Ecological impact of land reclamation on Jiangsu coast(China):A novel ecotope assessment for Tongzhou Bay

Jos R.M.Muller*,Yong-ping Chen,Stefn G.J.Arninkhof,Ying-Chi Chn,Theunis Piersm,Dirk S.vn Mren,f,Jin-feng To,Zheng Bing Wng,f,Zheng Gong

aFaculty of Civil Engineering and Geosciences,Delft University of Technology,Delft 2600 GA,the Netherlands

bBaggermaatschappij Boskalis BV,Papendrecht 3350 AE,the Netherlands

cCollege of Harbor,Coastal and Offshore Engineering,Hohai University,Nanjing 210098,China

dGroningen Institute for Evolutionary Life Sciences(GELIFES),University of Groningen,Groningen 9700 CC,the Netherlands

eDepartment of Coastal Systems,NIOZ Royal Netherlands Institute for Sea Research and Utrecht University,Den Burg 1790 AB,the Netherlands

fUnit of Marine and Coastal Systems,Deltares,Delft 2600 MH,the Netherlands

Abstract

Keywords:Mapping;Ecotope;Ecotope map;Intertidal mudflats;Migratory shorebirds;Reclamation;Tongzhou Bay;Jiangsu coast

1.Introduction

Over the lastfew decades,China has experienced immense economic development,during which the Chinese central government has promoted the urbanization and industrial development of coastal provinces by reclaiming large sections of coastal zones(Ma et al.,2014;Tian et al.,2016).A salient contemporary case is Tongzhou Bay on the Jiangsu coast(Fig.1).This area is characterized by large intertidal mudflats(e.g.,Yaosha and Lengjiasha)with deep tidal channels(e.g.,Xiaomiaohong and Sanshahong)and has a high potential for the development of agri-aquaculture,as well as the construction of the new deep-sea Tongzhou Bay port(Wang and Wall,2010;Zhang and Chen,2011).Recently,a new masterplan for the development of the Tongzhou Bay port before 2035 has been announced,proposing the reclamation of large sections of the Yaosha shoal(Nantong Leju Network,2019).

Fig.1.Overview of shoals and channels in Tongzhou Bay situated at southern Jiangsu coast(units of elevations:m).

However,the Yellow Sea coastal wetlands form an essential part of the East Asian-Australasian Flyway(Murray et al.,2015).Reclamation activities including those at the Jiangsu coast have shown to be linked to the decline of wetland ecosystem and endangered shorebird species(Murray et al.,2014,2015;Piersma et al.,2016).The proposed Tongzhou Bay port would strongly interfere with the habitat required by migratory shorebirds.Previous modeling efforts have provided a better understanding of the complex hydro-and morphodynamics at Tongzhou Bay(Wang et al.,2012;Xing et al.,2012;Xu et al.,2016;Yao,2016)and the impacts of reclamation activities on these processes(Huang and Liu,2017;Tao et al.,2011;Zhang et al.,2013).However,little is known about the consequences for the wetland habitat.Hydrodynamic conditions are a main determinant of spatial distribution of tidal wetland organisms,such as marsh vegetation(Wu et al.,2017),and ecotope distribution maps are a useful approach to assessing the impact of reclamation activities on Tongzhou Bay on potential habitat zones.These ecotope maps describe the relationship between the abiotic system and biotic elements and are often used as a tool in determining habitat maintenance and development policies(Bouma et al.,2005).

In this study we integrated a hydrodynamic model and unique distributional data of two of the main species of shorebirds,the bar-tailed godwit(Limosa lapponica,BTG)and great knot(Calidris tenuirostris,GK)in the Tongzhou Bay wetlands.First,we developed a hydrodynamic numerical model for the Tongzhou Bay region based on a large-scale model for the entire Jiangsu coast(Su,2016;Yao,2016).The validated model was then used to quantify ecosystem distribution at Tongzhou Bay for the 2012 situation,which was then compared with the distributional data of the shorebirds.Finally,the current reclamation(from 2014 to 2018)and proposed reclamation(from 2019 to 2035)were overlaid on the ecotope distribution to assess their combined impact.

2.Methodology

2.1.Ecotope classification

Ecotopes in the Tongzhou Bay system were quantified with the ZES.1 method(Bouma et al.,2005)that was developed for saline open water ecosystems in the Netherlands,and also applied to the entire Wadden Sea(Baptist et al.,2019).It was assumed that the ZES.1 salt water classification method was also applicable to the Jiangsu coast.An ecotope only describes a potential niche for the possible occurrence of a certain habitat and validation of the findings is therefore required.Regions with relatively homogeneous habitat characteristics can be determined,by selecting relevant abiotic features that dictate the occurrence of certain habitats in a hierarchical manner and defining class boundaries.

Field data were available for this analysis,including salinity,water level,bathymetry,and current velocities(CCCC Third Harbor Consultants Co.,Ltd.,2012;Yao,2016).Hence,the ecotope classification was based on the salinity,substrate,water depth,flow velocity,and dryfall period(e.g.,the period when the mudflats are not inundated during a tidal cycle)for the situation corresponding to 2012.Since the hydrodynamic data were only measured at several locations,a well-validated hydrodynamic model(Section 2.3)was refined and validated for the Tongzhou Bay region to generate the high-resolution spatial hydrodynamic data needed for ecotope mapping.The tidal system is driven by two large-scale rotary tidal currents in the Bohai,Yellow,and East China seas(BYECS),that converge at Jianggang,creating a “rope-skipping”pattern(Xing et al.,2012;Kang et al.,2015).The local hydrodynamic setting is dominated by a semi-diurnal tide,with a consequent tidal range of 3.9 m at Lu¨si Station(Su et al.,2015;Kang et al.,2015).To retrieve the average water depth,tidal current,and dryfall period,two consecutive tidal cycles were simulated.

The definition of the class bounds followed the method described by Bouma et al.(2005).Data analysis revealed small variations in salinity,as well as a homogenous soft substrate over the Tongzhou Bay region,which was considered constant.Depth classification was based on the tidal datum(e.g.,mean high water neaps(MHWN)and mean low water springs(MLWS)),consistent with Bouma et al.(2005).In this study,the classification of high hydrodynamic and low hydrodynamic features was based on the maximum linear velocity of 0.8 m/s.This value was derived from the initiation of bed forms,which implies a change in bed level and ecotope distribution as found for the Wadden Sea(Baptist et al.,2016;Bouma et al.,2005).Further classification based on bed soil composition was not included,due to the lack of highresolution soil data and the smaller impact that would have on habitat occurrence.However,low flow velocities often correspond to silt-rich bed soils.Classification based on the dryfall period was consistent with the ZES.1 method.However,a lower limit of high-range littoral zone of 60% of a tidal cycle was chosen to match the field observations.A color scheme was assigned to each defined ecotope to set up an ecotope distribution map,and the features and names of the ecotopes are shown in Table 1.

2.2.Bird satellite tracking analysis

To verify the ecotope map,satellite-tracking data of two migratory shorebird species,BTG and GK,were used.They are two of the main species using wetlands of the Jiangsu coast(Peng et al.,2017;Piersma et al.,2017)as an important fueling stop during their spring and autumn migration(Battley et al.,2012;Chan et al.,2019b;Piersma et al.,2017).Under the 1999 Environment Protection and Biodiversity Conservation Act of the Australian government,both BTG(the menzbieri subspecies being tracked in this study)and GK are listed as critically endangered(Australian Government,2019).Solar platform terminal transmitters(PTTs,Microwave Telemetry,USA)of 4.5 g and 9.5 g were deployed onto individual birds in Roebuck Bay,Broome(17.98°S,122.31°E)and Eighty Mile Beach (19.40°S,121.27°E),NorthwestAustralia.These transmitters send signals to Argos satellites,from which the birds’locations were estimated(for details see Chan et al.(2019b)).For each species,we estimated kernel density home ranges from all locations collected on the Jiangsu coast from April 2015 to September 2017.Kernel density contours were calculated to show the areas where these species generally occurred(90% and 95% home ranges),and the core(intensively used)areas(50%).

2.3.Numerical model and model setup

The hydrodynamics at Tongzhou Bay were calculated with the Delft3D numerical model(Lesser et al.,2004).In order to accurately predict the tidal flow,the model consisted of an online coupled domain of Tongzhou Bay within a stand-alone model of the Jiangsu coast(Fig.2).The Jiangsu regional model(JRM)encompasses the entire Jiangsu coast,with boundaries from north of the Yangtze River mouth to the Shandong Peninsula(Su,2016;Yao,2016).The JRM used a computational grid of 573×346 grid cells with a resolution varying from 600 to 1200 m.The Tongzhou Bay model(TBM)covered the Yaosha-Lengjiasha ridge as well as the more nearshore areas of Lu¨si and Rudong.The computational grid consisted of 269×468 cells with a resolution varying from 300 to 400 m.

Fig.2.Computational domains of Jiangsu regional model and Tongzhou Bay model(units of elevations:m).

Table 1Ecotope feature classification.

The JRM bathymetry was compiled from measurements through the radial sand ridge(RSR)system in 2006 and electronic navigational charts(Yao,2016).This bathymetry was updated with higher-resolution bathymetric data of the Tongzhou Bay region,which was surveyed in October 2010.

The Reynolds-averaged Navier-Stokes equations for shallow water and the Boussinesq assumption were solved numerically on a staggered grid using the finite difference scheme for a two-dimensional(2D)depth-averaged environment.The JRM was driven by a series of astronomical tides at its two open boundaries(Fig.2).These were derived from a set of 13 tidal harmonic components(i.e.,M2,S2,K2,N2,K1,O1,P1,Q1,M4,MS4,MN4,MF,and MM)from a large-scale tidal wave model of the BYECS for a full morphological year time-series(Su et al.,2015).

2.4.Model performance evaluation

As the model was refined for the Tongzhou Bay region,revalidation was required.The model was validated through quantitative assessments of the calculated and measured values at several monitoring locations at Tongzhou Bay.These measurements were retrieved during spring and neap tidal conditions from February 22 to 23,2012(spring tide)and from February 29 to March 1,2012(neap tide)at 14 mooring stations and six tidal gauges in Tongzhou Bay(Fig.1).

The quantitative validation was conducted through two statistical parameters.The first parameter was the Nash-Sutcliffe model efficiency coefficient(NSE)(Nash and Sutcliffe,1970),which is as follows:

where m is the measured value,p is the value predicted or calculated by the model,is the mean measured value,and Δm is the error in the measurements(van Rijn et al.,2003).The performance of the model was classified into the following classes:excellent(NSE≥0.65),very good(0.5≤NSE＜0.65), good (0.2≤NSE＜0.5), and poor(NSE＜0.2)(Allen et al.,2007;Henriksen et al.,2003).

The second parameter was the bias expressed in percentages(PB),which is as follows:

where the numerator can either be （m-p）-Δm(in the case of m-p＞0)or（m-p）+Δm(in the case of m-p＜0).The value of （m-p）±Δm is set to 0,if （m-p）±Δm＜0.The absolute values of PB were classified into four classes:excellent(|PB|≤10%),very good(10%＜|PB|≤20%),good(20% ＜|PB|≤40%),and poor(|PB|＞40%)(Allen et al.,2007;Henriksen et al.,2003).

3.Results and discussion

3.1.Model performance

The statistical validation of the calculated water levels for both NSE and PB at the six tidal gauges(with M1 through M6 representing Sanshahong,Xiaomiaohong,Lengjiasha,Lu¨si,Datang,and Yangkou stations,respectively)revealed an excellent performance of the model in predicting the water level,with an efficiency of NSE＞0.65 for all stations(Fig.3).Similar results were obtained for the percentage bias,scoring good for the Xiaomiaohong and Datang stations and very good for all the other stations(20%＜|PB|＜40% and|PB|＜20%,respectively).

The same statistical validation was performed for the current velocity and direction at the 14 mooring stations.Fig.4 shows the NSE values for the current velocity and current direction.The model performed well in predicting the current velocities at most stations with an excellent score(NSE＞0.65).The current velocity at Station N6(not shown in Fig.4)diverged from this trend and showed a considerably poor performance(NSE＜0.2),which demonstrates that the observed mean is better than the model prediction.At this station,significantly low velocities were measured during ebb tides(with the peak flow less than 0.3 m/s),which is presumably due to inconsistencies in bathymetry or error in measurements.Stations N10,N12,and N13 showed a good performance(0.2＜NSE＜0.5).This is mainly due to the fact that these stations were situated in relatively shallow water and the velocity was less bidirectional compared to that at the other stations.The performance of the model in calculating the current direction ranges from very good to excellent(NSE＞0.5)for most stations.Compared with the model performance in current velocity,the performance in current direction is poorer,due to the relatively large errors,even with a small difference in calculated and measured in current phase.Overall the model showed excellent to very good scores(|PB|＜20%).As discussed earlier,the exception is the poor score in current velocities at the N6 station(|PB|=100%).These scores are similar to performance scores found by Yao(2016),signifying that the refinement did not cause less accuracy in predicting the hydrodynamics in Tongzhou Bay.

Fig.3.Nash-Sutcliffe model efficiency coefficient and percentage bias for calculated water level at six tidal gauges.

3.2.Ecotope map of Tongzhou Bay

The model output was subsequently converted into the ecotope map for Tongzhou Bay(Fig.5),based on the definitions listed in Table 1.The Yaosha shoal was mainly classified as littoral ecotope,ranging from low-range littoral zone at the outer edges to high-range littoral and supralittoral zones close to the current seawall.This indicates that during mean high water springs(MHWS),nearly the whole extent of the Tongzhou Bay mudflats is inundated.The Lengjiasha shoal and other nearshore regions are within the shallow sublittoral zone,indicating constant inundation.The deeper sections of the Xiaomiaohong and Lengjiasha channels are mostly in the deep sublittoral zone.The hydrodynamic conditions roughly coincide with the overall elevations of the region,where the Yaosha and Lengjiasha shoals show low hydrodynamic conditions.

The bathymetry or elevation is the most influential factor in ecotope classification,as it also governs the hydrodynamics and dryfall periods.The created ecotope map showed that the hydrodynamic classification followed the elevation quite closely.As the elevation changes rapidly from deep channels to shallow flats with steep edges,a similar shift can be observed between low and high hydrodynamic conditions.This suggests that the hydrodynamic classification and chosen class bound are rather robust.As the more dominant elevation changes diminish within the littoral zone,the subclassification based on the dryfall period becomes more sensitive to the set class bounds.Specific validation of these intertidal dryfall periods,such as specific field measurements,is therefore required.

The retrieved ecotope map was subsequently overlaid with the 50%,90%,and 95% home ranges of the two shorebird species(Fig.6),with the degree of overlap shown in Table 2.Both species mainly occurred in the mid-range(30%-40%)and low-range(18%-19%)littoral zones.This is in line with the general knowledge of habitat preferences of the two shorebird species.Both species specialize in foraging on benthic organisms in intertidal mudflats.GK is a shellfish specialist(Zhang et al.,2019),while BTG is a generalist(Duijns et al.,2013).With its long bill,BTG can forage not only on shellfish,but also on worms deep in the sediment.Moreover,BTG is characterized as a tide-follower,moving up and down the mudflat following the tideline(Piersma et al.,2017).Hence,we can conclude that the mid-range and lowrange littoral zones,which overlap with 49%-60% of their 90% home ranges,represent the most used habitat conditions for both BTG and GK.

3.3.Impactofreclamationactivitiesonecotopedistribution

Fig.4.Nash-Sutcliffe model efficiency coefficient and percentage bias for calculated current velocity and direction at 14 mooring stations.

Fig.5.Ecotope map of Tongzhou Bay as of 2012.

The current reclamation activities on Tongzhou Bay(Fig.5)cover approximately 35.9 km2,including the U-shaped basin at Yaosha,covering 12.12 km2,and are similar to the work described in Huang and Liu(2017).The proposed port development expands these existing reclamations and covers approximately 126 km2,resulting in a total reclaimed area of 161.9 km2to be completed by 2035.

The impact of the current and proposed reclamations on the ecosystem was further assessed by calculating the area of each ecotope lost due to the current and proposed reclamation configurations,as a fraction of the extent of the original ecotope map(Fig.5).Results are shown in Fig.7.The reclamation from 2014 to 2018 was carried out at the upper reaches of the tidal flats,which are classified as supralittoral,high-range littoral,and mid-range littoral zones in low hydrodynamic conditions(5%,15%,and 9%,respectively).The proposed reclamation for 2018-2035 expands within these ecotopes.The port expansion will occur in the high-range,mid-range,and low-range littoral zones(27%,39%,and 34%,respectively).The combined current and proposed reclamations would lead to loss of high-range(42%),mid-range(48%),and low-range(38%)littoral zones(low and high hydrodynamic conditions combined).This corresponds to 44%-45% loss of the most important ecotopes(mid-range and low-range littoral zones)for BTG and GK.

3.4.Future opportunities

The current assessment of the impact of the ongoing reclamation on ecotope occurrence and potential habitat gives more insight into the potential risk that is involved with large-scale reclamation in coastal habitats,and the vital ecosystem services they provide.However,this assessment tool can also be used to determine the input value during the design stage.The composition of an ecotope map can provide additional ecosystem-based information to govern the design of coastal interventions.

In order to increase the accuracy of the created ecotope map,it is recommended to increase the set of abiotic parameters and data sources for the ecotope classification and subsequent validation.Including distributional data of more species of different ecosystem functions and investigating how their spatial distributions change throughout the tidal cycle(e.g.,Chan et al.,2019a),will enhance the insight into the value of certain ecotopes and the vital functions they provide within the coastal ecosystem.

Fig.6.Ecotope map of Tongzhou Bay and home range of bar-tailed godwit and great knot on southern Jiangsu coast from April 2015 to September 2017(refer to Fig.5 for ecotope color).

Table 2Degree of overlap of ecotopes corresponding with 90% home range of bar-tailed godwit and great knot.

Fig.7.Percentage of lost ecotope area with respect to total extent of ecotope map due to current,proposed,and total reclamations.

Reclamation has a severe effect on local hydrodynamics and transport agents and therefore the occurrence of ecotopes.In the classification process,the water depth is the most hierarchical parameter.Subsequent ecotope development could be investigated through morphodynamic modeling,which should be the next step in assessing the impact of reclamation on intertidal habitat.

4.Conclusions

In this study,an ecotope map was constructed to assess the current and proposed reclamation activities on Tongzhou Bay.The following conclusions were drawn:

(1)Spatial abiotic data(salinity and bathymetry)and hydrodynamic model output(dryfall period and flow velocity)were used to classify a set of 11 ecotopes,as bases for the ecotope map of Tongzhou Bay for 2012.This map was validated with satellite-tracked spatial data on two main threatened shorebird species(BTG and GK)occurring in this area.This showed a strong correspondence with mid-and low-range littoral zones,suggesting that these ecotopes represent the preferred habitat conditions for these species.This was further confirmed with previous knowledge on habitat preferences of these two species.

(2)The reclamation activities of 2014-2018 and proposed reclamation for 2018-2035,largely take place at the nearshore shoals(e.g.,Yaosha and Lengjiasha).The reclamation activities of 2014-2018 were carried out at the center of Yaosha shoal,reclaiming supralittoral or high-range littoral,and mid-range littoral zones in low hydrodynamic conditions(5%,15%,and 9% of the retrieved ecotope map extent,respectively).The proposed reclamation for 2018-2035 will expand on the current reclamation and mostly cover the highrange,mid-range,and low-range littoral areas(27%,39%,and 34% of the current ecotope map extent,respectively).The total of the proposed and current reclamations would cover highrange littoral(42%),mid-range littoral(48%),and low-range littoral(38%)areas,corresponding to a 44%-45% loss of the most important ecotopes(mid-range and low-range littoral zones)for BTG and GK.

(3)The use of ecotope maps as a basis for the quantification of the environmental impact of marine infrastructure over time enables early inclusion of habitat requirements in the design process of coastal works.

Acknowledgements

The authors would like to acknowledge Prof.Chang-kuan Zhang,Dr.Zeng Zhou,and Dr.Ao Chu at Hohai University,who contributed importantly to the data analysis and model development.We thank many volunteers who helped in the expeditions to deploy satellite transmitters onto shorebirds in Australia,including Chris Hassell who organized these expeditions,and Lee Tibbitts who curated the satellite tracking data.The satellite tracking was supported by the Spinoza Premium 2014 awarded by the Netherlands Organization for Scientific Research(NWO)to Theunis Piersma,by the MAVA Foundation,Switzerland,with additional support from WWFNetherlands and BirdLife Netherlands.Ying-chi Chan was supported by the Ubbo Emmius Fund of the University of Groningen and by the Spinoza Premium 2014 to Theunis Piersma.Finally,the financial support by the Lamminga Foundation to Jos R.M.Muller is also greatly acknowledged.