Field experiments investigating biodiversity and ecosystem functioning
require the observation of abiotic parameters, especially when carried out in the
intertidal zone. An experiment for biodiversity–ecosystem functioning was set
up in the intertidal zone of the back-barrier salt marsh of Spiekeroog Island
in the German Bight. Here, we report the accompanying instrumentation,
maintenance, data acquisition, data handling and data quality control as well
as monitoring results observed over a continuous period from September 2014
to April 2017. Time series of abiotic conditions were measured at
several sites in the vicinity of newly built experimental salt-marsh islands
on the tidal flat. Meteorological measurements were conducted from a weather
station (WS,
Biodiversity is changing at an unprecedentedly high rate (Mace et al., 2005), reflecting the anthropogenic alteration of Earth's ecosystems (Rockström et al., 2009). As a consequence, research on biodiversity–ecosystem function (BEF) relationships has become a major facet of ecology and evolutionary biology (Balvanera et al., 2006; Cardinale et al., 2006; Hillebrand and Matthiessen, 2009; Cadotte et al., 2008; Gravel et al., 2011). Research on BEF has been dominated by experimental studies manipulating the number of species in a trophic group (Hillebrand and Matthiessen, 2009). However, the change in the number of species is not the predominant pattern in biodiversity change, which more frequently comprises altered dominances and species turnover (Hillebrand et al., 2017). Moreover, the mechanisms involved in altering species composition, i.e., immigration and extinction, are usually experimentally prohibited in most BEF studies, e.g., via weeding. On the other hand, there have been recent advances in understanding the interaction between regional community assembly, local dynamics and the biodiversity-functioning aspects (Hodapp et al., 2016; Leibold et al., 2017) that should be incorporated in BEF experiments.
The intertidal zone represents an interface between terrestrial and marine processes and biodiversity. This area is sensitive to climate change and is heavily impacted by anthropogenic activities (Cooley and Doney, 2009; Ekstrom et al., 2015; Haigh et al., 2015; Mathis et al., 2015). Within the intertidal environment, salt marshes have increasingly gained attention in times of sea-level rise (Kirwan and Megonigal, 2013; Balke et al., 2016), but studies on meta-community dynamics remain scarce.
Based on this backdrop the project BEFmate “Biodiversity–Ecosystem Functioning across marine and terrestrial ecosystems” was conducted in March 2014–December 2017, aiming to quantify the dynamics of biodiversity and the associated functions of salt marsh and tidal flat ecosystems. For this purpose, a series of experimental islands were set up in September 2014 on the back-barrier tidal flat of Spiekeroog island in the German East Frisian Wadden Sea (Balke et al., 2017) and were mirrored by salt-marsh enclosed plots located within the nearby salt marsh. Here we report abiotic parameters observed from 23 sensors installed either near the experimental islands, within the island structures themselves or within the nearby salt marsh as well as meteorological data from a locally installed weather station. We describe the instrumentation, data handling and results observed over a period of 32 months starting from the middle of September 2014. We further discuss the data and results with respect to validity and potential limitations of the observational setup.
Experiment setup with
Scheme of an experimental island showing three elevational levels representing natural salt-marsh zones. The islands are built on an average of 0.8 m NHN (normal height null). The experimental islands' outer hull (galvanized steel) is patterned in the upper area to allow for a lateral water exchange. Lower area pattern is only meant to reduce weight and ease construction.
The island of Spiekeroog is located in the southern North Sea (Fig. 1) and is part of the Wadden Sea, which has been renowned as an UNESCO world natural heritage site since 2012. Twelve experimental islands (I1–I12) were built in the back-barrier tidal flats at distances of 240 m (I12) to 460 m (I1) from the southern salt marsh of the island of Spiekeroog. The experimental islands were distributed unevenly over 810 m from east to west at an elevation of 0.8 m NHN (standard height zero), with a mean tidal range of 2.7 m (Fig. 2, numbered 1–12 from east to west). The islands were built in a northeast–southwest direction, with the lowest elevation at the northeastern end of the island. The actual sensor position on the islands was determined by the local bathymetry, since the experimental islands encompass three different elevation levels (Fig. 3) that consist of the reflecting pioneer zone (Pio; 1.5 m NHN), lower salt-marsh zone (Low; 1.8 m NHN) and upper salt-marsh zone (Upp; 2.1 m NHN) of natural salt marshes. Half of the experimental islands were filled with mudflat sediment and left bare, whereas half of them were additionally transplanted with sods from the lower salt-marsh zone of the natural adjacent salt marsh. Experimental islands represent a treatment of dispersal limitation, constraining community assembly on the islands. Additionally 12 equally treated salt-marsh enclosed plots (S 1–12) were created (Fig. 2) that reflect unlimited dispersal. In addition to experimental plots, six control plots (C) per salt-marsh elevation zone were marked but left natural to compare established communities with the community assembly on dispersedly limited island plots and unlimited salt-marsh plots. The experimental design and setup of the experimental islands are not subject of this work and are described in detail in Balke et al. (2017). Abiotic conditions were measured at several sites due to the involvement of wide selection of parameters. Details concerning the individual sensors, their location, data provided and the associated methods of data handling are provided in the following subsections. All positions, coordinates and elevations of sensors are indicated and provided in Table 1.
Overview of all installed loggers at the experimental islands (I) and the salt-marsh enclosed plots (S) as well as the nearby installed sensors. WS: weather station, RCM: recording current meter, TWR: tide and wave recorder, DL: data logger.
Meteorological data were collected near the experimental setup (see Table 1),
with a locally installed weather station located approximately 500 m north
of the southern shoreline (53
A RCM9 LW recording current meter (AADI, Aanderaa, Bergen/Norway; RCM DCS
4220) with additional temperature (3621), conductivity (3919) and pressure
(4017) probes was deployed for deriving hydrographic conditions (see Table 1).
The device was bottom mounted through a buried H-anchor between islands
6 (I6) and 7 (I7) (53
Local tide and wave conditions were recorded with a RBRduo
TD
Several data loggers were installed on the experimental islands as well as in the salt-marsh enclosed plots for the observation of groundwater level, temperature, light and salinity (see Table 1). In all cases the date and time is given in UTC and all post-processing was performed using MATLAB (R2012b). Quality control was applied as described in Sect. 2.2 for the weather station (WS).
Structure of the weather station (WS) datasets at PANGAEA.
To continuously observe the flooding and the groundwater levels
on the experimental islands as well as in the salt marsh, pressure
loggers were deployed in dip wells within the experimental setup at
different elevational levels. Six HOBO® U20L Water Level
Logger (onset® HOBO® Data Loggers, Bourne,
MA, USA; S/N 10685287, 10685288, 10685289, 10685290, 10685291, 10685292;
Hobo-P) as well as a DEFI-D Miniature Pressure Recorder (S/N OA5K008;
DEFI-D) were deployed. All water level loggers were pre-calibrated by the
manufacturer. Recorded data were internal logged until the readout afield with
the Hobo Underwater Shuttle (U-DTW-1) and the HOBOware Pro (V3.7.4) software
respectively with the DEFI Series software (V1.02). For depth calculations,
pressure data were manually corrected by atmospheric pressure. Accordingly,
a HOBO® U20L Water Level Logger was installed outside the
dip wells at a higher elevation, attached on a steel pole at the upper zone
of island 3. All loggers were initially calibrated to get the exact height
inside the dip well. Data of one of the HOBO loggers had to be corrected with
Structure of the recording current meter (RCM) datasets at PANGAEA.
Structure of the tide and wave recorder (TWR) datasets PANGAEA.
Temperature in the sediment surface layer (in approximately 0.05 m depth) was measured with six DEFI-T miniature temperature recorders (JFE Advantech Co., Ltd., Tokyo; DEFI-T). The manufacturer pre-calibrated temperature recorders and were installed on the experimental islands and in salt-marsh enclosed plots at different elevation levels. Recorded data were internally logged until the readout with the DEFI Series software (V1.02).
Light availability was measured with six locally installed light intensity
loggers on the experimental islands as well as in the salt-marsh plots
at different elevation levels. The DEFI-L Miniature Light Intensity Recorder
(JFE Advantech Co., Ltd., Tokyo; DEFI-L) used here were pre-calibrated by
the manufacturer. Recorded data were internally logged until the readout with the
DEFI Series software (V1.02). Due to different calibrations for underwater
or above-water applications, raw data were processed differently and merged
depending on water levels. The processed pressure and depth data of the
RBRduo TD
Structure of the water level logger (DL-W) datasets at PANGAEA.
Structure of the temperature logger (DL-T) datasets at PANGAEA.
Structure of light logger (DL-L) datasets at PANGAEA.
Structure of conductivity logger (DL-C) datasets at PANGAEA.
Two HOBO conductivity loggers (Onset Computer Corporation, Bourne, MA, USA) were installed inside of dip wells on the experimental islands as well as in the salt-marsh enclosed plots at the pioneer zone. The conductivity logger used here was the HOBO U24 Conductivity Logger U24-002-C (S/N 10570000, 10599255). The conductivity loggers were pre-calibrated by the manufacturer. Recorded data were internally logged until the readout afield with the Hobo Underwater Shuttle (U-DTW-1) and in the following with the HOBOware Pro (V3.7.4) software. An automatic calculation of salinity was conducted within the software according to PSS-78 using the measured conductivity and temperature. Due to fluctuations in the groundwater level, conductivity loggers periodically became dry, especially in the beginning of the deployment. Data until October 2015 are therefore very scattered. The depth of conductivity loggers was thereupon adjusted to the bottom of the dip wells, assuring a constant coverage with water. Data from dry sensors was removed, using a salinity of 20 as a threshold value. As a reference, soil samples of all plots on the experimental islands and salt-marsh enclosed plots were sampled to analyze pore water salinity in laboratory (data not shown here). Comparative data of meteorological and hydrographic conditions for validation processes were taken from the nearby Time Series Station – Spiekeroog (TSS) at Otzumer Balje (Holinde et al., 2015; Baschek et al., 2017).
Available data shown in black bars for each sensor or logger type over the sampling period from 18 September 2014 to 18 April 2017.
Local wind roses (wind speed and wind direction) for storm seasons
(1 October–31 March) in 2014 and 2015
All datasets described herein are available at the World Data Center PANGAEA
(
Data of meteorological observations are available at PANGAEA from November 2014 to
April 2017 (
Current conditions in the storm season
Continuous current data are available on PANGAEA for September 2014 to October 2015
(
Tide and wave data are available on PANGAEA from October 2014 to April 2017
(
Data for groundwater level on the experimental islands and
salt-marsh enclosed plots are available on PANGAEA from June 2015 to April 2017 (
Temperature data for the surface sediment layer of the experimental islands
and salt-marsh enclosed plots are available on PANGAEA for September 2014 to
April 2017 (
Measured light availability on the experimental islands and
salt-marsh enclosed plots are available on PANGAEA from September 2014 to April 2017
(
Data of conductivity measurements inside dip wells on the experimental
islands and salt-marsh enclosed plots are available on PANGAEA for May 2015
to April 2017 (
An overview of sensor and logger data availability over the 944-day sampling period. D 1 – sum of days with data available; D 0 – sum of days with data absent; % 1 – proportion of days with data available; % 0 – proportion of days with data absent; D x – total available days specified for each sensor application, i.e., from first deployment to last measurement; % x1 and % x0 are the corresponding percent availabilities. Sum 1 – number of measurements available; Sum 0 – number of measurements missing; % S 1 – proportion of measurements available; % S 0 – proportion of measurements missing; Sum x – total available measurements specified of each sensor application, i.e., from first data point to last measurement; %S x1 and %S x0 are the corresponding percent availabilities. WS: weather station, RCM: recording current meter, TWR: tide and wave recorder, DL: data logger.
Sensor operation encompasses the time span of 32 months, from 18 September 2014 to 18 April 2017 (944 days). Figure 4 illustrates the data availability over the whole period, and Table 9 provides total availability in days as well as in percent.
Calculated water depth (black dots) at 0.71 m NHN, next to islands 6 and 7. The blue line represents the elevation of the pioneer zone (Pio), the green line shows the height of the lower salt marsh (Low) and the red line describes the upper salt-marsh zone (Upp). Thus, water level data can give information about flooding periods at the three elevation levels. Low-tide data (all data below 0.05 m) were removed before plotting the data. To provide m NHN as a basis, 0.71 m were added to the data points.
Calculated water depth (black line) during different weather
conditions.
Within this time frame meteorological data were available from 19 November 2014 to 18 April 2017 on 571 days (corresponding to 60.49 % availability). Absent days resulted from malfunctions and maintenances from 29 January 2015 to 11 March 2015, 4 November 2015 to 14 November 2015, 25 March 2016 to 28 October 2016 and 1 February 2017 to 21 March 2017.
Current meter operation was possible from 18 September 2014 until 6 October 2015 on 335 days (35.49 %). Further operations occurred in November and December 2013, March 2014, and June 2014 for preliminary investigations. Missing days were resulted from local maintenance and readouts. Due to a malfunction, the sensor had not been operating since October 2015.
Continuous observations of wave and tide data were feasible from 1 October 2014 to 18 April 2017 on 812 days (corresponding to 86.02 % availability). Absent days were the result of local maintenance and readouts as well as some extended services from 19 October 2014 to 28 October 2014, 17 August 2015 to 23 September 2015 and 8 November 2016 to 24 January 2017.
Groundwater level on the experimental islands and the salt-marsh enclosed plot.
Surface layer temperature and temperature differences over the whole time frame with the DL-T2 (I3 Low).
Within the total time period water level measurements were performed from 20 June 2015 to 18 April 2017 on 578 days (572 days within the salt-marsh enclosed plot), resulting in a 61.23 % (60.59 %) availability. Absent days resulted from malfunctions and maintenances from 16 February 2016 to 18 May 2016 as well as some local readouts. Groundwater level data of the DEFI-D logger ends at 10 January 2017 due to a missing readout in April 2017. The coverage of water level measurements is therefore 50.85 % (480 days).
Minima, maxima and mean values as well as median and standard deviation of the weather station (WS) for the time frame of 19 November 2014 to 18 April 2017 on 571 days. See Table 2 for parameter list.
Minima, maxima and mean values as well as median and standard deviation of the recording current meter (RCM) for its application time from 18 September 2014 until 6 October 2015 on 335 days.
Recordings of the surface layer temperature were available on 931 days, excluding the period from 10 January 2017 to 24 January 2017 due to maintenance (98.62 % availability). However, temperature data of two loggers within the salt-marsh enclosed plots (S2 Upp and S2 Low) exhibit a shorter time of deployment. An application was possible at 295/380 days (31.25 %/40.25 % availability) from 16 December 2014 to 6 October 2015 (S2 Upp and S2 Low) and further from 24 January 2017 to 18 April 2017 (S2 Low).
Light intensity and availability was recorded over the whole period from 18 September 2014 to 18 April 2017 for 931 days, except for the period from 10 January 2017 to 24 January 2017 due to maintenance (98.62 % availability). Two loggers (I3 Pole, Seafloor) were first installed on 7 October 2015 with a total application time of 547 days, corresponding to 57.94 % availability. A light logger within the salt-marsh enclosed plot (S3 Low) had to be removed from 19 November 2014 to 16 December 2014 due to maintenance. A light logger on the experimental island (I3 Low) had to be temporarily removed between 23 September 2015 and 11 April 2016.
Data of the conductivity logger on the experimental island (I3 Pio) are available from 6 May 2015 to 16 February 2016 and from 18 May 2016 to 18 April 2017 on 624 days (66.10 % availability). The logger located in the salt-marsh enclosed plot (S3 Pio) recorded water conductivity from 7 May 2015 to 8 October 2015, 3 December 2015 to 16 February 2016 and from 18 May 2016 to 18 April 2017, 567 days in total (60.06 % availability). Gaps in the dataset resulted from a malfunction and maintenance.
Minima, maxima and mean values as well as median and standard deviation of the tide and wave recorder (TWR) from 1 October 2014 to 18 April 2017 on 812 days.
Minima, maxima and mean values as well as median and standard deviation of the groundwater level data (DL-W 1-6) for each logger application time from 20 June 2015 to 18 April 2017 on 578 days (572 days within the salt-marsh enclosed plot).
Mean monthly temperature over the whole application time. Since the temperatures on the experimental islands are very similar, a clear offset of the salt-marsh enclosed plot temperature could be identified in winter and summer months.
Local wind diagrams for three winter storm seasons and one summer season are
shown in Fig. 5. Furthermore, minima, maxima and mean values as well as
the median and standard deviation of the weather station for the time frame of
19 November 2014 to 18 April 2017 on 571 days are listed in Table 10. Within
the application time, wind speeds of less than 25 m s
Current conditions for one storm season and one summer season are shown in
Fig. 6. A main current direction from the southwest to the northeast was clearly
identified, showing the good orientation of the experimental islands.
Furthermore, minima, maxima and mean values as well as the median and standard
deviation of the RCM data from 18 September 2014 until 6 October 2015 on 335 days
are listed in Table 11. A maximum current speed of 107.05 cm s
Calculated sum of light intensity per day.
Salinity values within dip wells on the experimental island and salt-marsh enclosed plot. Due to fluctuations in the groundwater level, conductivity loggers periodically became dry, especially in the beginning. Thus, data until October 2015 are scattered. The depth of conductivity loggers was thereupon adjusted deeper in the dip wells, assuring a constant covering of water.
Minima, maxima and mean values as well as median and standard deviation of the temperature data (DL-T 1-6) from 18 September 2014 until 18 April 2017 on 931 days.
Minima, maxima and mean values as well as median and standard deviation of the light intensity data (DL-L 1-6) for each logger application time.
Minima, maxima and mean values as well as median and standard deviation of the salinity data (DL-C 1-2) for each logger application time.
The water depth calculated from pressure data of the tide and wave recorder
(TWR) can be seen in Fig. 7. A seasonal pattern of water depth is
exhibited and the highest water depths were reached during storm season. However,
water level reached the upper salt-marsh zone (2.0 m NHN) several times
(e.g., April 2015, July 2015 and August 2016). During the storms Elon and Felix in
January 2015, the highest water level was observed at 3.62 m as well as the highest
wave at 2.14 m (Fig. 8). The highest wave energy (400.90 J m
The depth of the groundwater level achieved from pressure logger on the experimental islands and in the salt-marsh enclosed plots can be seen in Fig. 9. Differences in the water level were observed, especially during low water. This could be a result of various factors, ranging from diverse water consumption of plants to less flooding on higher elevations or leaks in the plastic bags inside an experimental island. Statistics of groundwater level data (DL-W 1-6) for each logger application time from 20 June 2015 to 18 April 2017 on 578 days (572 days within the salt-marsh enclosed plot) are listed in Table 13.
Surface layer temperatures for one experimental island and its three
elevational zones as well as one elevation zone within the
salt-marsh enclosed plots are shown in Fig. 10. To assess temperature differences the
DL-T2 Logger (I3 Low) was taken to compare both differences within one
island's elevational zone (DL-T1 I3 Pio, DL-T3 I3 Upp) and between the same
elevational level and those of the salt-marsh control plots (DL-T4, S3
Low). All temperatures are very similar and showing only less differences,
especially I3 Upp and I3 Low, with
To calculate the light sum per day, measured light intensity (Quantum (
Salinity values achieved from both conductivity logger on
the experimental island and the salt-marsh enclosed plot can be seen in Fig. 13.
Due to fluctuations in the groundwater level, conductivity loggers
periodically became dry, especially in the beginning. Thus, data until October 2015
are scattered. The conductivity loggers were thereupon adjusted deeper in
the dip wells assuring a constant covering of water. The mean salinity for the
experimental island is 27.95
The dataset described within this document is based on sensor information that has been uploaded to the PANGAEA database (see Sect. 2.3).
The BEFmate project included a variety of experiments dedicated to investigate biodiversity–ecosystem functioning across marine and terrestrial ecosystems, utilizing 12 experimental islands in the back-barrier tidal flat of Spiekeroog Island. Abiotic conditions were recorded from a suite of 23 different sensors installed at different locations in the vicinity of the experimental islands, on the islands themselves and in the nearby salt marsh. Data described here covers the period from September 2014 to April 2017 and has been published in seven datasets in the World Data Center PANGAEA. Data coverage within the period ranged from 35 % for the recording current meter (RCM) that failed in October 2015 to 99 % for six data loggers. With 17 sensors covering 80 % or more of the period of interest, a very good coverage was achieved. Additional data from pre-experiment investigations with the RCM between November 2013 and June 2017 was added to the dataset. Seasonal and tidal dynamics as well as storms were covered, and these data are available for interpretation in further contexts.
For future operations, data availability can be further increased if a rotational system for maintenance is applied, provided that spare sensors are available. Furthermore, the online data transfer of central information would not only increase the data availability, but meteorological and hydrographic conditions could also indicate the need to attend the experiment. For example, indicating a need to take action to prevent damage during extreme weather events. Non-invasive remote sensing sensor techniques can provide complementary data, to avoid fouling issues, as demonstrated successfully at the nearby Time Series Station – Spiekeroog (Garaba et al., 2014; Schulz et al., 2016). Additionally camera systems should be applied, providing a visual impression of the overall scene and detailed information, e.g., the process of flooding for different levels. Recently the RCM failure within the BEFmate project has inspired the development of a machine-learning environment that creates a virtual sensor enabling the compensation for single sensor dropouts (Oehmcke et al., 2017a, b). Finally, data quality assurance and quality measures should be further developed to reduce the workload of manual data curation while improving data availability in near-real time.
OZ was responsible for the scientific approach and performed the interpretation of results. He drafted and prepared the manuscript. DM was responsible for technical maintenance and laboratory analysis and performed fieldwork, data sampling and analysis, created figures, and contributed to the writing process of the manuscript. HH was the principle investigator of the BEFmate project and contributed to the writing process. MK was the principle investigator of one of the BEFmate subprojects and was responsible for the conception of the experimental islands. TB and KL were responsible for maintenance of the experimental islands as well as fieldwork coordination. All authors contributed to proofreading of the manuscript.
The authors declare that they have no conflict of interest.
The authors are very grateful to Kai Schwalfenberg and Claudia Thölen for their tremendous support in the field and laboratory. Sincere thanks to Daniela Voß, Ursel Gerken, Kathrin Dietrich, Nick Rüssmeier, Rohan Henkel, Jule Beßler, Franziska Wöhrmann and Hauke Haake for technical, laboratory and fieldwork support. Special thanks to Helmo Nicolai and Gerrit Behrens for their technical and logistical support. The support and cooperation with Nationalparkverwaltung Niedersächsisches Wattenmeer and the Umweltzentrum Wittbülten is acknowledged as well as the support of Rainer Sieger at PANGAEA. Thanks to the BEFmate colleagues and all other helping hands in the field during sampling campaigns, especially Regine Redelstein. The BEFmate project (Biodiversity–Ecosystem Functioning across marine and terrestrial ecosystems) was funded by the Ministry for Science and Culture of Lower Saxony, Germany under project number ZN2930. The feedback from two reviewers is gratefully acknowledged. Edited by: Giuseppe M. R. Manzella Reviewed by: Jaume Piera and one anonymous referee