A. Palermino, G. Coro, P. Bove, E.N. Armelloni, G. Scarcella, 2023
This repo contains the workflow developed to fill trawl survey haul gaps based on the paper: “Coro G, Bove P, Armelloni EN, Masnadi F, Scanu M and Scarcella G (2022) Filling Gaps in Trawl Surveys at Sea through Spatiotemporal and Environmental Modelling. Front. Mar. Sci. 9:919339. doi: 10.3389/fmars.2022.919339”.
The tool take into account the spatial mobility of the species through BIMAC tool (how far can a population go in one year and which is the mobility of the species?), the historical trends through SSA tool (what biomass do we expect in the missed hauls given the observation history?), the environmental aspect through MaxEnt tool (how inter-annual bio-geo-chemical change affect the species distribution in the survey year?). Finally through the HBIE tool,the results of the aforementioned tools are weighted and gather to obtain the final biomass index of missing hauls. In the present repo DIVA tool described in Coro et al. (2022) has been replaced by BIMAC (Coro 2024; https://doi.org/10.1016/j.envsoft.2023.105901).
- Download folder “R”, “data” and “java” and store them in a unique folder.
- Solea_solea.csv; HaulData.csv; StrataWeight.csv: example of input data
- gebco_30sec_8.asc: world depth data needed for BIMAC computation. Do not remove, move or change this file.
- Subfolder “Envirnmental inputs”: contains environmental variables of interest for MaxEnt tool. Data provided refers to the Adriatic Sea from 2019 to 2022
To apply “Survey_gap” to your own case study you need to provide the following input files:
This file contains the available information regarding your target species index (e.eg. abundance and biomass) for your study area. The file needs to have the following structure:
## station lat lon year species biomindex abunindex
## 1 1 44.55919 12.33923 2005 SOLEVUL 307.58783 6807.9443
## 2 2 44.58757 12.52665 2005 SOLEVUL 62.66961 716.2242
## 3 7 44.28683 12.38482 2005 SOLEVUL 55.14626 1246.4564
## 4 8 44.33746 12.50063 2005 SOLEVUL 94.46218 1208.7578
## 5 13 44.11446 12.54723 2005 SOLEVUL 54.91112 1402.5491
## 6 14 44.12538 12.70341 2005 SOLEVUL 84.19532 996.2042
where:
- station: identifier of the survey station. Needs to be consistent over the years, but admits gaps.
- lat: Latitude of the station in decimal degree
- lon: Longitude of the station in decimal degree
- year: year of the survey during which the station was carried out
- species: name of the species in the preferred format. Needs to be consistent over the years
- biomindex: biomass index of the station for the interested species
- abunindex: abundance index of the station for the interested species
This file contains a list of missing hauls that have to be computed with relative information. The file needs to have the following structure:
## station year lat lon
## 1 4 2022 45.09887 12.38338
## 2 5 2022 45.15225 12.50874
## 3 6 2022 45.25585 12.87548
## 4 10 2020 45.09946 13.25025
## 5 10 2022 45.09946 13.25025
## 6 14 2022 44.73999 13.08566
where:
- station: identifier of the survey station. Needs to be consistent over the years, but admits gaps.
- year: year of the survey during which the station was not carried out.
- lat: Latitude of the station in decimal degree
- lon: Longitude of the station in decimal degree
This file contains the information on the surveyed area. In case of missing haul use the swept area of the previous year:
## Station Stratum year SweptArea
## 1 1 STR1_17 2019 0.03599334
## 2 1 STR1_17 2020 0.02437610
## 3 1 STR1_17 2021 0.02349296
## 4 1 STR1_17 2022 0.02925132
## 5 10 STR2_17 2019 0.03656774
## 6 10 STR2_17 2020 0.03656774
where:
- Station: identifier of the survey station. Needs to be consistent over the years and includes all the stations (missing and not missing stations)
- Stratum: identifier of the stratum used for the computation of total biomass index of the area.
- year: year of the survey during which the station was carried out or not. Need to include the interested year to be filled
- SweptArea: Swept area of the station
This file contains information on the surveyed area:
## Stratum Area StratumWeight
## 1 STR1_17 11361 0.268023969
## 2 STR2_17 8410 0.198405209
## 3 STR3_17 22466 0.530008493
## 4 STR1_SLO 151 0.003562329
where:
- Stratum: identifier of the stratum used for the computation of total biomass index of the area.
- Area: Total area covered by each stratum
- StratumWeight: Weight of the stratum on the total surveyd area
Download the raster layer file in asc format for the interested environmental variables:
## [1] NA NA NA NA NA NA NA
## [8] NA NA NA NA NA NA 0.4100096
## [15] 0.3727627 NA 0.5278887 0.4521504 0.3670816 0.2999293 0.6281053
## [22] 0.5507973 0.3928404 0.2796727 0.2383853
## class : RasterLayer
## dimensions : 39, 62, 2418 (nrow, ncol, ncell)
## resolution : 0.1, 0.1 (x, y)
## extent : 12.3, 18.5, 41.9, 45.8 (xmin, xmax, ymin, ymax)
## crs : +proj=longlat +datum=WGS84 +no_defs
## source : CHL_summer_2019.asc
## names : CHL_summer_2019
where:
- file “variable.asc”: a matrix containing the values of environmental variable for each cell on the map. 0 values have to be set = -9999 in asc format in order to obtain NA.
- dimensions: identifier of matrix size. nrow, ncol and ncell need to be consistent across all the environmental variables
- resolution: identifier the resolution of the map determining the number of cells
- extent: extent of the map in longitude and latitude. Need to be consistent across all the environmental variables
- Folder “data” containing: missing_hauls.csv, Genus_species.csv, HaulData.csv, StrataWeight.csv,
- Subfolder “Environmental_inputs” containing the asc files of environmental variables to be used for MaxEnt run each year, file “gebco_30sec_8.asc”: world depth data for BIMAC computations (DO NOT EDIT)
- Folder “R” containing: Code BIMAC_no_advection.R for BIMAC computations (DO NOT EDIT), code Workflow_Surveygaps.R, code Workflow_Surveygaps_Solemon.R with species and year to be filled for Solemon survey
- Folder “java” containing: Code max_ent_cyb.jar for MaxEnt computations (DO NOT EDIT), code ssa.jar for SSA computations (DO NOT EDIT), folder “cfg” containing: operators.xlm for ssa computations (DO NOT EDIT)
Open Workflow_Surveygaps.R in RStudio and set working directory to “source file location”. In Workflow_Surveygaps_Solemon.R, data are already selected for the unusable years (2005 and 2006) and hauls (“ms” and “bis”)
-
Line 5: feature selection = FALSE/TRUE:
# feature_selection=T- if FALSE all the environmental variables stored in the subfolder are used during MaxEnt computation.
- if TRUE only the environmental variables with a certain level of importance (set at the 95% interval of the highest importance level among environmental variables) after the first run are included in MaxEnt computation
-
line 6: generate a vector with the name of the species to compute.
## [1] "Solea_solea"
- Terms of the species have to be consistent with the file names in folder "data":
- line 7: generate vector with years.
## [1] 2019 2020 2021 2022
- Years have to be consistent with Environmental_inputs subfolders
- line 8: select index to compute (e.g. abunindex or biomindex)
## [1] "biomindex"
Run the code from “Source”
In this step several functions are applied to compute temporal, spatial and ecological tools. The results are stored for HBIE computation
- Inputs are stored in folders named “ssa_input”, “MaxEnt_input” and “BIMAC_input” by species and year
- Output are stored in folders named “ssa_output”, “MaxEnt_output” and “BIMAC_output” by species and year
In this step the temporal (SSA), spatial (BIMAC) and ecological (MaxEnt) results are pulled together and weighted to give the estimation of mean, low and high biomass index of missing hauls.
- Inputs are stored in folder named “hbie_input” by species and year
- Outputs are stored in folder named “hbie_output” by species and year. The output csv file contain the final biomass index results per station in the column biomass_est_mean.
In this step the total index of the surveyed area is computed through the stratified mean method described in Coro et al. (2022) from the whole hauls including the filled missing hauls.
- Inputs are stored in folder named “biom_index_input” by species and year
- Outputs are stored in folder named “biom_index_output” by species and year. The output csv file contain the final biomass index results per area
At the end of the run all inputs and outputs are stored in the following folder tree structure per species
The main results are reported in:
hbie_output
## station lon lat ecological spatial temporal spatialerror
## 1 30 13.95125 43.94125 0.05007590 40.00000 0.000000 15.737645
## 2 69 14.24375 42.67375 0.06256380 66.83372 0.000000 16.638121
## 3 70 14.24375 42.57625 0.08174660 77.70821 15.008414 21.537151
## 4 71 14.63375 42.28375 0.04707840 52.24412 1.212261 12.507442
## 5 72 15.02375 42.28375 0.00784261 47.52965 0.000000 9.081131
## 6 73 14.92625 42.08875 0.04970930 48.73597 23.956892 19.635370
## index_est_mean index_est_low index_est_high
## 1 20.00000 0.000000 43.60647
## 2 33.41686 8.459682 58.37404
## 3 46.35831 14.052583 78.66404
## 4 26.72819 7.967027 45.48935
## 5 9.50593 4.057252 14.95461
## 6 36.34643 6.893373 65.79948
where:
- ecological depicts the results of MaxEnt provided in details in MaxEnt folders
- spatial and spetail_error depict the results of BIMAC provided in details in BIMAC folders
- temporal depicts the results of SSA provided in details in SSA folders
- index_est_mean, index_est_low, index_est_high depict the results of HBIE
biomindex_output
## # A tibble: 4 × 2
## method biomindex
## <chr> <dbl>
## 1 biomindex_known 74.6
## 2 biomindex_mean 69.5
## 3 biomindex_low 65.2
## 4 biomindex_high 73.9
where:
- biomass_known depicts the biomass computation from the known hauls (excluding the missing hauls)
- biomass_mean, biomass_mean, biomass_mean depict the biomass computations from the whole dataset (known hauls + missing hauls)
At the end of each run the biomindex_outptut table is shown along with a plot showing the index estimated per missing station (from hbie_output) and Area ( from biomindex_output). Both are stored in “biomindex_output” folder
