Levels and trends in marine contaminants and their biological effects – CEMP Assessment report 2015

The full assessment is available online at:  http://ices.dk/data/assessment-tools/Pages/ospar-cat-app.aspx and through the interactive map below.

The 2015‐16 MIME roll‐over assessed 7615 time series (of three years or more) in biota, of which 5216 were assessed for trends and 6727 for status, and 3453 time series in sediment, of which 2360 were assessed for trends and 2725 for status. A breakdown of trends and status by region and determinand is given in Tables 1‐4.

For organics, the prevailing trend is generally downwards. However, parent PAHs in sediments in Region II and Region III are tending to increase. For metals, there is a mixture of upwards and downwards trends, with a relatively high proportion of upwards trends for Cd and Hg in biota and As, Cu and Ni in sediments. The increase in sediment parent PAH concentrations is the most noticeable difference since the 2013‐14 assessment (OSPAR Monitoring and Assessment series, Levels and trends in marine contaminants and their biological effects – CEMP Assessment Report 2014, http://www.ospar.org/work‐areas/cross‐cutting‐ issues/cemp.

The assessment methodology is described in the help files that accompany the assessment. There was one major change to the methodology used for contaminants and biological effects (other than imposex) in the 2014‐15 assessment. Previously, a time series was modelled by summarising the concentrations each year by an annual index, and then using a weighted linear regression or loess smoother (depending on the number of years) to describe changes in the annual indices over time. This year, the concentrations were modelled directly using a mixed model, which also estimated the different variance components in the data. The advantage is that measurements below the detection limit could also be incorporated (by treating them as left‐censored data), which allowed some times series to be extended and others to be included in the assessment for the first time. At present, only those time series where the proportion of less‐than measurements is not too high can be modelled in this way, but it is hoped to extend the methodology to include all time series with at least three years of data for the 2016‐17 assessment.

There was also a major change to the methodology used for assessing VDS when individual stage measurements were available, rather than the summary indices. The individual stages were modelled using a proportional odds model and trends were allowed that included a change‐point between 2004 and 2008, the period when the ban on TBT was coming into force. The advantages are that the uncertainty in the trend and status assessments can be more correctly quantified and more flexible forms of trend, which match patterns seen in the data, can be fitted. One problem is that some time series are a mixture of individual measurements and indices, and the new methodology can only be applied to the subset of the time series with individuals. However, it is hoped that these time series can be extended back in time by the resubmission of legacy data including individual measurements.

The number of sediment time series has dropped since the 2014‐15 assessment. However, this is because some monitoring stations that are close together have been grouped for assessment purposes. This should provide time series that are more powerful for detecting trends and assessing status.