Coastal areas represent highly variable environments. The proximity to land, shallow water depths and the direct influence of river discharges affect oceanography, biology and meteorology of these regions (Sinex, 1994). Drastic changes in water parameters like temperature, oxygen level or pH can occur within hours and minutes, often only detected at local spatial scales (Bates et al., 2018). Marine organisms experiencing these conditions exhibit different strategies to cope with those circumstances. In the context of studies on global change such consideration is crucial since biological communities are not directly affected by climate itself, but rather by shorter-term variabilities in environmental conditions (i.e., weather) (Helmuth et al., 2014). However, scientists are still widely applying large-scale averages (temporal and spatial) in experimental approaches, risking misleading interpretations in an either too positive or too negative direction (Bates et al., 2018). This might be the result of not only misconception, but also limited availability and accessibility of necessary information. Until today, the descriptions of monitoring programs designed for pure data acquisition are commonly published in grey literature and the respective data are not publicly available. The fact that many locations are facing extreme events that are far beyond average conditions projected for the future (e.g., Mills et al., 2013) underlines the urgent need for fine-scale data. Identifying and describing environmental variabilities (e.g., extreme events) would be a first step towards better predictions of climate change impacts.

The Baltic Sea is a particular example for environmental variability of natural and anthropogenic origin. It exhibits gradients of critical environmental drivers caused by its semi-enclosed characteristics and the large drainage area of surrounding landmasses, mainly reflected in decreasing salinity and temperature towards the north (Snoeijs-Leijonmalm et al., 2017). The major contribution to environmental variability caused by human activities in the Baltic Sea is represented by eutrophication. Nutrient concentrations rose until the mid-1980s by riverine inputs of nitrate and phosphate (HELCOM, 2018). The resulting enhanced primary productivity led to an increase in organic matter deposition, which fueled respiration at the seafloor and created hypoxic or even anoxic areas of great spatial extent (HELCOM, 2018). Even though nutrient inputs decreased in the last 2 decades, so-called “dead zones” are persisting, not least because climate warming is boosting deoxygenation (Carstensen et al., 2014). Consequently, shallow areas at the shore can experience episodic hypoxia resulting from oxygen-depleted bottom waters and, in addition, upwelling events that transport water from deeper basins to coastal areas (Conley et al., 2011; Saderne et al., 2013). In the latter case, benthic communities living in these habitats will face not just one but a set of environmental shifts: in addition to low oxygen levels, organisms are subjected to increased nutrient concentrations, lower temperatures, higher salinities and elevated pCO2 levels (Lehmann and Myrberg, 2008; Saderne et al., 2013). However, the impact of upwelling events on benthic communities is not straightforward, as the lower temperatures and higher salinities represent short-term relaxations from climate-driven warming and desalination predicted for the Baltic Sea (Jonsson et al., 2018).

The interacting influence of large- and local-scale gradients in water parameters leads to a pronounced variability in environmental conditions of coastal ecosystems in the Baltic Sea. To further develop our understanding of the dynamics communities are experiencing in shallow waters, a monitoring of water parameters along the Baltic coast of Schleswig-Holstein (Germany) was designed. In this contribution we are presenting the obtained data for water temperature, salinity, dissolved oxygen content and nutrient concentrations recorded within the period of 2016–2018.

All datasets are deposited as a collection at PANGAEA and can be accessed via the following DOI: 10.1594/PANGAEA.895257 (Franz et al., 2018).

The present environmental monitoring provides data for dissolved inorganic nutrient contents recorded twice a month and continuous records of temperature, salinity and DO concentration. The obtained data provide a good picture of the highly dynamic shallow waters at the Baltic coast of Schleswig-Holstein, Germany. The total range (Fig. 6) as well as the recorded maximum diurnal changes (see Sect. 4.2) of continuously measured parameters highlight the enormous environmental fluctuations biota are facing in these habitats. However, the closer examination of data obtained for Stations 2 and 13, which were exemplarily depicted, shows that the extent of environmental variability locally depends on spatial and temporal factors (Fig. 7). This is especially evident for measurements of salinity and DO concentration. At both stations, the trends in salinity neither followed a recurring seasonal pattern nor was there a systematic difference between the stations, which could be expected according to the results of the applied regression using median salinities of the overall monitoring period (see Sect. 4.2). In contrast, the observed fluctuations have different timings at the stations, indicating that local influences like rainwater discharges could co-occur with large-scale processes, e.g., inflow events of North Atlantic waters (Snoeijs-Leijonmalm et al., 2017; Reusch et al., 2018). Local factors might also explain the rapid increase in salinity within a single day at Station 13 that was paralleled by DO concentrations dropping to 1.7 mg L-1, an indication of oxygen-depleted water being upwelled at the coast. The variability of measured DO concentrations also exhibits local differences and, in addition, seasonal patterns. The overall stronger variability of DO data at Station 13 could be related to higher nutrient loads compared to Station 2. The median nitrate concentrations at this station were 2 times higher; the recorded maximum exceeded even 7.5 times the value of Station 2 (Table 5). In consequence, biological productivity could be enhanced at this station, leading to larger diurnal fluctuations (Gubelit and Berezina, 2010), e.g., recorded as a drop by 50 % in DO concentration in August 2016. The generally lower variability in winter at both stations could be mediated by lower temperatures since biological processes are slowed down in this period, keeping oxygen concentrations more stable. The likelihood of detecting such events and processes by repeated single measurements is very low; thus continuous records provide the chance to unravel short-term dynamics that might have not been described by common monitoring efforts.

The measured nutrient loads exhibited seasonal fluctuations with the highest concentrations in autumn and winter (Fig. 5). Noticeably, the present extremes in nitrate concentrations exceed 6.6 times the highest recorded concentration measured at a close-by time series station between 1957 and 2014 at a depth of 1 m (https://www.bokniseck.de/, last access: 1 July 2019; Lennartz et al., 2014; Bange and Malien, 2015). The Boknis Eck time series station is situated 2.2 km away from Station 3, but located further offshore. Therefore, the time series station could be less affected by local river runoff and groundwater seepage that might explain the increased nitrate loads in water samples of this study (Szymczycha and Pempkowiak, 2016). Indeed, groundwater seepage has been described as a significant pathway in the hydrology of Eckernförde Bay (Schlüter et al., 2004), a location that is covered in this monitoring and where measured nitrate concentrations were among the highest (Stations 3 and 4; see Table 5).

The employment of self-contained monitoring systems, as presented here, always poses the risk of data losses due to failure of data loggers. Since the used data loggers have to be read out manually, a malfunction could be detected with temporal delay, leading to substantial gaps in the records. Real-time data systems like measurement buoys provide the advantage of wireless data transmission, allowing a continuous control of functionality. Furthermore, sensors attached to buoys are less susceptible to sediment dynamics. The coverage of the logger station by sand will potentially lead to extended gaps in coming datasets since four of the setups were not found during the last readout. Nevertheless, the advantages of real-time data systems go along with an increase in expenses for single measurement systems by up to an order of magnitude. Therefore, in this study, independent data loggers have been applied, keeping costs lower and favoring a higher replication of the measurement stations. The average retained number of data (96 %) from the raw dataset of temperature, salinity and DO measurements shows that this strategy can be a worthwhile alternative to more expensive monitoring systems. For future deployments the geological characteristics of the sampling sites should be examined in more detail, e.g., to avoid areas of pronounced sediment transport.

The obtained, temporally fine-scaled data of this study could be utilized for diverse purposes. The records could support the development and skill assessment of models (oceanographic, biogeochemical and ecological) or provide background information for the definition and probability of extreme events like heat waves or coastal upwelling (Bennett et al., 2013; Hobday et al., 2016; Bates et al., 2018). It may serve as a base for more realistic experimental approaches that not only apply mean treatment levels but also consider the range and magnitude of environmental variation (Wernberg et al., 2012). Moreover, the data could help to identify mechanisms behind observed changes in biological monitoring programs performed in the same area. Since the assessment will be continued for the coming years, the temporal coverage of the dataset is going to broaden in the future and may even enhance its applicability.

Versatile possible uses of the presented data underline the great value of observational programs. As the human population is concentrated along the coastlines, signals of anthropogenic changes are expected to be particularly strong. However, the detection of change in coastal areas is a challenging task since human pressures and natural variability caused by the sea–land interface vastly overlap (Cloern et al., 2016). At this point, environmental data of high resolution can be a useful tool, as they allow a better differentiation of natural variability from signs of human impact. Traditional monitoring strategies (e.g., ship-based surveys) can usually not provide the necessary information since temporal resolution is low and access to shallow waters limited. Self-contained monitoring systems therefore appear as a suitable alternative to bridge this knowledge gap. The presented approach displays an opportunity not only to researchers in better understanding the system dynamics, but also for policymakers in choosing appropriate measures in future environmental protection (Helmuth et al., 2014).