-
Computers & Geosciences 32 (2
xteoco
Kevin E. Kohler, Shaun M. Gill
calculation of the ShannonWeaver diversity index for each
species. Additional information can be found at http://
www.nova.edu/ocean/cpce/.
statistics of marine benthic communities (Carletonand Done,
1995). Previous methods included over-
ARTICLE IN PRESS
$Code available from server at http://www.iamg.org/CGEditor/
index.htmCorresponding author. Tel.: +1954 262 3641;laying
underwater photographic images with trans-parent sheets containing
randomly positioned
0098-3004/$ - see front matter r 2005 Elsevier Ltd. All rights
reserved.
doi:10.1016/j.cageo.2005.11.009
fax: +1954 262 4158.
E-mail address: [email protected] (K.E. Kohler).r 2005 Elsevier
Ltd. All rights reserved.
Keywords: Coral point count; Random point count; Coral reef
assessment; Coral reef monitoring; Coral area measurement
1. Introduction and rationale
Random point count methodology is commonlyused in many
population estimation applications,
e.g. forestry vegetation (Stoyan and Penttinen,2000), tree
canopy cover (Thomas and Winner,2000), bird population and
diversity (Ralph et al.,1995; Thompson III et al., 2002; Young and
Hutto,2002). It is also commonly used on frame-grabbedvideo or
still images to estimate the populationNational Coral Reef
Institute, Nova Southeastern University Oceanographic Center, Dania
Beach, FL 33004, USA
Received 19 July 2005; received in revised form 17 November
2005; accepted 21 November 2005
Abstract
Photographic and video methods are frequently used to increase
the efciency of coral reef monitoring efforts. The
random point count method is commonly used on still images or
frame-grabbed video to estimate the community statistics
of benthos. A matrix of randomly distributed points is overlaid
on an image, and the species or substrate-type lying
beneath each point is visually identied. Coral Point Count with
Excel extensions (CPCe) is a standalone Visual Basic
program which automates, facilitates, and speeds the random
point count analysis process. CPCe includes automatic
frame-image sequencing, single-click species/substrate labeling,
auto-advancement of data point focus, zoom in/out, zoom
hold, and specication of random point number, distribution type,
and frame border location. Customization options
include user-specied coral/substrate codes and data point shape,
size, and color. CPCe can also perform image calibration
and planar area and length calculation of benthic features. The
ability to automatically generate analysis spreadsheets in
Microsoft Excel based upon the supplied species/substrate codes
is a signicant feature. Data from individual frames can
be combined to produce both inter- and intra-site comparisons.
Spreadsheet contents include header information,
statistical parameters of each species/substrate type (relative
abundance, mean, standard deviation, standard error) and theCoral
Point Count with Excel eprogram for the determination
using random point006) 12591269
nsions (CPCe): A Visual Basicf coral and substrate coverageunt
methodology$
www.elsevier.com/locate/cageo
-
points (Murdoch and Aronson, 1999). Later im-provements to this
method included the use of
identication required to provide meaningful popu-
species/substrate type (relative abundance, meanthees.
ify-der,oralof
fterbefornedandof
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 125912691260lation estimates. It includes
many user-friendlydesign features, and provides a reliable means
ofdata archiving. This paper describes the featuresand
functionality of CPCe at the time of thiswriting, Version 3.3.
2. Overview
CPCe was designed specically to quickly andefciently calculate
statistical coral coverage over aspecied area. Underwater
photographic frames areoverlaid by a matrix of randomly distributed
points,and the fauna/ora species or substrate type lyingbeneath
each point is visually identied. The speciescode data for each
frame is stored in a .cpc lewhich contains the image lename, point
coordi-nates and the identied data codes. Data fromindividual
frames can be combined to produce bothinter- and intra-site
comparisons via automaticallygenerated Excel spreadsheets. The
transect datasetscan then be statistically analyzed to give
quantita-tive population estimates over the area of interest.In
addition, CPCe has the ability to measure bothplanar length and
area of selected benthic features.This capability allows the
statistical comparison
1Dustan, P., Leard, J., Meier, O., Brill, M., Kosmynin, V.,
1999. PointCount99 software. University of Charleston,
Southcomputer software to generate the random pointpatterns and to
identify the underlying benthicfeatures. PointCount991 is one such
computerprogram.Instead of requiring the counting of each
speci-
men, the random point method relies on statisticalpower analysis
to estimate the actual populationusing the random point samples as
a proxy (Lenth,2001). The technique consists of randomly
distri-buting a number of points onto an underwaterphotographic
image, and then visually identifyingthe features (e.g. coral,
algae, rubble, etc.) lyingunder each point. The percentage of
points over-lying each benthic category is calculated,
andstatistics can be compiled to estimate the populationof biota
such as stony coral, sponges, macroalgae,etc. over a region of
interest. Coral Point Countwith Excel extensions (CPCe) was
developed toincrease the efciency and ease of performing thelarge
number of image analyses and featureCarolina,
http://www.cofc.edu/coral/pc99/pc99.htm.3.1. Image and coral code
file specification
The user species a digital image in the form of a.jpg, .gif, or
.bmp le by selecting File-Open-Rawtheimse operations is detailed
below.
to simplify and speed the processing effort. Each
to ow logically from one operation to the next,
statistical analysis. CPCe was specically desig
automatically assembled into Excel spreadsheets
the images have been processed, the data can
the random points, and saving the data to le. A
species and/or substrate-type lying beneath each
overlaying random points, identifying the c
ing a digital image, dening a frame bor
The basic operations of CPCe consist of spec3. CPCe
operationShannonWeaver diversity index for each speci
and standard deviation), and the calculation ofbetween video
quadrat analyses and in situ mea-surements, and also to determine
growth trendsover time.The primary features of CPCe are:
automatic frame-image sequencing, single-click species/substrate
labeling, auto-advancement of data point focus, multiple point data
assignment, user-specied coral/substrate codes, variable random
point number/distribution, image scaling and calibration, planar
area analysis, batch data input and output, automatic Excel
spreadsheet generation, adjustable frame border dimensions,
color-coded category boxes, zoom in/out and zoom hold, hide/show
random points on image, customizable data point shape/color, code
le integrity checker,A signicant feature of CPCe is the ability
toautomatically generate analysis spreadsheets inMicrosoft Excel
based upon the supplied species/substrate codes. Individual image
frames can beanalyzed separately or multiple frames can becombined
into a single transect datasheet containingheader information,
statistical parameters of eachage le. Images smaller than the
available screen
-
area are expanded to ll the available space, whilelarger images
are reduced in size to t. A set ofmultiple image les can be specied
at once andCPCe will process each le sequentially in alphabe-tical
order. This eliminates the need to manuallyspecify each image le
for analysis, and allows theuser to quickly switch back and forth
betweenimages.
3.2. Border designation
The user then species the rectangular region ofthe image to be
covered with random points. Theregion border can be specied in one
of four ways
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 12591269 1261(Fig. 1). The rst method is to
click and drag tocreate a rectangular area on the digital image.
Theborder can then be manually stretched and movedinto the desired
size and position. The secondmethod is to use the entire image.
Thirdly, the usercan specify a border offset (in pixels) in the x-
andy-direction independently. The border is then drawnon the image
using the appropriate offsets from theimage edges. Lastly, the user
can specify thedimensions of the border region. For this method,the
user must either specify the scaling resolution ofthe digital
image, e.g. number of pixels percentimeter, or the image scaling
can be calculatedwithin CPCe (see Section 6: Program tools).
Afterthe image resolution is specied or calculated, aborder of
specic dimensions can be drawn andpositioned.
3.3. Random point specification
The random points to be overlaid within theborder perimeter can
be specied in one of fourFig. 1. Border boundary specication
options.ways (Fig. 2). The rst way is the simple randommethod where
every pixel within the marked borderboundary has an equal
likelihood of being chosenfor the position of the random points.
The x- andy-coordinates of the random points are
generatedseparately and are calculated as follows. The VisualBasic
random number generator is initialized withthe Randomize statement.
An array newrand of sizenumrand is lled with random numbers in the
range01. A random element of array newrand is chosenand the
corresponding random number chosen. Therandom numbers are then
scaled to producecoordinates which lie within the specied
borderedarea. The chosen newrand element is then relledwith a new
random number for the next iteration.The second border specication
type is the
stratied random method. In this case, the borderedregion is
sub-divided into m rows and n columns,and each cell is populated
with k random pointslying within the cell borders, using the
techniquedescribed for the simple random method. Thisreduces the
potential clumping of the randompoints using the simple random
method, andensures that some random points are present ineach image
cell region. The total number of points ism n k.The third and
fourth border specication types
allow the specication of a uniform grid over thebordered area.
The third type asks for the numberof points in the x- and y-
direction and creates thecorresponding matrix of point coordinates,
ttingthe points exactly within the border boundaries. Inthis case,
the x-spacing of the grid points does notnecessarily equal the
y-spacing.The fourth specication type dictates that the
x-spacing of the points equals the y-spacing, andthe user can
specify the number of points in eitherthe x- or y-direction. The
number of points that willt in the non-chosen direction (e.g.
y-direction if thenumber of points in the x-direction is specied)
iscalculated, and the point grid is centered in the non-chosen
direction.In all the four methods, the maximum number of
overlying points allowed is 500. After the pointdistribution
method is specied, the random pointsare placed inside the bordered
region of the image(Fig. 3).
3.4. Point data assignment
The random points are displayed and labeled
either alphabetically or numerically. Data codes are
-
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 125912691262assigned by clicking the point
label in the pointcodes box on the right side of the screen and
thenthe appropriate species code identier at the bottomof the
screen (Fig. 3). A sample species code le,shallow_coral_codes.txt,
is supplied with CPCe andcontains common corals from the Caribbean
region;however, a user can create a customized le withdata types
most appropriate for a specic analysis.The appendix provides
details for creating acustomized code le. The full name of the
speciesor substrate corresponding to the displayed codecan be seen
by allowing the cursor to hover over thecode box. After clicking on
the appropriate codebox, the corresponding coral code is inserted
in thepoint data table. Only data codes existing in thecode le may
be entered for each point. The NOTEScolumn is used to further
classify the data pointspertaining to disease, bleaching, etc.
Points thathave been assigned data then change color to give a
Fig. 2. Random point specication types: (a) simple random; (b)
stvisual representation of points still requiring identi-cation. If
the number of random points is greaterthan the number of data entry
boxes that can beshown on the screen at once, arrow buttons
areprovided to scroll up or down through the datapoint set.
3.5. Saving the data
At any time during the analysis, the data pointscan be saved to
a .cpc data le. The .cpc le containsthe image lename, the name of
the le containingthe species code identiers, the coordinates of
theborder boundaries, the number of overlying points,the
coordinates of the overlying points, andthe assigned code data for
each point. By savingto a .cpc le, the image and point data can
beretrieved at a later time for further analysis
and/ormodication.
ratied random; (c) uniform grid; and (d) equally spaced
grid.
-
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 12591269 12633.6. Excel spreadsheet
creation
The data from one or more .cpc les can becombined in various
ways to produce Excelspreadsheets. Individual frames can be
analyzed toprovide intra-site comparisons. Alternatively, multi-ple
image frames can be grouped together for singletransect evaluation.
Additionally, multiple transectscan be analyzed together as a
single site or samplingstation. To perform these analyses, the user
speciesone or more .cpc les to process, and thecorresponding Excel
spreadsheets are created. Thecoral categories and classes specied
in the datacode le are inserted automatically into the
spread-sheet. For each transect, two Excel worksheetsare createdone
containing basic statistical ana-lyses of the data such as point
frequency, percen-tage, ShannonWeaver diversity index for
bothspecies and category group (Fig. 4a), and another
Fig. 4. Example Excel spreadsheets generated by CPCe
showing:
(a) point frequency, percentage, and ShannonWeaver diversity
index and (b) raw point code data of a single frame.
Fig. 3. Screenshot of image with overlying random points.
Available coral codes are shown underneath image. Data entry area
is on right.
-
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 125912691264containing the raw point code
data (Fig. 4b). Thissecond worksheet provides an archival record
forthe point code data.Additional transects can be appended to
existing
Excel spreadsheets. A separate data summaryworksheet is created
in each workbook whichcontains statistics among transects including
mean,standard deviation, and standard error of categorygroups (Fig.
5).
4. Program controls
In the design of CPCe, an emphasis was placed onhaving the
necessary tools accessible on a singlescreen window, while
maximizing the screen areaallocated to the analysis image.
Additionally, CPCewas designed to be a self-contained program,
Fig. 5. Example Excel spreadsheet showing darequiring only
Microsoft Excel for automated dataanalysis.There are several
program controls designed to
increase the efciency of assigning the data pointcodes (Fig. 6).
In order to better view the underlyingimage features, all of the
data points can betemporarily hidden. Isolation mode can be used
toshow only the current point instead of all randompoints. Groups
of points can be assigned with asingle click. Point groups are
selected by using thetypical shift-click or ctrl-click
combinations. Click-ing on the data code box assigns that code to
allselected points. The data of the currently selectedpoints can be
cleared by clicking on the Clearselected points icon. Additionally,
all points havinga blank ID eld or blank Notes eld can be
selectedat once, and lled with a single data code with
oneclick.
ta summary for four individual transects.
-
The image can be zoomed in or out by left-clicking or
right-clicking on the image, respectively,with the zoom level
increasing or decreasing by 50%with each click. If the Maintain
zoom option ischecked in the Options menu, the current zoom levelis
maintained when moving among data points.Buttons are provided to
zoom to 100%, 300%, and600% with a single click. When the image is
zoomedin, a border designating the visible region relative to
the entire image is shown on an inset image at thelower right of
the screen.
5. Image enhancement
CPCe includes the capability of basic imageenhancement within
the program. The brightness,sharpness, and contrast of a selected
area of theimage can be adjusted to aid in species
identication(Fig. 7). This is very useful on indistinct object
edgeswhen trying to determine random point position.Since it is
rarely necessary to enhance an entireimage, the image enhancement
is performed on onlya user-selected subset of the image to speed
thecreation of the enhanced area display.To enhance an area of the
image, the user selects
the desired area by clicking and dragging. The typeand strength
of image enhancement (brightness,sharpness, and contrast) desired
is then specied,and the modied image area is displayed.
Thebrightness and sharpen lters are accumulative,
ARTICLE IN PRESS
Fig. 6. CPCe program controls.
K.E. Kohler, S.M. Gill / Computers & Geosciences 32 (2006)
12591269 1265Fig. 7. Image enhancintegrity checks.whereas the
contrast lter is not.
6. Program tools
Several program tools are provided to performtasks such as area
measurements and various dataement interface.
-
6.1. Image calibration and planar area analysis
CPCe includes the ability to calibrate an imageand to calculate
a planar area or length of adisplayed object. This is useful to
document the sizeand area of bottom features, and to aid in
theinvestigation of growth trends of a coral colonyover time.The
rst step in area or length analysis is to
determine the scaling resolution of the image, e.g.number of
pixels per centimeter. To determinethe scaling, an image is
selected, and two pointsthat are a known distance apart are marked
onthe image (Fig. 8). The user enters the distancespanned by the
points, and the scaling resolution iscalculated.Once the image
calibration is known, areas and
lengths on the image can be determined. Tocalculate a length,
points at the beginning and endof a linear segment are clicked, and
the distancebetween them is calculated and displayed. To
outlined by pressing shift-right-click. Instructions toaccess
these capabilities are displayed on the rightside of the screen. An
example of traced andcalculated areas is seen in Fig. 9.The image
including the traced area(s) can be
saved to an image (.bmp) le, either with or withoutthe area
information labels. Also, the traced area(s)information can be
saved to an .ara le which canbe retrieved for later re-analysis in
CPCe.
6.2. Data check and species search in .cpc files
This option searches a selected set of .cpc les forunassigned
points or the occurrence of a speciedspecies code. It provides the
ability to identifyframes with incomplete data entries, and to
locateframes containing possibly misidentied species.This is very
useful for analysis involving largenumbers of image frames as it
eliminates the need tomanually inspect each data le to ensure
complete
code or image les into other folder locations
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 125912691266calculate an area, a trace is
performed by left-clicking the mouse, moving the mouse around
theperiphery of the desired area, and then right-clicking the mouse
which connects the rst and lastpoints and closes the path. The area
of the enclosedregion is calculated and displayed. Other
featuresinclude the ability to temporarily stop a trace, andto
erase partial traces. Once an area is traced, theuser can toggle
between having the area lled orFig. 8. Screenshot of image
scalipost-analysis.data point assignments.
6.3. Change code file/image file directory location
This option changes the directory location ofeither the coral
code le or image le in a selectedset of .cpc les. This allows the
user to move theng and calibration process.
-
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 12591269 12676.4. Code file check
This option nds any obvious errors in a user-customized species
code le such as incorrectsyntax, missing required category elds,
and dupli-cate species codes.
7. Program options
There are several user-customizable programoptions available.
CPCe uses a conguration leto hold the current setting of the
various settingsand options. Each setting/option change is
writtento the conguration le, and the current setting/option
conguration is recalled each time theprogram starts. The program
options are listedbelow.
Fig. 9. Screenshot showing traced areas and lengths, outlined
area, a
bitmapped images.7.1. Data point graphical parameters
This option customizes the shape, size, and colorof the data
point objects. The available shapes arecircle, circle with
crosshairs, box, triangle, cross-hairs, all either lled (solid) or
outlined. The colorsof unassigned points, assigned points, the
currentfocus point, and the frame border can also bespecied, which
makes it easier to determine whichpoints remain to be identied.
Also, the size of thefont in the coral code boxes can be
specied.
7.2. Color-coded code category boxes
This option allows the user to specify customcolors for each of
the species and substrate codecategories appearing under the
analysis image. Thishelps in quickly nding a particular code box
whenusing many codes. Blank boxes can also be inserted
nd movable text boxes. Areas of traced regions can be saved
as
-
Walker, and Dr. Ivor Williams for their helpfulcomments which
signicantly improved the useful-
species identiers. The format of the coral code leis as
follows:
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 125912691268in the coral code le to separate
the coral codes forenhanced visual recognition.
7.3. Specify coral code file
The user can specify different data code les forindividual
analyses.
7.4. Letters/numbers
The user has the option of using either numbersor letters to
identify the data points on the image.
7.5. Maintain zoom
If this option in the control box is checked, thecurrent zoom
level is maintained after assigningeach data point.
7.6. Auto-advance and auto-follow
If the auto-advance option is checked, after apoint is assigned
a category, the next point insequence becomes the current point,
and the cursoris positioned at the corresponding data entry
label.This eliminates the need to manually click each datapoint
before assigning it a category value.If the auto-follow option is
checked, and main-
tain zoom is checked, the image shifts to center thecurrent
point in the display area, while maintainingthe same zoom
level.
8. Summary
Coral Point Count with Excel extensions (CPCe)is a standalone
Visual Basic program which auto-mates the random point count method
for thestatistical analysis of marine benthic communities.Features
such as group point selection, auto-follow,auto-advance, isolation
mode, batch processing ofboth input and output datasets, and image
enhance-ment are specically designed to speed the analysisof
photo-quadrats. In order to ensure data integrity,the data are
saved separately in .cpc format whichcan then be automatically
assembled into MicrosoftExcel spreadsheets. By saving the
individual .cpcles, the resultant Excel spreadsheets can be
easilyregenerated at any time. CPCe also provides thecapability to
perform planar area and lengthanalyses on individual coral colonies
and other
benthic features.The number of general coral categoriesFor each
coral category: category symbol,category nameFor each coral type:
coral code, coraldescription, category nameNOTES, NOTES, NOTES
(this line separatesthe coral names from the notes descriptors).For
each notes descriptor: notes code,ness and functionality of CPCe.
We also thank Drs.Bernhard Riegl and Richard Dodge for their
helpfulmanuscript suggestions. Thanks are also extendedto the many
researchers at the National Coral ReefInstitute, Florida Fish and
Wildlife ConservationCommission Fish and Wildlife Research
Institute,University of Miami Rosenstiel School of Marineand
Atmospheric Science, and the University ofHawaii for providing
feedback on the programsoperation and effectiveness.This work is a
result of research funded by the
National Oceanic and Atmospheric AdministrationCoastal Ocean
Program under award #NA03-NOS4260046 to Nova Southeastern
University forthe National Coral Reef Institute (NCRI). This isNCRI
contribution No. 75.
Appendix A. Coral code le creation
A coral code le is an ASCII text le containinggeneral coral
categories and individual codes andCPCe is continually updated, and
the authors arereceptive to suggestions from other researchers.
Thefeatures described in this paper are found in thecurrent version
of CPCe, Version 3.3. CPCe is beingmade available free of charge to
interested research-ers afliated with scientic institutions.
Moreinformation, as well as instructions for downloadingthe
software, can be found at
http://www.nova.edu/ocean/cpce/index.html.
Acknowledgments
The authors thank in alphabetical order Dr. JeanKenyon, Ryan
Moyer, Dr. James Stoddart, Briandescription, NA (not
applicable)
-
Example:
categories, as well as the category Coral. Addi-tionally, every
category must contain at least onemember.Blank boxes can be
inserted within the coral code
list to separate groups of coral codes. This is doneby entering
Blank,Blank,Blank. These boxesappear solid black on the screen.
Appendix B. Supplementary material
Supplementary data associated with this articlecan be found in
the online version at doi:10.1016/j.cageo.2005.11.009
References
Carleton, J.H., Done, T., 1995. Quantitative video sampling
of
coral reef benthos: large-scale application. Coral Reefs 14,
3546.
Lenth, R.V., 2001. Some practical guidelines for effective
sample size determination. The American Statistician 55,
187193.
Murdoch, T.J.T., Aronson, R.B., 1999. Scale-dependent
spatial
variability of coral assemblages along the Florida Reef
Tract.
Coral Reefs 18, 341351.
ARTICLE IN PRESSK.E. Kohler, S.M. Gill / Computers &
Geosciences 32 (2006) 12591269 1269Due to the requirements of the
associated Excelspreadsheets, the category TWS (Tape, wand,shadow)
must be included as one of the coralMonitoring bird populations by
point counts. US Forest
Service Gen. Technical Report PSW-GTR-149, Albany, CA;
US Department of Agriculture, Forest Service, Pacic
Southwest Research Station, 187pp.
Stoyan, D., Penttinen, A., 2000. Recent applications of
point
process methods in forestry statistics. Statistical Science
15
(1), 6178.
Thomas, S.C., Winner, W.E., 2000. Leaf area index of an old-
growth Douglas-r forest estimated from direct structural
measurements in the canopy. Canadian Journal Forestry
Research 30, 19221930.
Thompson III, F.R., Burhans, D.E., Root, B., 2002. Effects
of
point count protocol on bird abundance and variability
estimates and power to detect population trends. Journal of
Field Ornithology 73 (2), 141150.
Young, J.S., Hutto, R.L., 2002. Use of regional-scale
exploratory
studies to determine bird-habitat relationships. In: Scott,
J.M., Heglund, P.J., Morrison, M.L., Hauer, J.B., Raphael,
M.G., Wall, W.A., Samson, F.B. (Eds.), Predicting Species
Occurrences: Issues of Scale and Accuracy. Island Press,
Washington, DC, pp. 107119.Ralph, C.J., Sauer, J.R., Droege, S.,
(technical editors), 1995.
Coral Point Count with Excel extensions (CPCe): A Visual Basic
program for the determination of coral and substrate coverage using
random point count methodologyIntroduction and
rationaleOverviewCPCe operationImage and coral code file
specificationBorder designationRandom point specificationPoint data
assignmentSaving the dataExcel spreadsheet creation
Program controlsImage enhancementProgram toolsImage calibration
and planar area analysisData check and species search in .cpc
filesChange code file/image file directory locationCode file
check
Program optionsData point graphical parametersColor-coded code
category boxesSpecify coral code fileLetters/numbersMaintain
zoomAuto-advance and auto-follow
SummaryAcknowledgmentsCoral code file creationSupplementary
materialReferences