in situ SST observation and NOAA SST is shown in (c).

4.2 Linear skeleton extension (LSE) and its implications

The annual growth rate of linear skeleton extension (LSE) was determined based on the combination of soft X-ray photographs and annual signals from seasonal variations in δ¹⁸O_c (Fig. 3). The growth rate is quite variable, ranging from approximately 14 to 26.3 mm/year (av. 19.36 mm/ year). There were several reports on the coral growth rate and on the relationship between the instrumental SST and oxygen isotope of coral (Baker and Weber, 1975; Scoffin et al., 1992; Lough and Barnes, 2000; Asami et al., 2004). In their studies, the growth rate was closely related to coral calcification rate. This means that the variability of LSE in our study is associated with calcification rate of coral, further it was strongly related to local oceanographic environmental changes.

The highest LSE was observed in 2002, and the lowest LSE was observed in the 1992 and 1995 periods. The LSE observed at this site is generally higher than those in the Caribbean (6-12 mm/year), and very similar or slightly lower than those of the eastern Pacific (21-39 mm/year), even though the coral species is different (https://editors.eol.org/eoearth/wiki/coralgrowth). In the case of coral Porites lutea, the LSE is quite variable, differing from site to site. Extensive examination on LSE from 48 sites near the coast of Thailand clearly revealed a wide range of LSE variability from approximately 9.5 to 31 mm/year (Scoffin et al., 1992). They interpreted that the LSE is strongly associated with the calcification rate of the coral, the product of linear extension and skeletal bulk density. The hydraulic energy of the individual coral reef sites is the primary controller of the variation in LSE, which is inversely related with skeletal bulk density (Scoffin et al., 1992).

Usually, coral growth depends on sea temperature, irradiance, calcium carbonate saturation, turbidity, salinity, pH, nutrients (Shinn, 1966; Schenider and Smith, 1982; Hubbard and Scaturo, 1985; Shimamura et al., 2008), and climate changes (Al-Rousan et al., 2002). The levels of these components directly influence the physiological processes of photosynthesis and calcification of coral. In particular, coral growth is directly linked to the precipitation of the calcium carbonate skeleton in the form of aragonite. The rate of calcification is consequently controlled by the saturation state of carbonate ions in seawater. Thus, the concentration of carbon dioxide, which is dissolved in seawater as an acidic solution. However, the most important factors controlling the coral calcification are the surrounding SST and light availability, which eventually controls the environmental condition for coral growth (Lough and Barnes, 2000). Since the LSE is the most representative variable of coral calcification (Lough and Barnes, 2000), it is commonly compared with δ¹⁸O in regards to the local and global climatic events such as El Niño (e.g., Maier et al., 2004).

In the present study, the annual δ¹⁸O_c minima, i.e., the annual maximum temperatures, are not linearly in agreement with annual LSE as shown in Fig. 3. This means that annual LSEs may correspond to different environmental factors such as SSS and nutrients level of ambient seawater. In the whole scale, the LSE variation may reflect the decadal-scale in situ oceanographic variation. In this study, interval of small LSEs (i.e., less than 15 mm/year) correspond to the years of 1988, 1992, 1994 and 1996. Among these years, 1997-1998 arerecognized as El Niño years. Also, larger LES also appeared in higher water temperature as indicated in 2001 and 2002 yrs. Therefore, LSE variability might be related to the global scale ENSO signatures which affects the regional ocean environment in the study area.

4.3 Comparison with other coral sites and potential ENSO signal

As a coupled ocean-atmospheric interaction, ENSO is one of the most investigated climatic phenomena in low-latitude area. Because ENSO has played a crucial role in modulating global climate variation across the glacial-interglacial scale (Tudhope et al., 2001) and decadal timescale (e.g., Gu and Philander, 1997; Linsely et al., 2000), and also has widespread effects on the global climatic system and the ecology of the tropical ocean, its variability and uncertainty have been widely recognized based on past natural variations and its response, global warming (e.g., Correge et al., 2000; Latif and Keenlyside, 2009). For example, 10 years of SST observations by satellite clearly showed that the annual mean SST and size of the WPWP increased from 1983 to 1987. Yan et al. (1992) has interpreted this increase as a result of solar irradiance variability, ENSO events, and global warming. In a previous similar study, ENSO events were mostly and clearly recorded as SST signals in corals growing (Wellington and Dunbar, 1995).

Our isotopic data were compared with previous works of Morimoto et al. (2002, 2004) (Fig. 6). This Fig. 6 showed a coincident cyclic variation during 1991~1999, even though Kikai Island is located in the further northern part of the Pacific Ocean (more than 20°N). The isotopic excursion of the studied coral was strongly matched with those of Kikai even though the amplitude of δ¹⁸O_c our study is rather smaller than those of Kikai. Given that both δ¹⁸O_c is associated with latitude-dependence oceanic carbon chemistry, and ENSO phenomena influence on global climatic changes, the strong resemblance in both δ¹⁸O_c variation could be linked to global climatic and ocean environmental changes. The relatively small variation in δ¹⁸O_c duringthe years 1995 and 1996 in our study could be caused by a difference in the degree of SST seasonal variations to inter-annual variations.

Fig. 6.

Comparison of oxygen isotope ratios of corals between this study and previous works in Kikai Island (Morimoto et al., 2002).

As illustrated in Figs. 6 and 7, those excursions of δ¹⁸O_c during 1991 to 1999 in Kikai Island, and 1998 to 200 in Aulong and Drum Can Islands show fairly good coincident variation with some inconsistency in fluctuation. Additionally, the δ¹⁸O_c variation in our study is roughly consistent with those in a Guam coral in the western tropical Pacific Ocean (Asami et al., 2004). There are δ¹⁸O_c amplitude differences among the Aulong, Drum Can Island and our study during 1988 to 2001. Considering the δ¹⁸O_c is dependent on various factors such as water temperature, carbonate ion and the degree of calcification, these δ¹⁸O_c differences may attribute to local oceanographic variation. Even though the δ¹⁸O_c excursion is slightly smooth in Aulong and Drum Can Island data (Morimoto et al., 2002), the initiating point toward lighter δ¹⁸O_c ratio shifting is quite synchronized. In Kikai Island, δ¹⁸O_c ratio is much heavier than other reference sites implying much colder temperature. Therefore, it is expected that ENSO may influence on the area of Kikai Island’s oceanographic environments across the Aulong and Drum Can Island even though direct comparison by SST anomaly were not conducted.

Fig. 7.

Comparison of oxygen isotope ratios of corals between this study and previous works in Aulong and Drum Can Islands (Morimoto et al., 2004).

To compare the ENSO effect, our SST_con. data were normalized with average value for 22 years (28.86℃), and then expressed the difference value with respect to the averaged value as the SST_ano(av.). This method is similar with NOAA’s SST anomaly, but it differs from in terms of normalized interval. The comparison with El Niño-3/-4 in the east-central Pacific is adjusted for 22 years as shown in Fig. 8 (Kaplan et al., 1998). The comparison between the SST_ano(av.) of this study and the SST_Anomaly by NOAA is characterized by distinctive negative anomaly intervals occurred in several specific years such as 1995~1996, 1999~2000 yrs. On the whole, the entire phase variation between our data and El Niño-3/-4 were roughly analogous in 1987, 1997, 2002 and 2006 with positive anomaly. However, out of phase in two data sets occurred in 1989, 1991, 1992 and 2007.

Fig. 8.

Comparison between the SST anomalies in El Niño-3 (blue) and -4 (red) regions (5°S-5°N, 170°W-120°W) and converted SST anomaly (SSTano(av.)) based on oxygen isotope signals. SSTano(av.) were averaged, and expressed by deviated value from average value as anomaly. Ocean Nino Index (ONI) was cited from http://bmcnoldy.rsmas.Miami.edu/tropics/oni/. ONI implies a standard metric to define the phase of the El Niño Southern Oscillation (ENSO). Distinctive negative anomalies were observed during 1995-1996, 1999-2000 years. ONI index was cited from http://bmcnoldy.rsmas.miani.edu/tropics/oni/.

In particular, negative anomalies occurred at 1995~1996, 1999~2000 in NOAA’ SST anomaly were roughly matched with our SST_ano(av.) (Fig. 8). In between these two periods, there is a big positive anomaly during El Niño year of 1997~1998. Also, there are positive anomalies are shown during 1997, 2002 and 2006. Therefore, SST_ano(av.) calculated by δ¹⁸O_c in our study resembles to NOAA’s SST anomaly implying that oceanographic condition of the study area may have been associated with global scale ENSO phenomenon. In detail, positive warm phase is shown during 1991-1994 by NOAA’s SST_anomaly, however our SST_ano(av.) did not show coincident variation, indicating relatively colder conditions in our study area. Negative cold phase by NOAA’ s SST shown during a 1995-1996 ENSO cooling phase, but relatively smaller negative SST_ano(av.) indicate a slightly warmer SST in our SST_ano(av.). However, the signature of a strong ENSO cooling phase in 1999-2000 was well matched with our SST_ano(av.).

Our Micronesian data are clearly distinguished negative SST anomalies from the averaged value between the 1995 and 1996, and 1999 and 2000. Also distinctive positive anomalies were observed during 1987, 2002 and 2006. These positive and negative anomalies are coincident with NOAA’s SST anomalies. Also, there are remarkable positive and negative anomalies of SST are commonly recognized during ENSO warm phase in the eastern and western Pacific Ocean (Asami et al., 2004). Therefore, the coincident δ¹⁸O_c variation between our data and NOAA’s SST may imply that the ENSO effect influences low-latitude oceanographic environments. In addition, our coral-isotopic data suggesting that the study area may reflect the global phenomenon superimposed onto the local oceanographic condition.