In recent decades, acidification of the open ocean has shown a consistent
increase. However, analysis of long-term data in coastal seawater shows
that the pH is highly variable because of coastal processes and
anthropogenic carbon inputs. It is therefore important to understand how
anthropogenic carbon inputs and other natural or anthropogenic factors
influence the temporal trends in pH in coastal seawater. Using water quality
data collected at 289 monitoring sites as part of the Water Pollution
Control Program, we evaluated the long-term trends of the pH
The effect of ocean acidification on several marine organisms, including
calcifiers, is widely acknowledged and is the topic of various marine
research projects worldwide. Chemical variables related to carbonate cycles
are monitored in several ongoing ocean projects to determine whether the
rate of ocean acidification can be identified from changes in pH and other
variables in the open ocean (Gonzalez-Davila et al., 2007; Dore et al., 2009;
Bates, 2007; Bates et al., 2014; Midorikawa et al., 2010; Olafsson et al., 2009;
Wakita et al., 2017). Analysis of pH data measured in situ at the European
station in the Canary Islands (ESTOC) in the North Atlantic from 1995 to
2003 and normalized to 25
These long-term time series from various sites in the open ocean indicate
consistent changes in surface ocean carbon chemistry, which mainly reflect
the uptake of anthropogenic
Duarte et al. (2013) hypothesized that anthropogenic pressures would cause
the pH
These processes that occur only in coastal regions might cause increases or decreases in the rate of acidification, meaning that the outcomes for coastal ecosystems in different regions will vary. At present we have limited information about long-term changes in pH in coastal seawater, mainly because of the difficulty involved in collecting continuous long-term data from coastal seawater around an entire country at a spatial resolution that sufficiently covers the high regional variability in coastal pH.
The Water Pollution Control Law (WPCL) was established in 1970 to deal with
the serious pollution of the Japanese aquatic environment in the 1950s and
1960s. Several environmental variables, including pH
Regardless of any shortcomings, the WPCL coastal monitoring program in Japan includes more than 2000 monitoring sites that cover most parts of the coastline (Fig. 1), so the dataset provides the opportunity to estimate the overall trend in pH in Japanese coastal areas and the regional variability in the trends from data of known precision. Suitable analytical methods could make up for these shortcomings of the WPCL dataset. In this study, we focused on the general characteristics of the overall pH trends at all monitoring sites rather than examining the trend in pH at each site in detail, after carefully considering the accuracy of the dataset.
Coastal maps and monitoring sites in Japan. Red points in
In the present study, we examined the pH
Data for several environmental variables, including pH
The data were collected by the Regional Development Bureau of the Ministry of Land, Infrastructure, Transport, and Tourism and the Ministry of the Environment under the WPCL monitoring program. Monitoring protocols (sampling frequencies, locations, and methods) are outlined in the program guidelines (NIES, 2018, informed by the Ministry of Environment, MOE), originally published in Japanese, and we have provided a summary of these protocols in this paper.
Monitoring is carried out at 1481 sites along the Japanese coast, as shown in Fig. 1a. While most sites are in coastal sea areas, up to 10 % are in estuaries. At each monitoring site, basic surveys were carried out between 4 and 40 times a year, depending on the site. Information on the sampling frequency at the monitoring sites is presented in Table 1. During basic surveys, water samples were collected from 0.5 and 2.0 m below the surface at all sites. At sites where the bottom depths were greater than 10 m, water samples were collected four times a day to cover diurnal variation. At sites where the variation in the daily pH was large, samples were also collected over a period of 1 d at 2 h intervals (ca. 13 times a day) at least twice a year to check the adequacy of the basic water sampling protocol.
Number of samples (
The pH for each water sample was measured in accordance with the Japanese
Industrial Standard protocol JIS Z 8802 (2011), which is equivalent to
ISO10523 (
The monitoring operations were carried out by licensed operators as outlined in the annual plan of the Regional Development Bureau of each prefecture. These specific licensed operators were retained for the duration of the measurement period, which means that the same laboratories were always in charge of collecting the data. This approach helps to prevent systematic errors that might arise both between measurement facilities and over time and ensures the datasets are accurate.
We selected all the data for fixed sites in coastal seawater that had continuous time series from 1978 to 2009. There were 2463 regular and non-regular monitoring sites in 1978 and 2127 sites in 2009. While there were very few sites in some prefectures in Hokkaido and Tohoku, the monitoring sites covered almost all of the coastline of Japan (Fig. 1).
As explained in more detail later in this section, we applied a three-step quality control procedure. We excluded (1) discontinuous time sequences, (2) time sequences that had extreme outliers in each year, and (3) time sequences that included significant random errors and which were only weakly correlated with time sequences at adjacent sites.
When we excluded the sites that had discontinuous pH
Distributions of the monthly number of data points (
For the 302 sites, we evaluated whether the water temperature (Fig. 4a–b)
and pH
Examples of
Average mutual correlation coefficients among water temperature and pH
Correlations of water temperature and pH
Scatter plots of correlation coefficients for water temperature and
pH
The monitoring in each prefecture is carried out by different licensed
operators, decided by the Regional Development Bureau in each prefecture.
Intercalibration measurements have not been conducted between different
licensed operators. Even though all the operators follow the same JIS
protocol, manual monitoring can introduce systematic errors into the data.
Some adjacent monitoring sites are close to each other but are managed by
different operators, such as sites close to the boundaries between Osaka and
Hyogo (Fig. 6a), Hyogo and Okayama (Fig. 6b), Kagawa and Okayama (not
shown), and Kagawa and Ehime (not shown). The pH
Examples of time series for annual minimum and maximum pH
The histograms of the calculated pH
Histogram of pH trends, represented by
We detected both positive (basification) and negative (acidification)
trends, which contrasts with the findings of other researchers who reported
only negative trends (ocean acidification) in the open ocean (Bates et al., 2014; Midorikawa et al., 2010; Olafsson et al., 2009; Wakita et al., 2017). The
average (
The negative trends were relatively weak for the minimum pH
We examined the pH
Distributions of long-term trends in pH
By examining the average minimum and maximum pH
We found more acidification trends for the minimum pH
The JIS Z8802 (2011) allows a measurement error of
We used Welch's
We also applied a paired
The WPCL dataset did not discriminate between surface (0.5–2 m) and
subsurface (10 m) data when calculating the annual maximum and minimum
pH
Usually the pH is lower in subsurface water than in surface water, as primary production decreases and increases the DIC concentrations in surface and subsurface water, respectively, because of decomposition when particulate organic carbon (POC) is produced by primary producers. We therefore speculate that the annual maximum pH includes very little data from a depth of 10 m, thus this value does represent the winter pH of surface waters. In contrast, the annual minimum pH was somewhat difficult to interpret, as it may have contained data from 10 m at some monitoring sites but only surface data at other sites shallower than 10 m.
Results of statistical analysis (Sect. 4.1) confirm that the trends in
minimum and maximum pH
To facilitate our discussion of the factors that influenced the
pH
The pH
The DIC process (
It is difficult to observe general trends in both DIC (
Same as Fig. 7 but showing the pH
Our analysis was based on pH
To evaluate the direct thermal effects related to process
The distributions of the trends in pH
We used Welch's
The pH
The average highest temperatures observed at the minimum pH
Average highest and lowest temperatures observed for the minimum and
maximum pH
We estimated thermal effects and that the pH
We found regional differences in the pH
We used CO2sys (Lewis and Wallace, 1998) to predict how pH
The summer pH
Previous studies have reported that nutrient loadings in Japan have
decreased over recent decades (e.g., Yamamoto-Kawai et al., 2015; Kamohara et
al., 2018; Nakai et al., 2018), with variable effects on summer pH
Same as Fig. 7 but showing the highest and lowest temperature trends at 289 sites (selected by quality control step 3).
Correlation between trends in total nitrogen (TN) and trends in
For other stations, however, acidification and basification processes seem to
occur independently to the changes in TN input. The pH can change even with
a constant primary production rate if a residence time of coastal water
changes (for the case of autotrophic water, a shorter residence time could
cause lower pH). Some parts of stations with significant basification and
small
Nakai et al. (2018) reported that nutrient loadings decreased in the most parts of the Seto Inland Sea from 1981 to 2010 but that several areas remained eutrophic. Because of geographical variations in nutrient loadings and the uneven distribution of autotrophic and heterotrophic stations, there are significant spatial variations in pH trends in the Seto Inland Sea (Fig. 8). The pH trends in coastal areas of western Kyushu, where the anthropogenic nutrient loadings are relatively low, therefore reflect the decreases in nutrient discharges, resulting in variations between regions (e.g., Nakai et al., 2018; Yamamoto and Hanazato, 2015; Tsuchiya et al., 2018). Several cities in this area have introduced advanced sewage treatment to prevent eutrophication in coastal seawater (Nakai et al., 2018; Yamamoto and Hanazato, 2015).
Regional variations in coastal alkalinity, along with salinity, might be
related to changes in land use and might affect these trends (process Alk(
Regional differences in pH
We estimated the long-term trends in pH
There were striking spatial variations in the pH
Data are available in an institutional repository that does not issue datasets with DOIs (non-mandated deposition). The pH in situ data and information from monitoring sites are included in the Supplement. All data included in this study are available upon request via contact with the corresponding author.
The supplement related to this article is available online at:
Conceptulization, TT; Methodology, MI, YM, and TO; Validation, MI; Formal analysis, MI; Writing-Original Draft Preparation, MI, YM, TO; Writing-Review and Editing, MI, YM, and TO; Visualization, MI; Supervision, YM and TO; Project administration, TT and YM; Funding Acquisition, TT and YM.
The authors declare that they have no conflict of interest.
We thank the scientists, captain, officers, and personnel of the National Institute for Environmental Studies, Regional Development Bureau of the Ministry of Land, Infrastructure, Transport, and Tourism, who contributed to this study. We acknowledge financial support from the Sasakawa Peace Foundation of the Ocean Policy Research Institute. We also appreciate discussions with members of the Environmental Variability Prediction and Application Research Group of the Japanese Agency for Marine-Earth Science and Technology. Suggestions by two reviewers helped us to improve an earlier version of the paper.
This research has been supported by the Sasakawa Peace Foundation of the Ocean Policy Research Institute (OPRI-SPF).
This paper was edited by Jean-Pierre Gattuso and reviewed by Abed El Rahman Hassoun and one anonymous referee.