***** data in support of the paper: Subsidence reveals potential impacts of future sea level rise on inhabited mangrove coasts *****

authors: Celine.E.J. van Bijsterveldt*, Peter M.J. Herman, Bregje K. van Wesenbeeck, Sri Ramadhani, Tom S. Heuts, Corinne van Starrenburg, Silke A.J.Tas, Annisa Triyanti, Muhammad Helmi, Femke H. Tonneijck, Tjeerd J. Bouma

*corresponding author: Celine van Bijsterveldt

Contact information: 

celine.vanbijsterveldt@wur.nl

Aquatic Ecology and Water Quality Management (AEW)
Wageningen University & Research 
P.O. Box 47, 6700 AA Wageningen
The Netherlands

***** general introduction *****

To shed light on the future of low-lying rural areas in the face of sea level rise, we studied a 20 km long rural coastline neighbouring a sinking city in Indonesia (8 – 20 cm yr-1), 
hereafter called studyarea. Through the collection of data across 7 main topics, we show that villages experienced significant RSLR near the city. 
Mangroves also experienced RSLR near the city, although to a lesser degree, and were able to respond to RSLR rates 4.3 cm yr-1 through various root adaptations. 
The seven main investigated topics, and their respective datasets: 

0. 	village population migration
1.	village house experienced RSLR
2.	mangrove experienced RSLR
		2a. mangrove experienced RSLR
		2b. rainfall data
3.	foreshore dynamics
4.	mangrove bed-level dynamics based on pneumatophore markings
5.	mangrove root acclimation
		5.1. pneumatophore markings  (same dataset as 4.)
		5.2. rootmat formation
		5.3. sedimentation experiment
6.	lateral mangrove die-back
		6.1. dead trees
		6.2. pneumatophore mortality (same dataset as 4.)

******


0. village population migration

Response: migration flux (percentage of village population), based on census data (obtained from: https://demakkab.bps.go.id/publication.html)
Sample size: entire population of all villages within study area 

description:
The datafile (.csv) and shapefile (.shp) in this folder were used to show the migration flux and population densities of the villages in the coastal area of Demak. 
We used the village administrative boundaries and census data for 2009, 2010, 2014 and 2019 to create the shapefile. 
Census data were obtained from the website of the central bureau of statistics of Demak Regency, available at: https://demakkab.bps.go.id/publication.html. 
Through here, publications of census data per subdistrict of interest are available (our villages of interest lie withing the subdistricts of Bonang, Karang Tengah, Sayung and Wedung) 
Census data were joined with an existing shapefile of central Java's village administrative boundaries to create the shapefile in this folder.
The variablenames of the shapefile used to create Figure 3 in the Nature Sustainability paper are: 
- d_mig_fl_1: which is the migration flux in the villages in 2010. We chose this date as it is the year following a major erosion event in 2009, which was identified using the historic image slider in Google Earth Pro.
- population density per hectare: create new variable by: d_popula_1 / surface area in hectares (Desa_1)

The attribute table columns of the shapefile link to the columns in the csv file in the following manner:

columname in cvs	columname in attributetable .shp	meaning

Column1			d_field_1				field identifier
KODE			KODE					ID
DESA			DESA_1					village
KECAMATAN		d_KECAMATA				subdistrict
KABUPATEN		d_KABUPATEN				regency
PROPINSI		d_PROPINSI				province
KODE_KAB		d_KODE_KAB				code-regency
KABUPATEN_		d_KABUPA_1				subdistrict
KODE_PROP		d_KODE_PRO				code-province
PROPINSI_1		d_PROPIN_1				province
dfInside s		d_dfInside				filter(0/0.5/1) 1 = village inside aoi for this study (house data were collected), 0.5 & 0 = village outside aoi for this study) --> all these villages (0, 0.5 and 1) were used for visualation of the population fluxes in Figure 3 of the paper though 
X			d_X					longtitude coordinate centroid village EPSG:32749- WGS 84 / UTM zone 49S
Y			d_Y					latitude coordinate centroid village EPSG:32749- WGS 84 / UTM zone 49S
population.2009		d_populati				village population in 2009
population.2010		d_popula_1				village population in 2010
population.2014		d_popula_2				village population in 2014
population.2019		d_popula_3				village population in 2019
births.2009		d_births.2				births 2009
births.2010		d_births_1				births 2010
births.2014		d_births_2				births 2014
births.2019		d_births_3				births 2019
deaths.2009		d_deaths.2				deaths 2009
deaths.2010		d_deaths_1				deaths 2010
deaths.2014		d_deaths_2				deaths 2014
deaths.2019		d_deaths_3				deaths 2019
arrived.2009		d_arrive.				moved to village 2009
arrived.2010		d_arrive_1				moved to village 2010
arrived.2014		d_arrive_2				moved to village 2014
arrived.2019		d_arrive_3				moved to village 2019
left.2009		d_left.200				moved from village 2009
left.2010		d_left.201				moved from village 2010
left.2014		d_left.2_1				moved from village 2014
left.2019		d_left.2_2				moved from village 2019
mig_flux.2009		d_mig_flux				calculated migration flux 2009
mig_flux.2010		d_mig_fl_1				calculated migration flux 2010
mig_flux.2014		d_mig_fl_2				calculated migration flux 2014
mig_flux.2019		d_mig_fl_3				calculated migration flux 2019
migperc.2009		d_migperc.				migration percentage 2009
migperc.2010		d_migper.1				migration percentage 2010
migperc.2014		d_migper.2				migration percentage 2014
migperc.2019		d_migper.3				migration percentage 2019


******

1. village house experienced RSLR:

response: experienced RSLR (cm / year) by household
factor: distance from subsidence epicentre (km)
sample size: 194 houses across 14 villages spread across the studyarea

description: 
collection procedure: semi-structured interviews
Who collected: distributed by a group of students and local volunteers who spoke the local language, coordinated remotely (due to covid) by Sri Ramadhani, co-author on this paper.

column name						explanation

ID							village ID
NO							house identifier in village
Village							village name
seafront_village (yes/no)				filter used to selected the village houses to compare experienced sea level rise on land (behind the mangroves) with the experienced SLR of the mangroves
Distance_gradient					distance in kilometers from the subsiding harbour of Semarang
House Built/start living in house			Construction year of the house, or year that resident moved into the house
Number of floor increases since construction (times)	speaks for itself :)
The First Time Floor Raised				The First Time Floor Raised since the house was constructed or the person started living in the house
The Total Floor Construction (m)			The total amount that the floor was raised since house construction/person started living in the house
Gutter height survey year (m)				Gutter height during the survey in 2020 (m)
Gutter height at construction (m)			Estimated gutter height as the person remembers from house construction/moment started living in the house
eRSLR based on gutter					experienced relative sea level rise (m/year) based on the difference in gutter heights divided by time elapsed between construction/start living in house
comm_eRSLR						comments on gutter eRSLR calculations
eRSLR based on floor 					experienced relative sea level rise (m/year) based on the total amount of floor raised divided by time elapsed since construction/start living in house
comment_eRSRL based on floor				comments on floor eRSLR calculations

******

2. mangrove experienced RSLR

2a mangrove experienced RSLR

Collection procedure: Small pressure sensors (Onset HOBO Water level logger U20L-04), which were covered with a sock (to prevent theft), were tied to the trunks of eight two mangrove trees in the eight monitored mangrove stands located along 20 km of study area. The loggers were configured to measure the pressure every 15 minutes, and were deployed for a period of 2.25 years. All sensors were located well above MSL, ranging from 22 until 27 cm above MSL. The sensors were cleaned and redeployed repeatedly. Whenever a sensor-tree was lost by a storm, a new sensor was deployed on a tree further inside the mangrove forest, anticipating future storms.
Who collected: The first author

response: RSLR experienced by mangroves (cm/year)
factor: distance from subsidence epicentre (km)
sample size: 5 trees spread across the 20 km studyarea
Each tree represents a continuous waterpressure dataset of which more than 15 days overlapped in the dry season between years to calculate mean (+/- 95% CI) water level increase at the mangrove tree stem.

For pre-processing of the raw pressure data collected via this method. Please see the matlab script: "Supporting_script_subsidence_anlaysis.m" in the folder 2.Mangrove experienced RSLR preprocessing (raw data and code).
For statistics on differences in water level for each site resulting in an average experienced RSLR rate +/- 95 CI, see "subsidence_stats_repos_final.R" which also includes the names for relevant parameters from all preprocessed pressure files: "Comp_sitename_window.csv" files
The experienced RSLR values obtained with this analysis are combined in a datafile with the house experienced RSLR data called: 2. combined_experiencedRSLR_mangroves_villages.csv

2b. rainfall data

The potential effect of rainfall on the experienced RSLR in mangroves was verified by comparing average daily rainfall between the relavant time windows in the two years of interest.
rainfall data for the timeframe of interest were combined into one dataset: 2b_rainfall_data_aug2017-nov2019.csv, which is part of the folder: 2.Mangrove experienced RSLR and rainfall
Who collected: Data were publically available on: https://dataonline.bmkg.go.id/ . Data were collected by: Badan Meteorologi, Klimatologi dan Geofisika (BMKG)

Response: daily rainfall (mm)
Factor: 2018 vs 2019
Sample size: 182 days in 2018, and 182 days in 2019

Notes:
8888: Unmeasured Data
9999: No Data
Tn: Minimum temperature (°C)
Tx: Maximum temperature (°C)
Tavg: Average temperature (°C)
RH_avg: Average humidity (%)
RR: Rainfall (mm)
ss: The duration of solar radiation (h)
ff_x: Maximum wind speed (m/s)
ddd_x: Wind direction at maximum speed (°)
ff_avg: Average wind speed (m/s)
ddd_car: Wind direction most (°)


******

3. foreshore dynamics

Each mangrove site that was monitored for RSLR also had one PVC sediment monitoring pole on the foreshore at approximately 50 m seaward the mangrove fringe. 
Poles were 10 cm in diameter, 2 m long, marked at every 10 cm with tape, and driven 1 m deep into the sediment.

response: bed-level change (cm/month) at 50 m seaward from mangrove edge
response: foreshore depth
factor: experienced RSLR (measured in mangroves)
sample size: 8 sites across the studyarea, monitored 3 times in 2 years' time
who collected: Celine van Bijsterveldt

column name			explanation

empty				row number
afstand van semarang		distance from Semarang harbour (km)
site name			site name (derived from landward village name)
t 				date monitored
depth				depth at pole compared to MSL (cm)
accretion/month			accretion (cm) measered at the pole between t and t-1 and averaged out in accretion per month
PoleID				pole name specific per site. at some locations the PVC pole that had originally been placed had been removed by someone but georeference poles that we placed were still there. In those locations we used the georef poles, which we had originally placed for drone mapping of the foreshore, for depth monitoring, This is why some poles are called "georef".			
subsidence			mangrove experienced RSLR derived from dataset 2, this column was added to this dataset for the purpose of analysis


4. mangrove bed-level dynamics based on pneumatophore markings

To monitor morphological responses of mangrove trees to the RSLR they experienced, each tree that received a water level logger, 
also received markings on ten pneumatophores to quantify responses in root-growth and bed-level dynamics. 
To this end, a cable-tie was tied to each pneumatophore at 10 cm from the tip, 
and at 10 cm from the bed. We randomly picked 3 other trees per site that received a similar treatment. 

response: bed-level change (cm/month)
factor: foreshore depth
sample size: Each of the 8 sites across the study area had 4 marked trees that had 10 pneumatophores marked at 10 cm from the bed (and 10 cm from the tip for pneu growth (see 5.). 
Alls 8 sites and trees were monitored 3 times over a period of 1.5 years.
who collected: Celine van Bijsterveldt, pre-processing by Silke Tas (co-author)

column name				description

t					timepoint in months compared to baseline	
site					site name, derived from the village name landward
km from Semarang			distance from the site to the harbour of Semarang, assumed to be the "subsidence epicentre" in this study
tree #					treeID per site
subsidence				mangrove experienced RSLR derived from dataset 2, this column was added to this dataset for the purpose of analysis
mean_foreshoredepth			mean foreshore depth over the 1.5 years of monitoring per site --> derived from dataset 3
foreshoredepth_at_tn			foreshore depth at timepoint that the pneumatophores were measured. 
pneu-tie under-bottom t0 (March 2018)	distance from cable-tie to forestfloor at t0 (this is always 10 cm)
pneu-tie upper-top t0 (March 2018)	distance from cable-tie to tip of the pneumathophore at t0 (this is always 10 cm)
pneu-tie under-bottom tn		measured distance (cm) from lower cable tie to forest floor at timepoint of monitoring. 
pneu-tie upper-top tn			measured distance (cm) from upper cable tie to pneutophore tip at timepoint of monitoring. 
totel length pneu tn			total pneumatophore length at timepoint of monitoring.
bedlevel to baseline (cm)		difference between lower cable-tie to bedlevel at tn and distance from bed-to-cabletie at t0
pneu growth (cm)_tn			difference between upper cable-tie to tip at tn and distance from tip-to-cabletie at t0
pneu growth rate (cm/month)		pneu growth (cm)_tn divided by the time elapsed since baseline (months), thus average pneugrowth rate (cm/month). Note: pneu growth was calculated from t0 to tn only (not between t and t+1) because pneus didn't have IDs so we couldn't calculate growth between t-1 and t.
bedlevel change rate (cm/month)		bedlevel to basline (cm) divided by the time elapsed since baseline (months), thus average bedlevel change rate (cm/month)
mortality				pneumatophore still alive: 0, pneumatophore dead at timepoint tn: 1

5. mangrove root acclimation


5.1: pneumatophore markings (same dataset as 4)

response: pneumatophore extension (cm/month)
factor: experienced RSLR experienced by mangrove trees
sample size: Each of the 8 sites accross the study area had 4 marked trees that had 10 pneumatophores marked at 10 cm from the tip (and 10 cm from the bed (see 4.). Alls 8 sites and trees were monitored 3 times over a period of 1.5 years. 
who collected: Celine van Bijsterveldt

column names and descriptions: see dataset 4.


5.2: rootmat formation:

To investigate how mangrove trees had responded to ongoing experienced RSLR over the past decades, we rinsed out the root-zones of three living mangrove trees with a 60 m^3 / h motorized water pump at three key sites. 
At each site, the number of distinctly separate rootmats per tree was quantified in the top 60 cm of the sediment, deeper excavation and rinsing was not possible. 

response: number of rootmats
factor: experienced RSLR experienced by mangrove trees
sample size: 9: 3 trees per site. 3 sites distributed across the study area, including the outer edges of the studyarea
Who collected: co-author: Corinne van Starrenburg

column name			description

site				site name, derived from landward village name					
distance from Semarang		distance from the site to the harbour of Semarang, assumed to be the "subsidence epicentre" in this study
forest_accretion (cm/year)	accretion rates per year derived from the pneumatophore markings (dataset 4) 
subsidence			mangrove experienced RSLR derived from dataset 2, this column was added to this dataset for the purpose of analysis, the subsidence rate for Bedono island was interpolated by regression using dataset 2. combined_experiencedRSLR_mangroves_villages.csv 
DBH				Diameter at breastheight (cm) of the tree of interest
species				Avicennia species 
rootmats			number of distinctly seperated rootmats on the belowground stem.

5.3: sedimentation experiment:

The effects of extremely high sedimentation pulses were simulated in the field by exposing saplings (height = 60 cm, n = 6 per treatment group) and young trees (height = 2 m, n = 4 per treatment group) 
to a sediment increase of either 20 or 40 cm, with a control group where no sediment was added. 
A PVC tube (diameter = 30 cm) was put over the seedlings and filled with locally available sediment. 
The young trees were surrounded by fencing, constructed from bamboo and plastic and filled with sediment. 
Survival of saplings and trees in the experiment was assessed after 22 days at the end of that field campaign, survival of trees was assessed again at the start of the next field visit (60 days later).
Who collected: co-author: Tom Heuts

response: young mangrove tree survival
response: mangrove sapling survival
factor: sedimentation (3 treatments: 0 cm, 20 cm and 40 cm of sudden sedimentation)
sample size: saplings 6 per treatment, young trees: 4 per treatment

column name			description

ID				sapling and tree ID
age				age group (sapling or young tree)
Treatment			sedimentation treatment (cm) (0, 20 or 40 cm)
treatment_name			sedimentation amount - age group
survived			individual survived after 22 rep. 60 days: TRUE, individual died: FALSE

6. lateral mangrove die-back  

6.1: dead trees

Lateral mangrove erosion was quantified by counting the number of dead trees seaward of the mangrove fringe per 50 m coastline stretch at each of the 8 sites along the coast in the study area. 
collection procedure: Each site was approached by boat at low tide so that the trees with root zone were submerged. To get a general overview of the state of the mangrove fringe at the respective location, 
the number of dead trees within a 50 m stretch was counted alongshore, starting several meters from the boat. Thereby a 50 m measure tape was fixed (either on a tree or held), 
while one person walked along the coast counting the number of (visual) dead trees until the tape would reach its end. Fallen as well as standing trees were counted. 
Who collected: co-author Corinne van Starrenburg

response: number of dead trees per 50 m coastline stretch
factor: foreshore depth
sample size: 6 sites across the studyarea, including the outer edges of the study area

csv file: 6.summary_all_data_incl_dead mangrove counts per coastline strecht.csv --> this file was also used to create the graphs ín Figure 4 of the paper with relevant parameters and 95% CIs along the distance gradient from Semarang city.

column name												description

afstand van Semarang											distance from the site to the harbour of Semarang (km), assumed to be the "subsidence epicentre" in this study
site name												site name, derived from landward village name		
15earlysubsidence (cm/year)										mangrove experienced RSLR (cm/year) based on 15 day period in the early dry season (derived from dataset 2)
15earlysubtime_h											mangrove experience RSLR in increase of daily inundation time (h) based on 15 day period in the early dry season (derived from dataset 2)		
60earlysubsidence (cm/year)										mangrove experienced RSLR (cm/year) based on 60 day period in the early dry season (derived from dataset 2)
60earlysubtime_h											mangrove experience RSLR in increase of daily inundation time (h) based on 60 day period in the early dry season (derived from dataset 2)
60midsubsidence (cm/year)										mangrove experienced RSLR (cm/year) based on 60 day period in the mid dry season (derived from dataset 2)
60midsubtime_h												mangrove experience RSLR in increase of daily inundation time (h) based on 60 day period in the mid dry season (derived from dataset 2)
60latesubsidence (cm/year)										mangrove experienced RSLR (cm/year) based on 60 day period in the late dry season (derived from dataset 2)	
60latesubtime_h												mangrove experience RSLR in increase of daily inundation time (h) based on 60 day period in the late dry season (derived from dataset 2)
subsidence (cm/year)											average mangrove experienced RSLR (cm/year) based on 60 day period (and 15 day period for Bedono bay) displayed in figure 4
subsidence_95%CI_errorbar										95% confidence interval for excel errorbar with previous column (see dataset 2 for 95% CI for experienced RSLR calculations and stats)
n													timewindow overlap in the early dry season between year 1 and year 2 that experienced RSLR could be calculated from.
loglin_subfit_man											fitted values from logistic regression on mangrove experienced RSLR data and distance from Semarang (see stats for dataset 2), added to this dataset for trendline plotting
submergence time increase h/year									average mangrove experience RSLR in increase of daily inundation time (h) based on 60 day period (and 15 day period for Bedono bay)
subtime_95%CI_errorbar											95% confidence interval for excel errorbar with previous column (see dataset 2 for 95% CI for experienced RSLR calculations and stats)
no of dead trees											number of dead trees per 50 m coastline stretch
mean net accretion (cm/year)  (all red data averaged (april 2018 until october 2019))			mean net accretion of the forest floor in (cm/year) based on dataset 4
n_net accretion (cm/year)  (all red data averaged (april 2018 until october 2019))			number of pneumatophores that survived the 1.5 year period
95%CI_net accretion (cm/year)  (all red data averaged (april 2018 until october 2019))			95% confidence interval for excel errorbar with previous column (based on dataset 4)
mean_pneu_growth (cm/year) (all red data averaged (april 2018 until october 2019))			mean net pneumatophore growth in (cm/year) based on dataset 4			
n_pneu_growth (cm/year) (all red data averaged (april 2018 until october 2019))				number of pneumatophores that survived the 1.5 year period
95%CI_pneu_growth (cm/year) (all red data averaged (april 2018 until october 2019))			95% confidence interval for excel errorbar with previous column (based on dataset 4)
n_living trees												number of living trees excavated per site for rootmat assessement
avg_rootmat_number_living										average number of rootmats per site
95%CI_rootmat_number_living										95% confidence interval for excel errorbar with previous column
mean pneu mortality t19 %										mean percentage of dead monitored pneumatophores per tree after 1.5 years monitoring period
n pneu survival t0											number of monitored trees
95%CI pneu survival t19											95% confidence interval for excel errorbar with previous column (based on dataset 4)
sed pole name_baseline											Sediment pole name at original placement
sed pole name_end											Sediment pole name after monitoring campaign, if name is different, it had to be replaced. This is reflected in the number of observations at each site (2 columns later)
mean_foreshoredepth_during_campaign									mean bed level depth relative to MSL (cm) over the 1.5 year monitoring campaign
n_foreshoredepth											number of times the depth at each pole could be determined
95%CI_foreshoredepth											95% confidence interval for excel errorbar with previous column (based on dataset 3)
sed pole name_baseline											Sediment pole name at original placement
sed pole name_end											Sediment pole name after monitoring campaign, if name is different, it had to be replaced. This is reflected in the number of observations at each site (2 columns later)
mean_bedlevel_change_(cm/year) (all_data)								mean bed level change (cm/year) averaged over the times the pole was monitred over the 1.5 year monitoring campaign
n_bedlevelchangerate (all data)										number of times the bed level change at a pole could be monitored
95%CI													95% confidence interval for excel errorbar with previous column (based on dataset 3)
village_name												village name (based on adm_boundaries)
distance_from_semarang											distance from village centroid to the harbour of Semarang
mean subsidence cm											mean village experienced RSLR (cm/year) based on the houses that fell within the village administrative boundaries for the villages that lined the coastline directly (seafront village in dataset 1)
n													number of houses per village from which experienced RSLR was calculated
95%CI													95% confidence interval for excel errorbar with previous column (based on dataset 1)
loglin_subfit_vil											fitted values from logistic regression on house experienced RSLR data and distance from Semarang (see stats for dataset 2), added to this dataset for trendline plotting



6.2: pneumatophore mortality

In support of the number of dead trees counted along 50 m coastline stretches, the loss of monitored pneumatophores gave a clear indication of lateral erosion as well.
Marked pneumatophores (sample described in (4. Mangrove bed-level dynamics)) were often distributed around the tree, so 50% mortality of marked pneumatophores was often a result of lateral erosion in the field, 
which left the monitored tree at the seaward edge of the mangrove forest. Pneumatophore survival per site was monitored two times in 1.5 years.

response: proportion of marked pneumatohpores dead after 19 months of monitoring (same dataset as 4)
factor: foreshore depth
sample size: Each of the 8 sites accross the study area had 4 marked trees that had 10 pneumatophores, which were monitored for 19 months
Who collected: Celine van Bijsterveldt




