In JVAdams/artiFISHal: A Pelagic Fish Community Simulator

#library(knitr) 
knitr::opts_chunk$set(echo=TRUE, fig.width=5.5, fig.height=5.5)

The following is an example using the functions in the artiFISHal package. See README.md for instructions on installing the package.

Start by loading the package.

library(artiFISHal)

Simulate a Population of Fish

Create a data frame with 18 columns in which each row describes a sub-population of fish to be placed in the artificial lake. The first 11 columns must be completely filled in (no missing values).

• G = character, a one-letter nickname for the group (e.g., fish species and life stage) used in plotting
• Z = numeric, mean length (in mm)
• ZE = numeric, error around mean length, expressed as SD/mean
• LWC1, LWC2 = numeric, length-weight regression coefficients, where wt = LWC1*len^LWC2, (wt in g, len in mm)
• LWCE = numeric, error around weight estimate, expressed as SD(estimate)/estimate
• TSC1, TSC2 = numeric, target strength and length relation coefficients, ts = TSC1 + TSC2*log10(len/10), (ts in db, len in mm)
• TSCE = numeric, error around target strength estimate, expressed as SD(estimate)/estimate
• PropN = numeric, approximate proportion of population that the row represents (automatically adjusted to ensure they sum to 1)

The last 8 columns may have some missing values. However, in each row, either water depth (WD and WDE) or distance to bottom (D2B and D2BE) must be filled in, but not both.

• E = numeric, mean easting (m)
• EE = numeric, error around easting, expressed as SD/mean
• N = numeric, mean northing (m)
• NE = numeric, error around northing, expressed as SD/mean
• WD = numeric, mean water depth (m)
• WDE = numeric, error around water depth, expressed as SD/mean
• D2B = numeric, mean distance to bottom (m)
• D2BE = numeric, error around distance to bottom, expressed as SD/mean

You may find that the easiest way to create this data frame is by entering the information in an external file, e.g., a comma delimited file. In that case, you would read in the data frame using the read.csv function. The data frame should be all character and numeric variables, so if you use read.csv, you should specify as.is=TRUE.

ExInputs <- read.csv("C:/temp/ExInputs.csv", as.is=TRUE)

For the purposes of this vignette, I will create the data frame using code.

ExInputs <- data.frame(
  Description = c("alewife small", "alewife small", "alewife large", 
    "alewife large", "alewife large", "alewife large", "rainbow smelt small", 
    "rainbow smelt small", "rainbow smelt large", "rainbow smelt large", 
    "bloater small", "bloater small", "bloater large", "bloater large"), 
  G = c("a", "a", "A", "A", "A", "A", "s", "s", "S", "S", "b", "b", "B", "B"), 
  Z = c(65, 65, 127, 127, 127, 127, 55, 55, 127, 127, 70, 70, 190, 190), 
  ZE = c(0.15, 0.15, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.3, 0.15, 0.15, 0.3, 
    0.3), 
  LWC1 = 1e-8*c(1400, 1400, 1400, 1400, 1400, 1400, 481, 481, 481, 481, 102, 
    102, 102, 102), 
  LWC2 = c(2.8638, 2.8638, 2.8638, 2.8638, 2.8638, 2.8638, 3.0331, 3.0331, 
    3.0331, 3.0331, 3.3812, 3.3812, 3.3812, 3.3812), 
  LWCE = c(0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 
    0.05, 0.05, 0.05), 
  TSC1 = c(-64.2, -64.2, -64.2, -64.2, -64.2, -64.2, -67.8, -67.8, -67.8, 
    -67.8, -70.88, -70.88, -70.88, -70.88), 
  TSC2 = c(20.5, 20.5, 20.5, 20.5, 20.5, 20.5, 19.9, 19.9, 19.9, 19.9, 25.54, 
    25.54, 25.54, 25.54), 
  TSCE = c(0.03, 0.03, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.08, 0.08, 0.03, 
    0.03, 0.04, 0.04), 
  PropN = c(0.3, 0.3, 0.035, 0.035, 0.015, 0.015, 0.0375, 0.0375, 0.0375, 
    0.0375, 0.0375, 0.0375, 0.0375, 0.0375), 
  E = 100*c(93, 170, 6, 289, 90, 179, 93, 170, 6, 289, 93, 170, 34, 233), 
  EE = c(0.7, 0.4, 2.5, 0.1, 0.9, 0.6, 0.7, 0.4, 2.2, 0.1, 0.7, 0.4, 0.3, 0.1), 
  N = 100*c(100, 100, 100, 100, 100, 100, 100, 100, 170, 170, 100, 100, 75, 75), 
  NE = c(1, 1, 1, 1, 1, 1, 1, 1, 0.1, 0.1, 1, 1, 0.5, 0.5), 
  WD = c(14, 14, 11, 11, 11, 11, 16, 16, 29, 29, 20, 20, NA, NA), 
  WDE = c(0.2, 0.2, 0.2, 0.2, 0.2, 0.2, 0.2, 0.2, 0.3, 0.3, 0.2, 0.2, NA, NA), 
  D2B = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 15, 15), 
  D2BE = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 0.3, 0.3))

This data frame contains information on six groups of fish in 14 rows of data. To get some idea of how this works, have a look at rows 3-6 which includes specifications for group A, alewife large. The sizes of the fish (Z, ZE, LWC1, LWC3, LWCE, TSC1, TSC2, and TSCE) being simulated are the same in all four rows. But, the locations of the fish are different: 3.5% are at easting 600 (with SD 2.5 * 600), 1.5% are at easting 9,000 (SD 0.9 * 9,000), 1.5% are at easting 17,900 (with SD 0.6 * 17,900), and 3.5% are at easting 28,900 (SD 0.1 * 28,900). All of the groups of fish are located using water depth (WD and WDE), except for group B, bloater large, which are located using distance to bottom (D2B and D2BE).

ExInputs

Now you can use the SimFish function to simulate the fish population based on these parameters.

myfish <- SimFish(LakeName="My Lake", LkWidth=30000, LkLength=20000, 
  BotDepMin=20, BotDepMax=100, FishParam=ExInputs, TotNFish=100000, Seed=545)

Look at the true total number and weight of each fish group in the population.

myfish$Truth

Look at the fish population that you just simulated. Each row represents a single individual fish. Columns describe the group, location, and size of the fish.

head(myfish$Fish)

Simulating a fish population that looks the way you intended may take some trial and error. If you specify PlotsPdf = TRUE in the SimFish() function, a pdf file is created with a series of graphs illustrating the fish population that you've simulated to aid in this trial and error process. Some examples are shown below.

.BMP) .BMP)

In the process of creating the population, SimFish will eliminate fish based on their size (length or weight less than zero or target strength outside the allowable range) or location (easting, northing, water depth, bottom depth, or distance to shore outside their allowable ranges). You will likely end up with fewer fish than you requested. But, if you end up with far fewer fish than expected, you should look at the proportion of fish excluded to troubleshoot where in the inputs the problem might lie.

myfish$PropExcluded

The output from SimFish also keeps the inputs that were supplied as arguments for use in later sampling and summarization.

myfish$LakeInfo
myfish$FishInfo
head(myfish$FishParam)

Sample the Fish Population

Once you have the population the way you want it, you can sample the population with a virtual acoustic and midwater trawl survey.

mysurv <- SampFish(SimPop=myfish, NumEvents=10, AcNum=15, AcInterval=30000, 
  AcLayer=10, AcAngle=7, MtNum=30, MtHt=10, MtWd=10, MtLen=2000, Seed=341)

in addition to a few graphs summarizing the catch. The output from SampFish has two sample data frames, one for the acoustic targets detected and one for the midwater trawl catch.

head(mysurv$Targets)
head(mysurv$MtCatch)

The output from SampFish also keeps the inputs supplied as arguments for use in later sampling and summarization.

mysurv$SurvParam

Generate Estimates from the Sample

Estimate the number and biomass of fish per unit area sampled using the AcMtEst function which combines the acoustic and midwater trawl data.

perf <- AcMtEst(SimPop=myfish, AcMtSurv=mysurv, Seed=927)
head(perf)

The virtual acoustic and midwater trawl surveys carried out by SampFish are assumed to be perfect with no issues of fish availability, acoustic dead zones, or trawl selectivity. If you wish to introduce availability or selectivity functions, you may do so in the AcMtEst function.

For the target data, fish are excluded from the sample if they are located in zones unavailable to the acoustic transducer specified with the argument AcExcl.

For the catch data, fish are excluded from the sample if they are located in zones unavailable to the midwater trawl specified with the argument MtExcl. In addition, selectivity curves can be specified for different panels of the midwater trawl using the arguments PanelProps and SelecParam. The PanelProps argument is used to denote the relative sizes of four different panels in the trawl, from the mouth (outermost panel) to the cod end (innermost panel). You can get a fish's-eye view of the panel sizes denoted using the ViewZones function.

ViewZones(PanelProp = c(0.4, 0.3, 0.2, 0.1))

.PNG)

The SelecParam argument is used to denote the mesh selectivity for each fish group (species, life stage) and trawl panel zone. You can use the MeshPass function to determine the largest height (or depth) of a fish that can pass through a single diamond-shaped mesh of a net, provided the size of the mesh BarMesh, the height to width ratio of the mesh H2WRatio, and the length to height ratio of the fish L2HRatio.

MeshPass(BarMesh=2, H2WRatio=0.3, L2HRatio=4)

A double logistic regression function is used to specify selectivity as a function of fish total length using two slopes and two inflection points. If you have some notion of what the selectivity curve should look like, you can use the TuneSelec function to discover the corresponding parameters using interactive sliders.

TuneSelec()

.PNG) .PNG)

Once you have chosen the selectivity parameters you want to use, put them in a data frame.

selec <- data.frame(
  G = c("A", "a", "A", "a", "A", "a"),
  Zone = c("mouth", "mouth", "middle", "middle", "aft", "aft"),
  MtL50Small = c(100, 90, 60, 50, 30, 2),
  MtSlopeSmall = c(40, 40, 30, 30, 20, 20),
  MtL50Large = c(180, 180, Inf, Inf, Inf, Inf),
  MtSlopeLarge = c(20, 20, 100, 100, 100, 100))
selec

Then you can visually compare the selectivity curves for the different fish groups using the ViewSelec function.

ViewSelec(selec)

If all looks well, you can now derive estimates from the survey, incorporating these availabilities and selectivities.

imperf <- AcMtEst(SimPop=myfish, AcMtSurv=mysurv, AcExcl=c(5, 10), 
  MtExcl=c(2, 2), SelecParam=selec, Seed=204)
head(imperf)

JVAdams/artiFISHal documentation built on May 7, 2019, 10:14 a.m.