Adult Index Estimation

The R package GLFC includes functions to estimate indices of adult sea lamprey abundance in the Great Lakes. The Great Lakes Fishery Commission made the decision to switch to a lake-wide index of adult abundance in December 2014 (GLFC-TTF 2014). Prior to that decision, lake-wide population estimates (PE) were generated following Mullett et al. (2003). This vignette walks you through the process of generating Adult Index estimates.

One overarching function, AIpresto can be used to carry out all the steps of the Adult Index estimation process in one fell swoop. This function prepares and error checks the data, estimates the Adult Index, calculates the targets, and generates a draft report.

Install

Install the latest version of R from CRAN.

Install the latest version of the GLFC package following the instructions at Readme. The easiest way to do this, if you already have the package devtools installed, is:

devtools::install_github("JVAdams/GLFC")

Prepare the data

Three csv files are needed to generate Adult Index estimates, one each with new stream data, previous stream data, and previous lake data.

(1) STREAMDATANEW should contain stream mark-recapture estimates of adult sea lampreys for the year of interest.

year	lake	lscode	trapcatch	PEmr	CVmr	comments
2015	1	1.032	1	NA	NA	no recaps
2015	2	2.005	5971	9121	6
2015	2	2.006	575	910	16
2015	2	2.015	619	1548	9
2015	2	2.024	1010	1316	12

The first row of STREAMDATANEW should be a header row with the following column names:

year - survey year
lake - lake code (1=Superior, 2=Michigan, 3=Huron, 4=Erie, and 5=Ontario)
lscode - lake-stream code, a combination of lake and stream codes e.g., lscode = 1.064 = lake code + (stream code)/1000
trapcatch = no. of adult sea lampreys captured
PEmr = stream population estimate of adult sea lampreys from mark-recapture study
CVmr = coefficient of variation of PEmr
comments = comments

where \[ CV_{mr} = 100\% \frac{\sqrt{Var(PE_{mr})}}{PE_{mr}} \]

(2) STREAMDATAPREV should contain stream mark-recapture estimates of adult sea lampreys for all of the years prior to the year of interest.

year	lake	lscode	PEmr	CVmr	indexContrib	indexContribCV
2000	1	1.026	NA	NA	NA	NA
2001	1	1.026	NA	NA	NA	NA
1989	1	1.028	NA	NA	NA	NA
1990	1	1.028	NA	NA	NA	NA
1991	1	1.028	NA	NA	NA	NA

The first row of STREAMDATAPREV should be a header row with the following column names:

year - survey year
lake - lake code
lscode - lake-stream code, a combination of lake and stream codes
PEmr = stream population estimate of adult sea lampreys from mark-recapture study
CVmr = coefficient of variation
indexContrib = the stream population estimate used in the Adult Index
indexContribCV = the stream CV used in the 95% confidence interval

(3) LAKEDATAPREV should contain annual lake-wide adult sea lamprey indices for all of the years prior to the year of interest.

year	lake	index	ilo	ihi
2010	5	14864	11176	18552
2011	5	24808	15232	34384
2012	5	25848	21379	30318
2013	5	9456	7532	11380
2014	5	10644	8435	12854

The first row of LAKEDATAPREV should be a header row with the following column names:

year - survey year
lake - lake code
index - estimated lake-wide index of adult sea lamprey abundance
ilo - lower 95% confidence interval
ihi - upper 95% confidence interval

For all 3 files,

the names of the columns must be exactly as described here (same spelling, same case)
the order of the columns is not important
additional columns may be included
there should be no missing values in year, lake, or lscode

All 3 files should all be stored in a single directory, DIRECTORY.

Assign the name of the directory (use double backslashes, \\, in the file path) and the names of the three files. For example,

Dir <- "C:\\temp"
StreamNew <- "AdultStream2015.csv"
StreamPrev <- "AdultStreamThru2014.csv"
LakePrev <- "AdultLakeThru2014.csv"

Run the works

Load the GLFC package to have access to the Adult Index functions.

library(GLFC)

Use the AIpresto() function to prepare and error check the data, estimate the Adult Index, calculate the targets, and generate a draft report. Simply provide the function with the name of the directory where the files are stored, the names of the new and previous stream data, and the name of the previous lake data.

AIpresto(DIRECTORY=Dir, NEWDATARAW=StreamNew, STREAMDATAPREV=StreamPrev, LAKEDATAPREV=LakePrev)

Behind the scenes

Within the AIpresto() function, the following steps are carried out.

Manipulate the data

The AIprep() function is used to read in the adult sea lamprey stream trapping data and prepare them for error checking and estimation.

streamall <- AIprep(csvDir=Dir, csvNew=StreamNew, csvOld=StreamPrev)

The AIprep() function brings in additional information built in to the GLFC package by default, including

trappedStreams a data frame with stream information,

lake	lscode	country	strcode	estr	strname	lat	long
1	1.060	Canada	60	4	Little Carp	46.50	-84.43
1	1.059	Canada	59	5	Big Carp	46.50	-84.45
1	1.072	Canada	72	23	Cranberry	46.70	-84.43
1	1.058	Canada	58	24	Goulais	46.71	-84.45
1	1.071	Canada	71	39	Harmony	46.85	-84.36

lsIndex a list of lake-stream codes identifying index streams,

lake	index.streams
Superior	1.001, 1.002, 1.009, 1.029, 1.031, 1.032, 1.062
Michigan	2.005, 2.006, 2.015, 2.024, 2.026, 2.035
Huron	3.01, 3.016, 3.027, 3.032, 3.999, 3.038
Erie	4.001, 4.002, 4.003, 4.007, 4.009
Ontario	5.005, 5.009, 5.022, 5.023, 5.036

lsKeep a list of lake-stream codes identifying streams which will continue to have ongoing trapping even if not part of the Adult Index.

lake	maintenance.streams
All	1.023, 1.059, 2.013, 2.068, 4.001, 5.013, 5.021

The returned object from the AIprep() function is a single data frame (called streamall in this example) with information from all of the inputs, including new variables:

index - a logical (TRUE/FALSE) identifying the index streams
maintain - a logical identifying the streams that will continue to have ongoing trapping even if not part of the Adult Index
indexContrib - the stream population estimate that will be used in the Adult Index (initially set to missing for csvNew)
indexContribCV - the stream CV that will be used in the 95% confidence interval (initially set to missing for csvNew)
complete - a logical identifying streams and years for which the Adult Index has already been estimated (initially set to FALSE for csvNew)

lscode	year	trapcatch	PEmr	CVmr	comments	index	maintain	complete	indexContrib	indexContribCV	lake	country	strcode	estr	strname	lat	long
5.036	2010	252	1242	18	NA	TRUE	FALSE	TRUE	1242	18	5	US	36	73	Sterling	43.34	-76.69
5.036	2011	85	703	39	NA	TRUE	FALSE	TRUE	703	39	5	US	36	73	Sterling	43.34	-76.69
5.036	2012	36	NA	NA	NA	TRUE	FALSE	TRUE	1726	35	5	US	36	73	Sterling	43.34	-76.69
5.036	2013	280	938	18	NA	TRUE	FALSE	TRUE	938	18	5	US	36	73	Sterling	43.34	-76.69
5.036	2014	216	847	31	NA	TRUE	FALSE	TRUE	847	31	5	US	36	73	Sterling	43.34	-76.69

Error checking the data

The AIcheck() function is used to check the data for errors and generate a report of tables and figures.

AIcheck(streamDat=streamall, csvDir=Dir)

## New RTF document created, C:\temp/2015 Adult Index - error checking.doc

The generated report is an rtf (rich text format) file but it has a *.doc extension so that it will be automatically opened by Microsoft Word. The report is named YYYY Adult Index - error checking.doc, where YYYY is the latest year represented in streamDat, and placed in the directory csvDir. In this example, the path to the error checking report is C:\temp\2015 Adult Index - error checking.doc.

The following conditions are highlighted as potential errors in the data:

mark-recapture estimates without trap catches or CVs
population estimates less than trap catches
for the most recent year
- streams missing population estimates or CVs
- unusually large trap catches
- unusually small trap catches

The error checking report also includes two multi-panel (one panel per stream) time series plots:

adult sea lamprey trap efficiency estimates (trap catch / mark-recapture PE)
adult sea lamprey mark-recapture estimates

Estimating the Adult Index

The AIestimate() function is used to estimate the Adult Index of sea lampreys in a single Great Lake. For example, for Lake Ontario,

ainewOnt <- AIestimate(streamDat=streamall[streamall$lake==5, ])

The annual Adult Index is simply the sum of stream population estimates for each year. Missing stream estimates are estimated by a lake-specific ANOVA model relating the log of the stream estimates to the main effects of each stream and each year, weighted by the inverse of the CV squared.

The 95% confidence interval is estimated from the sum of the mark-recapture variances of the streams.

The returned object from the function (ainewOnt in this example) is a list with 2 components:

streamPE - a data frame of stream mark-recapture and Adult Index contributions for the incomplete rows in streamDat, with the same variables as streamDat

lscode	year	trapcatch	PEmr	CVmr	comments	index	maintain	complete	indexContrib	indexContribCV	lake	country	strcode	estr	strname	lat	long
5.005	2015	233	441	45.35	NA	TRUE	FALSE	TRUE	441	45	5	Canada	5	117	Duffins	43.81	-79.03
5.009	2015	25	NA	NA	no recaps	TRUE	FALSE	TRUE	893	29	5	Canada	9	131	Bowmanville	43.88	-78.66
5.022	2015	1249	1733	12.51	NA	TRUE	FALSE	TRUE	1733	13	5	Canada	22	100	Humber	43.63	-79.47
5.023	2015	2316	6434	8.00	NA	TRUE	FALSE	TRUE	6434	8	5	US	23	19	Black	43.99	-76.06
5.036	2015	149	797	25.00	NA	TRUE	FALSE	TRUE	797	25	5	US	36	73	Sterling	43.34	-76.69

lakeIndex - a data frame of annual lake-wide Adult Indices with 5 columns: lake, year, index, and the 95% confidence intervals ilo and ihi

lake	year	index	ilo	ihi
5	2015	10297.93	8557.329	12038.52

The index estimates for the lakes, lakeainew, looks like this:

lake	year	index	ilo	ihi
1	2015	20224.32	10152.318	30296.32
2	2015	14695.00	12772.938	16617.06
3	2015	23968.00	20539.559	27396.44
4	2015	7112.00	2755.585	11468.41
5	2015	10297.93	8557.329	12038.52

Determining the targets

The AItarget() function is used to calculate lake-wide targets for the Adult Index of sea lamprey populations in the Great Lakes from the mean of specified years. The AItarget() function takes three arguments, but only the first one needs to be provided.

targ <- AItarget(lakeIndex=lakeprev)

lakeIndex - a data frame of annual lake-wide Adult Indices with (at least these) 3 columns: lake, year, and index
years - A list of 5 vectors (one for each Great Lake), specifying the 5 years during which there were acceptable sea lamprey wounding rates on lake trout
adjust - A vector of 5 adjustments (one for each Great Lake) to be made to the mean indices when calculating the target (this is needed if the wounding rates were not acceptable during the specified years)

Default values are provided for years and adjust based on the decisions made by the Great Lakes Fishery Commission. Currently, these values are

lake	years	adjust
Superior	1994-1998	1.0000000
Michigan	1995-1999	0.5617978
Huron	1989-1993	0.2500000
Erie	1991-1995	1.0000000
Ontario	1993-1997	1.0000000

The returned object from the AItarget() function (called targ in this example) is a data frame with the calculated targets for the Adult Index and expanded PE of each Great Lake.

lake	targInd
1	9664
2	24874
3	24113
4	3039
5	11368

Scaling up indices to PEs

Lake-specific conversion factors built in to the GLFC package, index2pe, are used to scale up indices of adult sea lamprey abundance in the Great Lakes to lake-wide population estimates.

	index2pe
Superior	3.95
Michigan	1.87
Huron	2.87
Erie	1.45
Ontario	2.37

lakeall <- plyr::rbind.fill(lakeprev, lakeainew)
lakeallpe <- merge(
  lakeall[, c("lake", "year", "index", "ilo", "ihi")],
  cbind(lake=1:5, i2pe=index2pe))
lakeallpe$pe <- with(lakeallpe, index*i2pe)
lakeallpe$pelo <- with(lakeallpe, ilo*i2pe)
lakeallpe$pehi <- with(lakeallpe, ihi*i2pe)
knitr::kable(tail(lakeallpe, 5), row.names=FALSE)

lake	year	index	ilo	ihi	i2pe	pe	pelo	pehi
5	2002	15675.00	11527.165	19822.83	2.37	37149.75	27319.38	46980.12
5	2003	11378.00	8827.062	13928.94	2.37	26965.86	20920.14	33011.58
5	2004	32586.00	22271.373	42900.63	2.37	77228.82	52783.15	101674.49
5	2005	26260.00	20577.328	31942.67	2.37	62236.20	48768.27	75704.13
5	2015	10297.93	8557.329	12038.52	2.37	24406.09	20280.87	28531.30

Generating a draft report

The AIreport() function is used to create a draft template-style report of the Adult Index estimates of sea lamprey in the Great Lakes.

streamaiall <- rbind(streamall[streamall$complete==TRUE, ], streamainew)
AIreport(streamPEs=streamaiall, lakeIPEs=lakeallpe, targets=targ, csvDir=Dir)

## New RTF document created, C:\temp/2015 Adult Index - draft report.doc

As with the error checking report, the document generated by AIreport() is an rtf file with a *.doc extension, named YYYY Adult Index - draft report.doc.

The report has few paragraphs summarizing the latest estimates, along with 3 tables,

adult sea lamprey indices for the year of interest for each lake with errors and targets
adult sea lamprey indices for all years and all lakes
lake-wide adult sea lamprey abundances for all years and all lakes

and 3 figures,

5-panel time series plot of adult sea lamprey indices (with 95% confidence intervals) and targets for each Great Lake
5-panel bar plot of adult sea lamprey abundance estimates for index streams
map showing the index streams in the Great Lakes and the relative size of the most recent population estimates