Environ Sei Pollut Res 
Ö Springer 
Content courtesy of Springer Nature, terms of use apply. Rights reserved. 
Temporal Aspects 
The general overview of the data set, depicting the median 
concentrations of the indicator contaminants of the 13 surveys 
from 2001 to 2014, is presented in Fig. 3. Additional statistical 
data is presented in the supplements (Tab. S10). For some 
compounds, the data indicates that overall both clear upward 
(e.g. PFF1XA, Metazachlor) and downward (e.g., ATR, SIM) 
temporal trends are detectable. Flowever, for many com 
pounds, high variability is observed, which might indicate 
an influence of special local or seasonal effects making trend 
analysis difficult. The data set is divided into two sampling 
periods: from 2001 to 2008, sampling took place during sum 
mer (May to August, Fig. 5), and from 2009 to 2014, sampling 
occurred in January and February. In addition, the early 
cruises in summer contained some additional coastal stations, 
semi-enclosed bay areas, in the western part of the Baltic Sea. 
Unfortunately, there was no sampling in the winter and sum 
mer of the same year. Thus, there remains some uncertainty 
concerning the interpretation of the observed temporal effects. 
The contaminants can be subdivided into three groups ac 
cording to their temporal and spatial behavior (Fig. 5). One 
group exhibits a uniform behavior without great differences 
between summer and winter and no significant spatial differ 
ences. ATR, SIM, and PFASs belong to this group. They can 
be characterized by the absence of large local acute inputs. 
The temporal behavior of no seasonal variation for ATR, 
SIM, and PFASs has also been observed in other coastal wa 
ters (Carafa et al. 2007; Flu et al. 2010; Zhao et al. 2015). 
A second group (DIU, IRG, and 2,4-D) is characterized by 
high variability in summer, with distinct local hot spots and 
high variability between single sampling campaigns (during 
summer sampling). DIU and IRG show high concentrations in 
the most western part of the Baltic Sea (Figs. 4, 5). 
Remarkably, 2,4-D shows elevated concentrations in an area 
north of Rügen and the Arkona basin (TF030, TF113, and 
TF109). It is difficult to decide on trends for these compounds, 
due to their high and variable concentrations, in the early years 
of the monitoring period. The observable downward “trend” 
can be caused by the seasonal effect. Furthermore, Kot-Wasik 
et al. (2004) observed higher phenoxyactic herbicide concen 
trations during springtime in the Gulf of Gdansk (Baltic Sea) 
as well. 
In a third group, compounds can be summarized without 
any significant high local elevations in the summer but with 
slightly elevated concentrations during summer. Trends are 
often detected for these compounds. In the case of decreasing 
temporal trends, it remains more or less a great uncertainty. In 
the case of upward trends, like TERB and METOLA, they can 
be accepted as reliable (Tab. S10, S11). The seasonal varia 
tion, with elevated concentrations during the spring/summer 
time, for TERB and METOLA has been observed in the Sacca 
di Goro lagoon (Italy) as well (Carafa et al. 2007). 
Long-time trends Based on the above-described sensitivity 
towards possible seasonal effects (summer/winter time), the 
time courses of the compounds were analyzed for selected 
areas (from west to east); TF360, TF010 and MB3, DZ1 and 
TF030, and TF113 and TF109 (pairing of stations was select 
ed based on the similarity of the stations’ data and due to 
statistical representativeness). In Tab. SI 1, the results of the 
trend analysis, as well as the sensitivity for seasonal effects, 
are summarized. The trends were identified by critical visual 
inspection of the graphically displayed data (Fig. 7). The pre 
sented calculated linear regression lines support the visual 
inspection but were not used for quantitative evaluations. 
The class of PFASs shows no seasonal influence (group 1, 
sec. 3.4), therefore, trend analyses is quite reliable although 
the time period is shorter (2005 to 2015) (Fig. 7a, Tab. SI 1). 
The main compounds PFOA and PFOS showed a clear down 
ward trend at all stations during the entire period. In contrast, 
the shorter chain C6- and C7-compounds PFFlxA and PFFlpA 
exhibit slightly increasing trends. Evidently, the voluntary 
abandonment of the C8 technology in 2002, by the main pro 
ducer (3M), shows positive results with decreasing concentra 
tions in the Baltic Sea. The PFAS concentration range over 
time is in a similar range as former conducted studies in the 
Baltic Sea (Theobald et al. 2007; Rostkowski et al. 2009; 
Ahrens et al. 2010; Kirchgeorg et al. 2010; Theobald 2011) 
(Tab. SI5). 
The group of herbicides exhibits more complex behavior. 
The triazines ATR and SIM are not influenced by seasonal 
effects (group 1, sec. 3.4) and show clear downward trends 
from 2001 to 2014 at all stations (26.0-1.0ng/L, 4.3-0.9ng/L, 
respectively) (Fig. 7b). These findings fit well with investiga 
tions of the literature and demonstrate positive effects of the 
ban in the mid 1990 (European Commission 2005c, a; Mai 
et al. 2013). In the 1990, levels of ATR near the coast and SIM 
in the south-western Baltic Sea reached up to 20 ng/L and 30 
ng/L, respectively (Bester and Hrihnerfuss 1993; Graeve and 
Wodarg 1996). Further offshore, they were detected at lower 
levels of 1.8-5.1 ng/L and 2.4-6.1 ng/L, respectively 
(Pempkowiak et al. 2000). The mean long-time trend (2001— 
2014) values (2.9 ng/L, 2.7 ng/L, respectively) are consistent 
with the study of (Pempkowiak et al. 2000). Moreover, the 
mean values of ATR and SIM (2.2 ng/L, 2.0 ng/L, respective 
ly) of 2014 are consistent with the measurements done by 
Orlikowska et al. (2015) in the southern Baltic Sea (1.9 ± 
0.3 ng/L, 2.3 ± 0.4 ng/L, respectively). ATR and SIM are 
listed as priority substances in the European Water 
Framework Directive (environmental quality standards 0.6 
|rg/L, 1.0 |rg/L, respectively) (Union 2013). All detected con 
centrations during the long-time series did not exceed these 
environmental quality standards. PROM does not show a sea 
sonal variability as well (Tab. S8), but at most stations, no 
time trend could be identified (Fig. 7b, Tab. Sll). For 
TERB, a slight seasonal influence cannot be excluded, but