This study (Gescha 2) analyses the impact of the construction of eleven offshore wind farms (OWFs) and offshore converter platforms (OSS) built in the German North Sea and adjacent Dutch waters in the period 2014-2016 on harbour porpoises (Phocoena phocoena). The study is a direct follow-up to the Gescha 1 project, which investigated the effects of pile driving of the first eight wind farms in the German Bight (2009-2013). For the present study, also a combined dataset from 2010 to 2016 could be considered, containing most Gescha 1 and all Gescha 2 data. The dataset combines porpoise monitoring data from passive acoustic monitoring using Porpoise Detectors (CPODs) and digital aerial survey data with measured data on noise levels in 750 m and 1500 m distance from the piling location as well as other piling characteristics. These data were analysed in order to describe the response of harbour porpoises to pile driving activities, most of which took place under operation of noise-mitigation systems with the aim to reduce disturbance effects on porpoises.
Prior to investigating piling effects on harbour porpoises, baseline analyses were conducted to identify the seasonal distribution of porpoises in different geographic subareas. Daily CPOD data and digital aerial survey data uncovered seasonal patterns with often higher densities in spring and summer, especially in the north-eastern part of the German Bight. Lowest porpoise abundance was found in the central part of the German Bight with deep waters, whereas higher densities were found in the more coastal subareas. Seasonal patterns differed from subarea to subarea. In the northern German Bight porpoises showed a pronounced summer peak, whereas in the eastern subarea animals also showed considerable activity in autumn. In overall, this resulted in highest porpoise densities and detections in summer within and next to the SAC Sylt Outer Reef. Another high-density area was identified near the SAC Borkum Reef Ground in the south-western German Bight and adjacent Dutch waters. Here, highest densities were found in late winter. These results are in line with previous findings and point towards a very stable occurrence of porpoises within the German Bight. Both CPOD and aerial survey datasets showed matching regional porpoise phenology trends, indicating a high degree of correlation between these completely independent datasets.
Noise measurements at 750 m and 1500 m distance from piling locations for all German wind farms constructed between 2014 and 2016 were combined with CPOD porpoise monitoring data. Analyses of noise levels revealed a high variability within each wind farm. However, both the variability of the sound values and the average noise level were significantly reduced with the noisemitigated pile drivings of Gescha 2, and noise levels were mostly below the BSH's mandatory noise limit of 160 dB SEL05 at a distance of 750 m from a piling location. Pile driving carried out during the study period of Gescha 2 had on average 9 dB lower noise levels than those of Gescha 1, and had on average more than 15 dB lower noise levels than those measured during unmitigated piling, which can be attributed to a significant improvement in noise-mitigation systems in recent years. Still, some variability in the noise measurements remained, which may result from certain environmental factors affecting sound propagation, such as water depth, substrate, and wind speed. The present study shows that noise-mitigation systems used between 2014 and 2016 had improved considerably when compared to 2010-2013 and worked consistently well in the German Bight.
Establishing of the relationship of noise level to porpoise response is crucial for environmental impact assessment based on the noise prognosis for specific projects.
Since noise-mitigation technology became more efficient over the last years and noise levels were reduced significantly, the expectation was that the displacement range and duration for porpoises due to piling noise should have been reduced accordingly.
However, this was not the case. The effect range regarding porpoise detection rates based on all hourly CPOD data during mitigated pile driving from all projects within Gescha 2 was at 17 km (std. error range: 15-19 km), and the effect duration in close range lasted from 28 hours (lower s.e.: not available; upper s.e.: 22 hours) before until 48 hours (lower s.e.: 35 hours; upper s.e.: not available) after stop of pile driving. These values were similar to those obtained at Gescha 1, and thus no reduced displacement effect could be shown when comparing Gescha 2 to Gescha 1. Analyses of 12 digital aerial surveys, conducted during or up to 12 hours after stop of pile driving, showed an effect range of 11.4-19.5 km to piling sites and thus confirmed the effect range found by CPOD data.
We discuss this outcome with five explanatory approaches which might be relevant alone or in combination.
1. Stereotypical escape distance within a certain noise-level range: Results of the hourly CPOD data indicate that there is no correlation between noise level and displacement range below noise levels of 165 dB SEL05 at 750 m from piling locations. Below this value the effect range seemed not to be further reduced. This might be explained by animals maintaining a certain minimum escape distance independent of the respective noise level if it is below this value and within a certain intermediate range. Thus, animals may react stereotypically as soon as pile-driving noise exceeds a certain individually differing unknown threshold level that must be regarded in the context of a seasonally and sitespecific different condition of animals. However, regarding piling-noise levels we only had access to the broadband SEL05 cut off at 20 kHz, and could not refer to noise levels being weighted according to the hearing spectrum of harbour porpoises; hence, we might not have dealt with the noise most relevant for porpoises.
2. Increasing relative importance of the displacement effect of seal scarers with better noise reduction: Based on theoretical considerations and on measured values at a distance of 750 m from piling sites, it can be assumed that seal scarer noise up to a distance of approximately 20 km is clearly better audible for harbour porpoises than pile-driving noise mitigated by well-functioning noise-mitigation systems. This cannot explain the farreaching effect by its own because seal scarers were also used in projects where the response range was rather short. Thus, seal scarer effects cannot be the only explanation, but might have contributed to the fact that no improvement of effect range and duration from Gescha 1 to Gescha 2 was found.
3. Other construction-related noise, commencing already prior to the start of deterrence and driving a large part of the animals away from those noise sources: A reduction of detection rates before deterrence and pile driving was shown for all wind farms investigated during both Gescha studies. Thus, there has to be a reason why animals leave the area up 5 to 24 hours before the start of the seal scarer. However, displacement during pile driving is clearly stronger than the effect before piling, at least in the close range of a few kilometres. Since the animals' only sense of perceiving a disturbance over many kilometres is their hearing system, it is assumed that anthropogenic sounds associated with pile driving are the trigger for the reaction.
4. Cumulative effects due to tight piling sequence: Piling schedules became much tighter with latest OWFs. By this fact it cannot be completely excluded that the time between consecutive pilings was partly so short that animals did not have enough time to come back before the next piling started. Even though we tried to include several variables into our models to capture this possible phenomenon these were often thrown out during the model-selection process. By applying to the hourly CPOD dataset a model approach with a reduced dataset where at least a break of 72 hours occurred between two consecutive pilings, we tried to minimise the possible influence of tight piling sequences. Still, the displacement range was similar to the range for OWFs built in the period 2010-2013 when piling sequences were much less tight. However, the cumulative effect might not have been captured adequately by only three days of break between pilings.
5. Habitat characteristics at different OWF areas: Above all, there also exists a high variability of porpoise occurrence due to different small-scale habitat structures, which might have consequences on a seasonal and inter-annual time scale in the North Sea. A good example in this context is the patchy presence of sandeel and sand goby in and around the DanTysk and Sandbank project areas, which may have contributed to different displacement radii within these two areas. Hence, since the response of harbour porpoises to disturbance also depends on habitat use and habitat characteristics, the unexpectedly high effect range for Gescha 2 might partly be attributable to habitat differences between and among the Gescha 1 and 2 OFW projects.
When looking at long-term trends of daily harbour porpoise detection rates obtained by CPODs, spatial differences occurred among the investigated subareas. In the eastern and, less pronounced, the southern part of the German North Sea and adjacent Dutch waters, we found an increasing trend from 2010 to 2016, whereas porpoise detections remained relatively constant in the northern part, and decreased in the central part of the study area. The latter subarea, however, was less important for harbour porpoises, as it generally showed low porpoise detection rates. Regarding the entire study area, porpoise detection rates increased from 2010 to 2016. Hence, cumulative OWF construction activities in the German Bight apparently did not have any measurable negative effect on population level.
In conclusion, the future development of noise-reduction measures, with the aim of reducing the radius of disturbance of harbour porpoises, must be critically reviewed, as no improvement regarding piling effects on harbour porpoises was found. Nevertheless, despite of large disturbance radii no negative effect on population level was observed.