This action might not be possible to undo. Are you sure you want to continue?
https://www.scribd.com/doc/37253355/dfdinstructions
10/29/2012
text
original
IRC Ho use
The Sq uar e
Penningt o n
Lymingt o n
SO4 1 8 GN
Engl and
Pisc es@ir c ho use.demo n.c o .uk
www.ir c ho use.demo n.c o .uk
Pho ne 44 (0) 1590 676622
Fax 44 (0) 1590 675599
DENSITY FROM
DISTANCES
A pr ogr am f or c al c ul at i n g an i mal an d pl an t
den si t i es usi n g pl ot l ess an d di st an c e
sampl i n g met h ods
PISCES LICENCE AGREEMENT
This is a legal agreement between you the end user and PISCES Conservation Ltd. Lymington (PISCES). BY OPENING THIS
PACKAGE YOU ARE AGREEING TO BE BOUND BY THE TERMS OF THIS AGREEMENT. IF YOU DO NOT AGREE TO
THE TERMS OF THIS AGREEMENT PROMPTLY RETURN THE UNOPENED PACKAGE AND ALL ACCOMPANYING
ITEMS (including written material ) TO THE PLACE YOU OBTAINED THEM FOR A FULL REFUND.
1. GRANT OF LICENSE  This PISCES License Agreement (‘License’) permits you to use one copy of the PISCES software
product acquired with this License (SOFTWARE ) on any single computer, provided the SOFTWARE is in use on only one
computer at any time. If you have multiple Licenses for the SOFTWARE then at anytime you may have as many copies of
the SOFTWARE in use as you have Licenses. The SOFTWARE is ‘in use’ on a computer when it is loaded into the
temporary memory (i.e. RAM) or installed into the permanent memory (e.g. hard disk, CD ROM. or other storage device)
of that computer, except that a copy installed on a network server for the sole purpose of distribution to other computers is
not ‘in use’. If the anticipated number of users of the SOFTWARE will exceed the number of applicable Licenses then you
must have a reasonable mechanism or process in place to assure that the number of persons using the SOFTWARE
concurrently does not exceed the number of Licenses. If the SOFTWARE is permanently installed on the hard disk or other
storage device of a computer (other than network server) and one person uses that computer more than 80% of the time it
is in use then that person may also use the SOFTWARE on a portable or home computer.
2. COPYRIGHT  The SOFTWARE is owned by PISCES or its suppliers and is protected by all applicable national laws.
Therefore, you must treat the SOFTWARE like any other copyrighted material (e.g. a book) except that if the software is
not copy protected you may either (a) make one copy of the of the SOFTWARE solely for backup or archival purposes, or
(b) transfer the SOFTWARE to a single hard disk provided you keep the original solely for backup or archival purpose. You
may not copy the Product manual(s) or written materials accompanying the SOFTWARE.
3. DUAL MEDIA SOFTWARE  If the SOFTWARE package contains both 31/2" and 51/4" disks, then you may use only one
set (either the 31/2" or 51/4') of the disks provided. you may not use the other disks on another computer or computer
network, or lend, rent, lease, or transfer them to another user except as part of a transfer or other use expressly permitted
by this PISCES License Agreement.
4. OTHER RESTRICTIONS  You may not rent or lease the SOFTWARE, but you may transfer your rights under this PISCES
License Agreement on a permanent basis provided you transfer all copies of the SOFTWARE and all written materials, and
the recipient agrees to the terms of this Agreement. You may not reverse engineer, decompile or disassemble the
SOFTWARE. Any transfer must include the most recent update and all prior versions.
LIMITED WARRANTY
LIMITED WARRANTY  PISCES warrants that (a) the SOFTWARE will perform substantially in accordance with the
accompanying Product Manual(s) for a period of 90 days from the date of receipt; and (b) any PISCES supplied hardware
accompanying the SOFTWARE will be free from defects in materials and workmanship under nominal use and service for a
period of one year from the date of receipt. Any implied warranties on the SOFTWARE and hardware are limited to 90 days and
one (1) year respectively or the shortest period permitted by applicable law, whichever is greater.
CUSTOMER REMEDIES  PISCES'S entire liability and your exclusive remedy shall be, at PISCES option, either (a) return of
the price paid or (b) repair or replacement of the SOFTWARE or hardware that does not meet PISCES’S Limited Warranty, and
which is returned to PISCES with a copy of your receipt. This Limited Warrant is void if failure of the SOFTWARE or
hardware has resulted from accident, abuse or misapplication. Any replacement SOFTWARE will be warranted for the
remainder of the original warranty period or 30 days, whichever is longer.
NO OTHER WARRANTIES  TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, PISCES DISCLAIMS
ALL OTHER WARRANTIES. EITHER EXPRESS OR IMPLIED, INCLUDING BUT LIMITED NOT TO IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, WITH RESPECT TO THE
SOFTWARE, THE ACCOMPANYING PRODUCT MANUAL(S) AND WRITTEN MATERIALS, AND ANY
ACCOMPANYING HARDWARE. THE LIMITED WARRANTY CONTAINED HEREIN GIVES YOU SPECIFIC LEGAL
RIGHTS.
NO LIABILITY FOR CONSEQUENTIAL DAMAGES  TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW
PISCES AND ITS SUPPLIERS SHALL NOT BE LIABLE FOR ANY OTHER DAMAGES WHATSOEVER (INCLUDING,
WITHOUT LIMITATION, DAMAGES FOR LOSS OF BUSINESS PROFITS, BUSINESS INTERRUPTION, LOSS OF
BUSINESS INFORMATION, OR OTHER PECUNIARY LOSS) ARISING OUT OF THE USE OF OR INABILITY TO USE THIS
PISCES PRODUCT, EVEN IF PISCES HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN ANY CASE,
PISCES’S ENTIRE LIABILITY UNDER ANY PROVISION OF THIS AGREEMENT SHALL BE LIMITED TO THE AMOUNT
ACTUALLY PAID BY YOU FOR THE SOFTWARE.
This Agreement is governed by the laws of England.
Should you have any questions concerning this Agreement, or if you desire to contact PISCES for any reason, please use the
address information enclosed in this product to contact PISCES or write:
PISCES Conservation Ltd
IRC House, The Square
Pennington
LYMINGTON
Hampshire
England
SO41 8GN
Contents
I. INTRODUCTION............................................................................................................. 1
II. INSTALLATION AND GENERAL INSTRUCTIONS......................................................... 1
A. SYSTEM REQUIREMENTS ................................................................................................. 1
B. INSTALLATION................................................................................................................ 1
C. GENERAL INSTRUCTIONS................................................................................................. 1
D. OBTAINING HELP............................................................................................................ 2
E. OPENING A DATA SET ..................................................................................................... 2
F. DEMONSTRATION DATA SETS............................................................................................ 2
G. DATA ENTRY.................................................................................................................. 2
H. SETTING AND CHANGING UNITS........................................................................................ 5
I. EDITING EXISTING DATA...................................................................................................... 5
J. MAXIMUM SIZE OF THE DATA SET...................................................................................... 5
K. PRINTING AND EXPORTING YOUR RESULTS.......................................................................... 5
L. ZOOMING AND PANNING ON GRAPHS.................................................................................. 6
III. LINE TRANSECT METHODS..................................................................................... 6
A. POINT AND LINE SURVEY THEORY...................................................................................... 6
1. Line transect methods: the Fourier series estimator................................................ 6
2. Point transects ........................................................................................................ 9
B. FOURIER MODEL FOR FULL DISTANCE LINE TRANSECT DATA ................................................ 10
1. Viewing the histogram of distances....................................................................... 10
2. Truncating the data ............................................................................................... 10
3. Viewing the results................................................................................................ 11
C. HALFNORMAL MODEL FOR FULL DISTANCE LINE TRANSECT DATA......................................... 12
1. Viewing the histogram of distances....................................................................... 12
2. Truncating the data ............................................................................................... 12
3. Viewing the results................................................................................................ 12
D. ALL OBJECTS DETECTED DURING LINE TRANSECT SURVEYS ................................................ 12
E. FOURIER MODEL FOR NBELT LINE TRANSECT DATA............................................................ 13
1. Viewing the histogram of distances....................................................................... 13
2. Viewing the results................................................................................................ 13
3. Goodness of fit...................................................................................................... 13
F. LINEAR DETECTION MODEL FOR 2BELT LINE TRANSECT DATA ............................................. 13
1. Viewing the results................................................................................................ 14
G. EXPONENTIAL DETECTION MODEL FOR 2BELT LINE TRANSECT DATA .................................... 14
1. Viewing the results................................................................................................ 14
H. POINT TRANSECT  2BELT MODEL .................................................................................. 14
1. Viewing the results................................................................................................ 14
I. POINT TRANSECT  HALF NORMAL MODEL............................................................................. 14
1. Viewing the results................................................................................................ 15
IV. PLOTLESS DENSITY ESTIMATORS....................................................................... 15
A. CLOSEST INDIVIDUAL (COTTAM ET AL., 1953).................................................................. 15
B. NEAREST NEIGHBOUR (COTTAM & CURTIS, 1956) ............................................................ 16
C. KENDALLMORAN (KENDALL AND MORAN, 1963) ............................................................. 16
D. ORDERED DISTANCE CLOSEST INDIVIDUAL (MORISTA, 1957) ............................................. 17
E. ORDERED DISTANCE SECOND CLOSEST INDIVIDUAL (MORISTA, 1957)................................. 17
F. ORDERED DISTANCE THIRD CLOSEST INDIVIDUAL (MORISTA, 1957) .................................... 17
G. ANGLEORDER NEAREST INDIVIDUAL (STEARNS, 1949) ..................................................... 18
H. ANGLEORDER SECOND NEAREST INDIVIDUAL (MORISTA, 1957) ......................................... 18
I. DATA SIMULATION........................................................................................................... 18
1. Plotless density estimation.................................................................................... 18
2. Line transect simulation ........................................................................................ 20
V. IMPORTANT REFERENCES.................................................................................... 21
Pisces Conservation Ltd 1
I. Introduction
DfD (Density from Distances) is a Windows® program that offers a range of analytical
techniques commonly used by ecologists to estimate animal and plant density. Researchers
in other fields may also find these methods useful for estimating density. DfD has been
designed to be easy to use and is particularly suitable for ecological teaching because it
allows students to quickly enter or simulate data and explore a range of methods within a
familiar Windows setting.
The survey methods used in Density from Distances aim to estimate density using
observations on the distance between objects or from a selected line or point to the animals
or plants and thus do not require the worker to accurately map out or define the sampling
area. They are particularly appropriate for the estimation of population density for plants or
large animals living at low density in difficult to traverse habitat. For example, distance
sampling using a Fourier series is frequently the method of choice for estimating population
size of primates in neotropical forest. The animals are counted while walking along specially
cut forest trails. The high density of the forest and the low density and mobility of the
monkeys would make it futile to try to census a number of predefined quadrates. Distance
methods are particularly appropriate for static objects such as plants and nests. For bird
surveys the difficulty of accurately estimating distance results in the need for simple point or
line transect methods where distance is divided into two or more categories. Such methods
are available in DfD. An important feature of the program is the range of simulations that can
be undertaken. These are useful for both students who wish to learn about the various
methods and researchers who need to appreciate the level of accuracy they are likely to
obtain. The simulated data can be the basis of a classroom practical of the merits of the
various methods.
II. Installation and General instructions
A. System requirements
1. A PC running Windows® 95 or later operating system.
2. Minimum of 2 MB spare hard disk space.
B. Installation
1. Place the DfD CD in the CD drive.
2. The installation program should start automatically; follow the onscreen instructions.
3. If the CD does not autoplay, browse the CD in Windows Explorer or My Computer,
and doubleclick Setup.exe in the root directory.
When installation is complete there will be a DfD entry on your Start: Programs menu and a
folder, C:\Program Files\DfD on your hard disk that holds the program files. An uninstall
program will also be created to remove the program if you wish. The instruction manual (in
Adobe Acrobat PDF format) is also available from the Start: Programs: DfD menu.
C. General instructions
Start DfD in the normal Windows fashion either by clicking on the program icon or from the
start button. Along the top bar are a number of pulldown menus. These work in the same way
as standard Windows programs.
File: To open, initiate, print, export and save data sets.
Edit: The standard Windows menu.
Options: Allows the setting of defaults, window views and units of measurement.
Simulations: To simulate plotless and transect data.
Help: to enter the Help system.
About: Details about the program version.
Pisces Conservation Ltd 2
When the program is started, you will be presented with a blank data grid. From the File
menu choose Open to load an existing data set and New to start entering new data. To enter
new data select the type of data you wish to enter from the dropdown menu above the data
grid and fill in the grid with appropriate data. To see the type of data that is required run
simulations from the Simulation dropdown menu and view the data grid.
D. Obtaining help
For most active windows context sensitive help can be obtained by pressing F1, clicking on
the help button or selecting the help drop down menu. Help and frequently asked questions
are available on our web site at http://www.piscesconservation.com
If you have problems using the program or entering data which you cannot solve then contact
Pisces Conservation Ltd by email pisces@irchouse.demon.co.uk or by phone to England 44
(0)1590 676622 during office hours (09.00 to 17.00).
E. Opening a data set
Use FileOpen to start the file dialogue to select an existing data file for analysis. The default
file extension is .csv as the data is stored in simple comma delimited form.
F. Demonstration data sets
DfD is supplied with a number of demonstration data sets as follows:
Stakes.csv The stakes demonstration data set for a line transect from page 62 of Burnham
et al (1980).
p68_belt.csv An example of a line transect with distance belts from page 68 of Burnham et
al. (1980).
Multi_point.csv Multiple point data for a simulated data set with an actual density of 0.05
objects per square metre.
Closest.csv A closest neighbour data set from a population with a density of 1 per square
metre.
KM.csv A data set of distances from a random point to its closest object and from this object
to its nearest neighbour for a population with a density of 1 object per square metre.
G. Data entry
Data is entered in the Data window that is opened by clicking on the Data tab. To create a
new data set select File New from the pulldown menu. Because the different methods
require different data structures you must then select the method you want to apply from the
drop down menu in the top panel. The different options and the data they require are
described below. The data simulator will automatically organise the data correctly. See
Setting and Changing Units to select the units used to measure distances.
Transect, Perpendicular Distance
Select this option if you want to estimate
density for line transect data and you have a
series of measurements of perpendicular
distance from the transect line. The data grid
that will be formed is shown below. You must
also select the units of measurement in the
dropdown menu on the righthand side of the
top panel.
Pisces Conservation Ltd 3
To enter data click on a cell and type. To move down a row press the return key or move the
mouse to the cell and click. Cells holding perpendicular distances will be coloured blue. To
add another transect click on the Add New Transect button. All the data for a transect can be
removed by double clicking its title cell and selecting Delete Column. This dialogue box also
allows the transect title to be changed.
Transect, Distance and Angle
Select this option if you want to estimate density for line transect data and you have a series
of measurements of the distance to the objects and their angle from the transect line. The
data grid that will be formed is shown below. You must also select the units of measurement
in the dropdown menu on the righthand side of
the top panel.
To enter data click on a cell and type. To move
down a row press the return key or move the
mouse to the cell and click. Cells holding
perpendicular distances will be coloured blue. To
add another transect click on the Add New
Transect button. All the data for a transect can
be removed by double clicking its title cell and
selecting Delete Column. This dialogue box also
allows the transect title to be changed.
Transect, Perpendicular Distance  belted
Select this option if you want to estimate density for line transect data and you have a series
of the number of counts of objects in set belts of distance away from the transect line. The
data grid that will be formed is shown below. In the outer limit column place the distance to
the outer edge of each band. You must also select the units of measurement in the drop
down menu on the righthand side of the top
panel.
To enter data click on a cell and type. To
move down a row press the return key or
move the mouse to the cell and click. Cells
holding perpendicular distances will be
coloured blue. To add another transect click
on the Add New Transect button. All the data
for a transect can be removed by double
clicking its title cell and selecting Delete
Column. This dialogue box also allows the transect title to be changed.
Point transect
Select this option if you want to estimate density for point transect data and you have a
series of measurements of the distance to the objects. The data grid that will be formed is
shown below. You must also select the units
of measurement in the dropdown menu on
the righthand side of the top panel.
To enter data click on a cell and type. To
move down a row press the return key or
move the mouse to the cell and click. Cells
holding perpendicular distances will be
coloured blue. To add another transect click
on the Add New Transect button. All the data
Pisces Conservation Ltd 4
for a transect can be removed by double clicking its title cell and selecting Delete Column.
This dialogue box also allows the transect title to be changed.
Point transect – with belts
Select this option if you want to estimate density for point transect data and you have a
series of the number of counts of objects in set belts of distance away from the observation
point. The data grid that will be formed is shown below. You must also select the units of
measurement in the dropdown menu on the
righthand side of the top panel.
To enter data click on a cell and type. To
move down a row press the return key or
move the mouse to the cell and click. Cells
holding perpendicular distances will be
coloured blue. To add another transect click
on the Add New Transect button. All the data
for a transect can be removed by double
clicking its title cell and selecting Delete
Column. This dialogue box also allows the transect title to be changed.
Nearest Neighbour/Closest Individual
Select this option if you want to estimate density using a plotless technique such as nearest
neighbour, closest individual or an ordered distance method. The data grid that will be formed
is shown below. You must also select the units of measurement in the dropdown menu on
the righthand side of the top panel.
To enter data click on a cell and type. To
move down a row press the return key or
move the mouse to the cell and click. Cells
holding perpendicular distances will be
coloured blue. To add another transect click
on the Add New Transect button. All the data
for a transect can be removed by double
clicking its title cell and selecting Delete
Column. This dialogue box also allows the
transect title to be changed.
AngleOrder
Select this option if you want to estimate density using the angleorder plotless method. The
data grid that will be formed is shown below. You must also select the units of measurement
in the dropdown menu on the righthand side of the top panel.
To enter data, click on a
cell and type. To move
down a row press the
return key or move the
mouse to the cell and click.
Cells holding perpendicular
distances will be coloured
blue. To add another
transect click on the Add
New Transect button. All
the data for a transect can
be removed by double
Pisces Conservation Ltd 5
clicking its title cell and selecting Delete Column. This dialogue box also allows the transect
title to be changed.
Kendal Moran
Select this option if you want to estimate density using the KendalMoran plotless method.
The data grid that will be formed is shown below. You must also select the units of
measurement in the dropdown menu on the righthand side of the top panel.
To enter data, click on a cell and type. To move down a row press the return key or move the
mouse to the cell and
click. Cells holding
perpendicular distances
will be coloured blue. To
add another transect click
on the Add New Transect
button. All the data for a
transect can be removed
by double clicking its title
cell and selecting Delete
Column. This dialogue box
also allows the transect
title to be changed.
H. Setting and Changing Units
The units of measurement are selected using the Select units dropdown dialogue box in the
top panel of the data tab sheet. Units available are millimetres, centimetres, metres,
kilometres, feet, yards, miles and nautical miles.
The units of measurement for the calculated density estimates are chosen from
OptionsSelect density units of measurement. The units available are square metres,
kilometres, nautical miles, miles, yards and feet. You can convert between units at any time
even after the density has been calculated.
I. Editing existing data
The raw data grid can be edited by using the mouse to click on a cell to select it and typing in
a new value. These changes will not alter a saved file until FileSave is undertaken. Use
FileSave as to save your data under another name.
J. Maximum size of the data set
The maximum number of observations in any column is 500.
K. Printing and exporting your results
Any active window can be printed using FilePrint and data or graphs can be copied to the
clipboard using edit copy in the normal Windows fashion. When print is selected a print
preview window is opened in which size, margins, colour etc. can be specified. Copies can
be made to the Windows clipboard or to a file from FileExport clicking on the Copy to
clipboard or Copy to file buttons in the dialogue box. Images of graphs can be stored as
bitmaps, metafiles or enhanced metafiles. For windows users, who wish to include a graph in
another document, the metafile or enhanced metafile formats are to be preferred as they will
allow the image to be changed in size without loss of resolution of the text.
Pisces Conservation Ltd 6
L. Zooming and panning on graphs
To zoom in on an area, move to the top left corner of the area to be enlarged then hold the
left hand mouse button and drag to the lower right hand corner and release the button. An
enlarged view of the selected area will be displayed. To return to the original view, hold down
the lefthand mouse button and move upwards and to the left and release. To pan the graph
hold down the right hand mouse button and move the mouse.
III. Line transect methods
Counting the number of sightings forms the basis for estimating density for many animal
groups. This is particularly the case for large or easily seen animals such as birds, large
grassland mammals, whales and large, active insects such as butterflies. While it may be
possible to count animals from a suitable vantage point or while moving along a transect, the
count can only be converted to a density estimate if the area scanned can be estimated.
This simple approach is often difficult to undertake for two reasons, firstly, it may not be
possible to estimate sufficiently accurately the area scanned and secondly, not all of the
animals present may have been spotted. Distancing sampling methods have been developed
to allow for these problems by assuming that the likelihood that an individual will be observed
will decline in a mathematically definable way with distance. The methods available within
DfD are also useful for sessile or slow moving organisms such as barnacles, corals and
molluscs. They can be applied to data collected nonvisually if distance can be estimated, for
example a bird census may be based on bird song or an electric fish survey on the detection
of the distinctive electrical signals. One problem, as can be the case with a visual survey, is
to ensure that the same individual is only counted once.
If it is possible to count all of the individuals, n, within a known area, a, then this is termed a
census and the estimated density, D, is simply:
Counting often requires the observer to move over the census area and thus favours the use
of strip transects (long, thin quadrats) of length, L, and width, 2w, along which the observer
moves in a straight line. The counts obtained suffer from error, which varies with the counting
rate and bias because of the tendency of observers to undercount. LeResche & Rausch
(1974) from a study of bias during aerial surveys of moose Alces alces concluded that this
bias was sufficient to invalidate the method as a means of absolute population estimation. A
survey of 17 studies on large mammals by Caughly (1974) found that the proportion of the
population counted varied from 23 to 89%. Dodd & Murphy (1995) made a comparison of
census methods for bird nest estimation.
A. Point and line survey theory
1. Line transect methods: the Fourier series estimator
Line transect methods have been developed for situations when it is not possible to count all
the animals within a strip transect. The methods are based on the idea that only animals lying
on the centre line of the of the strip transect along which the observer moves will be certain
to be detected and that the probability of detection will fall with perpendicular distance from
this line. The techniques presented here have been reviewed in detail by Buckland et al.
(1993). For many animals the observer may walk the transect, but counts can also be
undertaken from motor vehicles, ships and remote operated vehicles.
For these methods it is assumed that:
1. Objects on the line are always detected.
a
n
D ·
ˆ
Pisces Conservation Ltd 7
2. The observer does not influence the recorded positions. For mobile animals, the position
must be that prior to any response to the presence of the observer: The theory has been
developed under the assumption that the objects are immobile, but slow movement in
relation to the observer creates little inaccuracy.
3. Distances and/or angles are measured accurately.
4. The objects are correctly identified
The basic field procedure is for the
transect route to be a straight line of
length, L, randomly placed with
respect to the animals or objects to
be counted and the perpendicular
distance to each detected object of
interest, x, recorded. In practice a
number of lines, arranged as a
regular grid, and randomly placed in
the study may be used. Further, it is
usually easier to record the sighting
distance, r, which is the distance
from the observer to the object, and
the sighting angle, θ, which is the
angle of the object from the transect
line and calculate x rather than
measuring x directly. Methods of population estimation based on r and θ were reviewed by
Hayes & Buckland (1983) but are not available in DfD as they are considered inferior to
those based on the perpendicular distance (Buckland et al., 1993).
If all n objects in a strip of length, L, and width, 2w are counted then the estimated density is:
.
If a proportion, P, of the animals present are detected then the equation becomes:
.
Line transect methods use the distribution of the perpendicular detection distances to
estimate P.
It is assumed that the number of observations will decline with perpendicular distance from
the line and a detection function, g(x) which describes this reduction is fitted to the data.
Given assumption (1) above, g(0), the probability of detecting an object lying on the line, is
assumed to equal 1.
The probability of detecting an object within a strip of area 2wL, P, is:
,
which on substitution gives
wL
n
D
2
·
)
wLP
n
D
2
·
)
w
dx x g
P
w
∫
·
0
) (
∫
·
w
dx x g L
n
D
0
) ( ˆ 2
ˆ
Angle
Sighting distance
Perpendicular distance
Pisces Conservation Ltd 8
As it is assumed g(0) =1, the probability density function (pdf) evaluated at x = 0 is
and thus the general estimator of density is often expressed as:
.
Density is expressed as numbers per square unit of length where length of the transect and
distance to the objects are expressed in the same length units such as metres or kilometres.
A flexible function with the desired properties to use as a detection function is a cosine
series. This is the Fourier model of Crain et al. (1979) and is a general model that has been
shown to give good results for a wide variety of data.
When perpendicular distance has been measured as a continuous variable and not grouped
into size classes the Fourier series f(0) is given by:
,
where
w
*
is the transect halfwidth; without truncation this is the largest perpendicular distance
observed; x
i
the perpendicular distance of the ith animal, n the number of observations and m
the number of cosine terms determined by the stopping rule. that you choose the first value
of m such that
.
Experience suggests that m should be < 7.
Having calculated f(o) D is estimated. The sampling variance of D requires considerable
computation. First calculate the variance and covariances of the parameters a
k
,
∫
·
w
dx x g
f
0
) (
1
) 0 (
L
f n
D
2
) 0 (
ˆ
ˆ
·
∑
·
+ ·
m
k
k
a
w
f
1
*
ˆ
1
) 0 (
ˆ
]
]
]
,
`
.

·
∑
·
n
i
i
k
w
x k
nw
a
1
* *
cos
2 π )
1
2
1
*
ˆ
1
2 1
+
≥
,
`
.

+
m
a
n w
]
]
]
−
,
`
.

+
−
·
2
* 2 *
2 1
1
1
) ˆ var(
k k k
a
w
a
w n
a
Pisces Conservation Ltd 9
.
Then sum the variancecovariance matrix to give
.
Finally, the estimated variance is calculated using
.
It is normally assumed that the variance of the number of animals counted, var(n) is a
Poisson variable so that it is equal to n. As this need not be so, Burnham et al. (1980)
suggest that a sounder approach is to undertake a number of replicate transects, calculate D
for each and then find the variance of these replicates.
Buckland et al. (1993) argue that extreme observations of perpendicular distance should be
removed from the data set prior to calculating population density. As a rough rule, they
suggest that 5% of the data should be truncated.
The Fourier series method can also be applied to grouped data. Such data are generated
when it is impossible to assign observations to accurately measured distances so they are
allocated to distance bands such as 05, 510,1015 and 1520 metres. It may also be
advantageous to group data when bias in favour of certain distances is suspected. A
description of these methods is beyond the scope of the present text. If possible grouping
data should be avoided as it reduces accuracy (Southwell & Weaver, 1993).
From the above Fourier model it can be seen that the key computational issue is the
estimation of g(x) or f(0).
2. Point transects
Instead of traversing a transect the observer may move to a number of fixed points and
record the distance, r, to individual animals. These methods are almost only used for bird
surveys where the patchy suitability of the habitat to the birds may make line transects
inappropriate because they cut across a number of habitat types. Point transects are often
easier to undertake because the observer needs only estimate distance and markers may be
placed in advance to aid the estimation of distance. Population density
is given by:
Where n is the number of animals observed, k the number of point transects undertaken and
h(0) the slope of the probability density function of detection distances evaluated at zero
distance. In similar fashion to what was described for the line transect method the central
problem is to estimate h(0).
For the case of a half normal detection function the maximum likelihood estimator for density
has the particularly simple form:
.
( )
]
]
]
− +
−
·
− + j k j k j k j k
a a a a
w n
a a
*
1
1
1
) ˆ , ˆ cov(
∑ ∑
· ·
·
,
`
.

m
j
m
k
k j
a a
a
1 1
) ˆ , ˆ cov(
ˆ
1
var
( )
( )
( )
]
]
]
]
]
+ ·
2
2
ˆ
1
ˆ
1
var
) var(
ˆ ˆ
var
a
a
n
n
D D
k
h n
D
π 2
) 0 (
ˆ
ˆ
·
∑
·
·
n
i
i
r k
n
D
1
2
2
2
ˆ
π
Pisces Conservation Ltd 10
B. Fourier model for full distance line transect data
To apply this method you will require data of either the perpendicular distances from the
transect line to the objects or the distance to the objects plus their angle from the line.
If you have perpendicular distances for each transect then select Transect, Perpendicular
Distance from the drop down menu on the data form. If you wish to enter distance and angle
select Transect, Distance, angle. All the calculations are undertaken using the perpendicular
distance which is calculated by DfD when distance/angle data is supplied. The distance
observations for each transect are entered in columns and additional transects are added
using the Add New Transect button on the data form. The length of each transect must be
entered at the head of the first column for each transect and the units of measurement
selected from the dropdown menu. The units of measurement for both the distances to
objects and the length of the transects must be in the same.
To obtain a density estimate either click on the F button at the top of the window or click on
the Select Model tab and click on the Fourier model button
1. Viewing the histogram of distances
Select the Histogram tab to view the distribution of distances observed. When multiple
transects have been entered into the data grid the transect to display is selected from the
Plot transect dropdown menu. The number of bins (frequency classes) for the distance data
is selected using the Select bin number dropdown. If a line transect model has been run
then the detection function with be plotted on the graph as a green line.
2. Truncating the data
It is generally advisable to truncate distance data to remove some of the largest distances
observed. Click on the truncation tab to view the histogram selected in the Histogram tab.
The percentage of observations to remove is then selected using the radio buttons at the top
of the graph. To select a particular cut level select Custom and use the edit box to select the
Pisces Conservation Ltd 11
percentage required. The observations that will be omitted from the calculation are shown in
green.
Once a cut level has been applied this will be used for future calculations when a model is
selected. Truncation is not applied to the half normal point transect model.
3. Viewing the results
Once an analysis has been undertaken the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects. See Setting and Changing Units above
to change the units of measurement.
The parameters calculated for the Fourier model are viewed using the Model Parameters
tab. The parameters for each transect are presented on separate rows. Column 2 gives the
estimate for f(0) which is used to calculate the estimated density. In the next column is given
the number of terms of the Fourier series that were used for the calculation. DfD determines
automatically using a standard stopping rule the appropriate number of terms. The first six
terms for the Fourier series are given in columns 4 to 9. If only 2 terms were used for the
calculation then only a(0) and a(1) would be used to form the detection function.
Pisces Conservation Ltd 12
A plot of the detection function used for the calculation can be viewed by selecting the
Histogram tab.
C. Halfnormal model for full distance line transect data
To apply this method you will require data of either the perpendicular distances from the
transect line to the objects or the distance to the objects plus their angle from the line.
If you have perpendicular distances for each transect then select Transect, Perpendicular
Distance from the drop down menu on the data form. If you wish to enter distance and angle
select Transect, Distance, angle. All the calculations are undertaken using the perpendicular
distance which is calculated by DfD when distance/angle data is supplied. The distance
observations for each transect are entered in columns and additional transects are added
using the Add New Transect button on the data form. The length of each transect must be
entered at the head of the first column for each transect and the units of measurement
selected from the dropdown menu. The units of measurement for both the distances to
objects and the length of the transects must be in the same.
To obtain a density estimate either click on the N button at the top of the window or click on
the Select Model tab and click on the Half normal button
1. Viewing the histogram of distances
See Section B.
2. Truncating the data
See Section B.
3. Viewing the results
Once an analysis has been undertaken the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
The parameters calculated for the model are viewed using the Model Parameters tab.
A plot of the detection function used for the calculation can be viewed by selecting the
Histogram tab.
D. All objects detected during line transect surveys
This option is included so that you can see what the density estimate would be if all the
objects were detected. The program assumes that the counts are for a strip transect with a
width equal to twice the maximum perpendicular distance observed. If truncation is applied
then the maximum distance for the truncated data set is used.
To obtain a density estimate either click on the All button at the top of the window or click on
the Select Model tab and click on the All objects detected button.
Once an analysis has been undertaken the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
Pisces Conservation Ltd 13
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
E. Fourier model for nbelt line transect data
To apply this method you will require data of the number of objects counted in defined belts
away from the transect line. The observations for each transect are entered in pairs of
columns. In the first column is entered the outer limit of each belt and in the second the
number of objects observed within this belt. The length of each transect must be entered at
the head of the first column for each transect and the units of measurement selected from the
dropdown menu. The units of measurement for both the distances to objects and the length
of the transects must be in the same.
To obtain a density estimate either click on the F button at the top of the window or click on
the Select Model tab and click on the Fourier model button
1. Viewing the histogram of distances
See Section B.
2. Viewing the results
Once an analysis has been undertaken the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
The parameters calculated for the Fourier model are viewed using the Model Parameters
tab. The parameters for each transect are presented on separate rows. Column 2 gives the
estimate for f(0) which is used to calculate the estimated density. In the next column is given
the number of terms of the Fourier series that were used for the calculation. DfD determines
automatically using a standard stopping rule the appropriate number of terms. The first six
terms for the Fourier series are given in columns 4 to 9. If only 2 terms were used for the
calculation then only a(0) and a(1) would be used to form the detection function.
A plot of the detection function used for the calculation can be viewed by selecting the
Histogram tab.
3. Goodness of fit
The goodness of fit tab shows the observed and expected frequencies in each belt and a chi
squared test for the goodness of fit. The expected values are calculated using the calculated
Fourier parameters.
F. Linear detection model for 2belt line transect data
Simple two belt models are often used for animals such as birds where it is difficult to
measure distance accurately. The counts simply comprise the numbers observed within and
beyond a set distance from the transect line.
The density of the objects is calculated as
where w is the distance to the outer edge of the inner belt in metres, N the total objects
observed, N
1
the number of objects within the inner belt and L the length of the transect in
kilometres.
wL
N
N
N D
,
`
.

− −
·
1
1 1
10
Pisces Conservation Ltd 14
1. Viewing the results
Once an analysis has been undertaken, the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
G. Exponential detection model for 2belt line transect data
Simple two belt models are often used for animals such as birds where it is difficult to
measure distance accurately. The counts simply comprise the numbers observed within and
beyond a set distance from the transect line. The density of the objects is calculated as
where w is the distance to the outer edge of the inner belt in metres, N the total objects
observed, N
1
the number of objects within the inner belt and L the length of the transect in
kilometres.
1. Viewing the results
Once an analysis has been undertaken, the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
H. Point transect  2belt model
Point transects are most frequently used for bird surveys where it is usually impossible to
accurately record the distance to each sighting and line transect methods are inappropriate
because the birds are localised into particular parts of the habitat. Given these limitations a
simple two belt method is appropriate. The density of the objects is calculated as
where N is the total number of objects counted and N
2
the number beyond the fixed radius r.
1. Viewing the results
Once an analysis has been undertaken, the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
I. Point transect  half normal model
This method can be used when point transects have been used to measure the distances to
each object detected. The detection function is modelled as a half normal function and the
density is estimated using:
where n is the number of objects observed and d
i
the distance to the ith object.
2
2
ln
r
N
N
N
D
π
,
`
.

·
wL
N
N
N D
,
`
.

,
`
.

− −
·
1
1 ln
5
∑
·
·
n
i
i
d
n
D
1
2
2
π
Pisces Conservation Ltd 15
1. Viewing the results
Once an analysis has been undertaken, the density estimates for each transect are viewed
using the Density Estimates tab. The results for each transect are arranged in a grid. In
addition to the density the grid gives the variance and the standard error of the estimates
and gives the average density for all the transects.
IV. Plotless density estimators
These methods are used to estimate the density of stationary objects in the field. Although
most obviously applicable to plants they can also be used for many static or slow moving
animals or their signs such as ant hills and nests. They are especially useful in situations
where the use of quadrate sampling would be difficult or too costly. However, while the
accuracy of the density estimate obtained by quadrate sampling is relatively insensitive to
the spatial distribution of the object under study this is not the case with plotless methods.
Unfortunately, there is no clearly superior plotless density estimator as the methods differ in
their ability to handle different types of spatial distribution. In practice, it is important to select
the method that offers the easiest and quickest field measurements commensurate with the
accuracy required. To help you with this task DfD offers a data simulation facility. Methods
that are more robust could be developed, but they would be too costly to use. Engeman et al
1994 reviewed the quality of the density estimates obtained using the various estimators
available and DfD implements all of the less costly methods that they found best. However,
their study used simulated data and the situation in the field can be even more adverse than
even they assumed. There can be no substitute for a pilot study in which a known density of
the objects of interest is estimated by a variety of plotless estimates to determine the most
costeffective method.
The following points should be considered when considering a survey using plotless methods.
1. You must have some idea of the effort required to collect measurements in the field.
2. You should have some idea of the spatial distribution of the objects under study  can
they be assumed to be random, regularly distributed, clumped or highly clumped?
3. If the distribution is clumped (as is often the case) then the best methods offered by DfD
are Angleordersecond nearestindividual, ordereddistancethird nearest individual and
the KendallMoran estimators.
4. The Angleorder estimators are more costly than the other methods to implement so it is
wise to use simulations to decide if ordereddistance or KendalMoran estimators might
be acceptable. Engeman et al 1994 suggest that angleorder methods are unlikely to
reward the user for their increased cost by producing greatly superior estimates.
5. Generally, it seems better to use more sampling points rather than investing a large effort
at a small number of points. This is because the spatial pattern of the study objects can
change over the sampling area. It is therefore often essential to place sufficient sampling
points to give good coverage.
6. If you can assume that the objects are randomly distributed then you can use either
closestindividual or nearestneighbour estimators which are the simplest and least
costly of methods to use. If these "basic distance" methods are to be applied tests of
randomness must be carried out. For populations sampled in quadrates, with a mean
greater than one, the Poisson Index of Dispersion is a satisfactory test of randomness.
Seber (1982) gives other test procedures.
A. Closest individual (Cottam et al., 1953)
To use this estimator you will require a series of measurements of the distance from a
randomly selected point to the closest individual. For each study area, the distances are
entered as a single column.
Pisces Conservation Ltd 16
Density, D, is calculated using:
where R
i
is the distance from the ith random point to its closest individual and N the number
of random sampling points.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
B. Nearest neighbour (Cottam & Curtis, 1956)
To use this estimator you will require a series of measurements of the distance from a
randomly selected individual to its closest neighbour. The problem is that it is not really
possible to randomly select the individuals to measure from in the field. For each study area,
the distances are entered as a single column.
Density, D, is calculated using:
where H
i
is the distance from the ith individual to its closest neighbour and N the number of
measurements made.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
C. KendallMoran (Kendall and Moran, 1963)
For this method you will require a series of pairs of measurements of the distance from a
randomly selected point to the closest individual and from this individual to its nearest
neighbour. For each study area, the distances are entered as over two columns. The first
column holds the distance to the closest individual and the second column the distance to its
nearest neighbour. The program calculates the area searched which is the sum of the
circular areas within the radii of the closest individual and its nearest neighbour minus the
overlap.
Density, D, is calculated using:
where P is the number of closest individuals, N the number of nearest neighbours and B
i
the
combined search area from the ith sampling point for the closest individual and its nearest
neighbour.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
,
`
.

]
]
]
·
2
4
1
N
R
D
i
,
`
.

]
]
]
·
2
4
1
N
H
D
i
∑
− +
·
i
B
N P
D
1
Pisces Conservation Ltd 17
D. Ordered distance closest individual (Morista, 1957)
To use this estimator you will require a series of measurements of the distance from a
randomly selected point to the closest individual. For each study area, the distances are
entered as a single column.
Density,D, is calculated using:
where R
i
is the distance from the ith random point to its closest individual and N the number
of random sample point.
Seber (1982) as gives the variance of D:
.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
E. Ordered distance second closest individual (Morista, 1957)
To use this estimator you will require a series of measurements of the distance from a
randomly selected point to the second closest individual. For each study area, the distances
are entered as a single column.
Density, D, is calculated using:
where R
I
is the distance from the ith random point to its closest individual and N the number
of random sample point. Seber (1982) as gives the variance of D as:
.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
F. Ordered distance third closest individual (Morista, 1957)
To use this estimator you will require a series of measurements of the distance from a
randomly selected point to the third closest individual. For each study area, the distances are
entered as a single column. Density, D, is calculated using:
where R
i
is the distance from the ith random point to its closest individual and N the number
of random sample point.
Seber (1982) as gives the variance of D as:
( )
( )
∑
−
·
2
1
i
R
N
D
π
( )
( )
∑
−
·
2
1 2
i
R
N
D
π
( )
( )
∑
−
·
2
1 3
i
R
N
D
π
( ) 2
ˆ
]
ˆ
var[
2
−
·
N
D
D
( ) 2 2
ˆ
]
ˆ
var[
2
−
·
N
D
D
Pisces Conservation Ltd 18
.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
G. Angleorder nearest individual (Stearns, 1949)
This is a 'pointcentredquarter' method. For each randomly selected point the area around it
is divided into equal quarters and the distance to the nearest individual in each quarter
recorded. Density, D, is calculated using:
where R
ij
is the distance from the ith random point to its closest individual in the jth quarter
and N the number of random sample points.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
H. Angleorder second nearest individual (Morista, 1957)
This is a 'pointcentredquarter' method. For each randomly selected point the area around it
is divided into equal quarters and the distance to the nearest individual in each quarter
recorded. Density, D, is calculated using:
where R
ij
is the distance from the ith random point to the second closest individual in the jth
quarter and N the number of random sample points.
DfD can simulate data collected using this method so that you can appreciate the likely
errors.
I. Data Simulation
Select Simulation and either Transect Data or Plotless Density Estimate from the top menu
to simulate transect sampling observations or plotless density estimation protocols. In both
cases, following selection, a dialogue window opens in which simulation parameters are
selected. These simulators can be used to explore the efficiency of the different methods
under a variety of assumptions. Simulated data will be added to data grid and can be viewed
and saved in the normal fashion.
1. Plotless density estimation
The dialogue for this simulator is shown below. Within the Method panel on the righthand
side select the type of data that you wish to simulate.
∑
·
2
/ 1
12
ij
R
D
π
∑
·
2
/ 1
28
ij
R
N
D
π
( ) 2 3
ˆ
]
ˆ
var[
2
−
·
N
D
D
Pisces Conservation Ltd 19
The distribution of the
objects that will be
sampled is selected
within the Distribution
Type panel. This is
useful as methods
differ in their ability to
handle nonrandom
data. Select Random
to position the objects
at random with the
study area. This is
simulated by using a
random number
generator to give x
and y coordinates.
Select Regular to
arrange the objects in
a regular grid pattern
and Clumped to
produce a contiguous
distribution. If Regular
is selected then it is possible to select the degree of regularity by the numerical value placed
in the Noise edit box. A value of 0 (zero) will be perfectly regular: as the value increases a
potentially larger random displacement from the regular distribution is applied to each object.
If clumped is selected the Number of Clumps can be altered to give the number of clusters of
objects within the sample area. The objects are randomly placed within each cluster and the
Tightness parameter can take any value from 0. The smaller the value the tighter the
clusters. Values for tightness above 1 will produce ever more randomlike distributions.
Place in the Density edit box the required density of the objects in the study area. The default
value of 1 is good starting point as it produces a reasonable number of objects.
The value entered into the Number of Samples edit box gives the number of samples that will
be simulated. For example, if closest individual is selected this will select the number of
distances measured from random points to their nearest object.
The length and width of the sampling area is chosen using the Dimension edit box. For most
studies a 10 by 10 area, which is the default, will be found suitable.
If Random Seed is set to 0 (zero) the program selects the random number seed using the
date/time function of the computer. Thus, every simulation will be different. To be able to
generate exactly the same simulation a number of times enter a number in Random Seed
and remember it!
Once the simulation parameters have been entered select Ok to generate the simulated data
or Cancel to leave without data generation.
The distribution of the objects and the results of the simulation are shown graphically (next
page).
Pisces Conservation Ltd 20
The individual objects are shown as red squares, the selected point or individual as a yellow
cross and the distance calculated by the program as a green line.
2. Line transect simulation
The dialogue for this simulator for both point and line transect data is shown below. Within
the Method panel on the lefthand side select the type of data that you wish to simulate.
The detection function
(the decline in the
likelihood of an object
being spotted with
distance) is selected
from the Detection
Function panel.
Depending on the
function chosen other
parameters may have
to be entered in the
parameter list on the
righthand side.
Actual density is the
density of the objects
in the simulated study
area. Any real,
positive, number can be chosen. However, large numbers may result in an error if the number
of objects detected becomes too large. The density is assumed to be measured in objects
per square metre.
Transect length is the length of the line transect in metres. Any positive integer can be used
although large numbers may result in an overload error if the number of objects detected
becomes too large.
Maximum detection distance is the maximum distance at which an object will be detected in
metres. It is used by the simulator to fix the scale of the sighting experiment you are
envisaging.
Inner belt distance is the distance to the outside limit of the inner belt in metres. This
parameter is required when simulating a 2belt line or point transect.
Pisces Conservation Ltd 21
Std. dev. of detection function is the standard deviation used for a halfnormal detection
function. The larger the value the more objects will be detected at extreme distances. This
should be a real, positive value.
Decay constant is the negative exponential detection function parameter. It is a real, positive
number. The larger the value the more rapidly the probability of detection declines with
distance.
For both point and line transects the results of the simulations are shown graphically. For line
transects the detected objects are shown as red squares and the perpendicular distance as
green lines. Undetected objects are shown as blue squares. For point transects detected
objects and the sighting distance is shown in red or green and undetected objects in blue.
For two belt surveys the colour of the detected object and the sighting line is green if the
point is within the inner belt and red if it is the outer belt.
V. Important references
Bibby, C. J., Burgess, N. D. & Hill, D. A. (1992). Bird census techniques. Academic Press, San
Diego. 257 pp.
Blackith, R. E. (1958). Nearestneighbour distance measurements for the estimation of animal
populations. Ecology 39, 14750.
Buckland, S. T. 1987. On the variable circular plot method of estimating animal density. Biometrics
43:363384.
Buckland, S. T., Anderson, D. R., Burnham, K. P. & Laake, J. L. (1993). Distance sampling:
estimating abundance of biological populations. Chapman and Hall, London.
Burnham, K. P. & Anderson, D. R. (1976). Mathematical models for nonparametric inferences from
line transect data. Biometrics 32, 325336.
Burnham, K. P., Anderson, D. R. & Laake, J. L. (1980). Estimation of density from line transect
data. Wildlife Monograph 72.
Byth, K. 1982. On robust distancebased intensity estimators. Biometrics 38:127135.
Clark, P. I. & Evans, F. C. (1954). Distance to nearest neighbor as a measure of spatial relationships
in populations. Ecology 35, 44553.
Clayton, G., & T, F. Cox. 1986. Some robust density estimators for spatial point processes.
Biometrics 42:753767.
Cottam, G. 1947. A point method for making rapid surveys of woodlands. Bulletin of the Ecological
Society of America 28:60.
Cottam, G., & J. T. Curtis. 1956. The use of distance measures in phytosociological sampling.
Ecology 37:451460.
Cottam, G., J. T. Curtis, & A. J. Catana. 1957. Some sampling characteristics of a series of
aggregated populations. Ecology 38:610621.
Cottam, G., J. T. Curtis, & B. W. Hale. 1953. Some sampling characteristics of a population of
randomly dispersed individuals. Ecology 34: 741757.
Craig, C. C. (1953). On a method of estimating biological populations in the field. Biometrika 40, 216
18.
Crain, B. R., Burnham, K. P., Anderson, D. R. & Laake, J. L. (1979). Nonparametric estimation of
population density for line transect sampling using Fourier series. Biometrical Journal 21, 731748.
Delince, J. 1986. Robust density estimation through distance measurements. Ecology 67:15761581.
Diggle, P.J. 1975. Robust density estimation using distance methods. Biometrika 62:3948.
Dodd, M. G. & Murphy, T. M. (1995). Accuracy and precision of techniques for counting great blue
heron nests. J. Wildlife Management 59(4), 667673.
Pisces Conservation Ltd 22
Engeman, R. M., Sugihara, R. T., Pank, L. F. & Dusenberry, W. E. (1994). A comparison of plotless
density estimators using Monte Carlo simulation. Ecology 75(6), 17691779.
Ensign, W. E., Angermeier, P. L. & Dolloff, C. A. (1995). Use of line transect methods of estimate
abundance of benthic stream fishes. Can. J. Fish. Aqu. Sci. 52(1), 213222.
Gates, C. E. (1969). Simulation study of estimators for the line transect sampling method. Biometrics
25, 317328.
Gates, C. E., Marshall, W. H. & Olson, D. P. (1968). Line transect method of estimating grouse
population densities. Biometrics 24, 13545.
Kelley, J. R., Jr. (1996). Linetransect sampling for estimating breeding wood duck density in forested
wetlands. Wildlife Soc. Bull. 24(1), 3236.
Kendall, M. G. & P. A. P. Moran, 1963. Geometrical probability. Griffin, London, England.
Keuls, M., Over, H. I. & De Wit, C. T. (1963). The distance method for estimating densities. Statistica
Neerlandica 17, 7191.
Kovner, J. L. & Patil, S. A. (1974). Properties of estimators of wildlife population density for the line
transect method. Biometrics 30, 225230.
Lewis, S. M. 1975. Robust estimation of density for a twodimensional point process. Biometrika
62:519521.
Morisita. M. 1954. Estimation of population density by spacing method. Memoirs of the Faculty of
Science Kyushu University, Series E, Biology 1:187197.
Morisita. M. 1957. A new method for the estimation of density by spacing method applicable to non
randomly distributed populations. Physiology and Ecology 7:134144 (in Japanese). [Available as
Forest Service translation number 11116, USDA Forest Service, Washington, D.C., USA.]
Parker, K. R. 1979. Density estimation by variable area transect. Journal of Wildlife Management
43:484492.
Patil, S. A., K. P. Burnham & J. L. Kovner, 1979. Nonparametric estimation of plant density by the
distance method. Biometrics 35:597604.
Patil, S. A., J. L. Kovner, & K P. Burnham, 1982. Optimum nonparametric estimation of population
density based on ordered distances. Biometrics 38:243248.
Persson, O. 1971. The robustness of estimating density by distance measurements. Pages 175187
in G. P. Patil, E. C. Piciou, and W. E. Waters. editors. Statistical ecology. Volume 2. Pennsylvania State
University Press, University Park, Pennsylvania, USA.
Pielou, E. C. (1969). An Introduction to Mathematical ecology. Wiley_Interscience, New York and
London.
Pollard, S. H. 1971. On distance estimators of density in randomly distributed forests. Biometrics
27:9911002.
Seber, G. A. F. (1982). The estimation of animal abundance and related parameters. Griffin,
London.
Skellam, J. G. (1958). The mathematical foundations underlying the use of line transects in animal
ecology. Biometrics 14, 385400.
Sen, A. R., Tourigny, I. & Smith, G. E. J. (1974). On the line transect sampling method. Biometrics
30, 329340.
Southwell, C. & Weaver, K. (1993). Evaluation of analytical procedures for density estimation from
linetransect data: Data grouping, data truncation and the unit of analysis. Wildlife Research 20(4), 433
444.
Starr, R. M., Fox, D. S., Hixon, M. A., Tissot, B. N., Johnson, G. E. & Barss, W. H. (1996).
Comparison of submersiblesurvey and hydroacousticsurvey estimates of fish density on a rocky bank.
U S National Marine Fisheries Service Fish. Bull. 94(1), 113123.
Stearns, F. W. 1949. Ninety years' change in a northern hardwood forest in Wisconsin. Ecology
30:350358.
Pisces Conservation Ltd 23
Symmons, P. M., Dean, G. J. W. & Stortenbere, R. C. W. (1963). The assessment of the size of
populations of adults of the red locust. Nomadacris septemfasciata (Serville), in an outbreak area.
Bull. ent. Res. 54, 549569.
Thompson, H. R. (1956). Distribution of distance to nth neighbour in a population of randomly
distributed individuals. Ecology 37, 391394.
Turner, F. B. (1960). Size and dispersion of a Louisiana population of the cricket frog, Acris gryllus.
Ecology 41, 258268.
Waloff, N. & Blackith, R. E. (1962). The growth and distribution of the mounds of Lasius flavus
(Fabricius) (Hym.: Formicidae) in Silwood Park, Berkshire. J. Anim. Ecol. 31, 421437.
Yapp, W. B. (1956). The theory of line transects. Bird Study 3, 93104.
PISCES LICENCE AGREEMENT This is a legal agreement between you the end user and PISCES Conservation Ltd. Lymington (PISCES). BY OPENING THIS PACKAGE YOU ARE AGREEING TO BE BOUND BY THE TERMS OF THIS AGREEMENT. IF YOU DO NOT AGREE TO THE TERMS OF THIS AGREEMENT PROMPTLY RETURN THE UNOPENED PACKAGE AND ALL ACCOMPANYING ITEMS (including written material ) TO THE PLACE YOU OBTAINED THEM FOR A FULL REFUND. 1. GRANT OF LICENSE  This PISCES License Agreement (‘License’) permits you to use one copy of the PISCES software product acquired with this License (SOFTWARE ) on any single computer, provided the SOFTWARE is in use on only one computer at any time. If you have multiple Licenses for the SOFTWARE then at anytime you may have as many copies of the SOFTWARE in use as you have Licenses. The SOFTWARE is ‘in use’ on a computer when it is loaded into the temporary memory (i.e. RAM) or installed into the permanent memory (e.g. hard disk, CD ROM. or other storage device) of that computer, except that a copy installed on a network server for the sole purpose of distribution to other computers is not ‘in use’. If the anticipated number of users of the SOFTWARE will exceed the number of applicable Licenses then you must have a reasonable mechanism or process in place to assure that the number of persons using the SOFTWARE concurrently does not exceed the number of Licenses. If the SOFTWARE is permanently installed on the hard disk or other storage device of a computer (other than network server) and one person uses that computer more than 80% of the time it is in use then that person may also use the SOFTWARE on a portable or home computer. 2. COPYRIGHT  The SOFTWARE is owned by PISCES or its suppliers and is protected by all applicable national laws. Therefore, you must treat the SOFTWARE like any other copyrighted material (e.g. a book) except that if the software is not copy protected you may either (a) make one copy of the of the SOFTWARE solely for backup or archival purposes, or (b) transfer the SOFTWARE to a single hard disk provided you keep the original solely for backup or archival purpose. You may not copy the Product manual(s) or written materials accompanying the SOFTWARE. 3. DUAL MEDIA SOFTWARE  If the SOFTWARE package contains both 31/2" and 51/4" disks, then you may use only one set (either the 31/2" or 51/4') of the disks provided. you may not use the other disks on another computer or computer network, or lend, rent, lease, or transfer them to another user except as part of a transfer or other use expressly permitted by this PISCES License Agreement. 4. OTHER RESTRICTIONS  You may not rent or lease the SOFTWARE, but you may transfer your rights under this PISCES License Agreement on a permanent basis provided you transfer all copies of the SOFTWARE and all written materials, and the recipient agrees to the terms of this Agreement. You may not reverse engineer, decompile or disassemble the SOFTWARE. Any transfer must include the most recent update and all prior versions. LIMITED WARRANTY LIMITED WARRANTY  PISCES warrants that (a) the SOFTWARE will perform substantially in accordance with the accompanying Product Manual(s) for a period of 90 days from the date of receipt; and (b) any PISCES supplied hardware accompanying the SOFTWARE will be free from defects in materials and workmanship under nominal use and service for a period of one year from the date of receipt. Any implied warranties on the SOFTWARE and hardware are limited to 90 days and one (1) year respectively or the shortest period permitted by applicable law, whichever is greater. CUSTOMER REMEDIES  PISCES'S entire liability and your exclusive remedy shall be, at PISCES option, either (a) return of the price paid or (b) repair or replacement of the SOFTWARE or hardware that does not meet PISCES’S Limited Warranty, and which is returned to PISCES with a copy of your receipt. This Limited Warrant is void if failure of the SOFTWARE or hardware has resulted from accident, abuse or misapplication. Any replacement SOFTWARE will be warranted for the remainder of the original warranty period or 30 days, whichever is longer. NO OTHER WARRANTIES  TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, PISCES DISCLAIMS ALL OTHER WARRANTIES. EITHER EXPRESS OR IMPLIED, INCLUDING BUT LIMITED NOT TO IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, WITH RESPECT TO THE SOFTWARE, THE ACCOMPANYING PRODUCT MANUAL(S) AND WRITTEN MATERIALS, AND ANY ACCOMPANYING HARDWARE. THE LIMITED WARRANTY CONTAINED HEREIN GIVES YOU SPECIFIC LEGAL RIGHTS. NO LIABILITY FOR CONSEQUENTIAL DAMAGES  TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW PISCES AND ITS SUPPLIERS SHALL NOT BE LIABLE FOR ANY OTHER DAMAGES WHATSOEVER (INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS OF BUSINESS PROFITS, BUSINESS INTERRUPTION, LOSS OF BUSINESS INFORMATION, OR OTHER PECUNIARY LOSS) ARISING OUT OF THE USE OF OR INABILITY TO USE THIS PISCES PRODUCT, EVEN IF PISCES HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN ANY CASE, PISCES’S ENTIRE LIABILITY UNDER ANY PROVISION OF THIS AGREEMENT SHALL BE LIMITED TO THE AMOUNT ACTUALLY PAID BY YOU FOR THE SOFTWARE. This Agreement is governed by the laws of England. Should you have any questions concerning this Agreement, or if you desire to contact PISCES for any reason, please use the address information enclosed in this product to contact PISCES or write: PISCES Conservation Ltd IRC House, The Square Pennington LYMINGTON Hampshire England SO41 8GN
................................ D.......................... 12 Truncating the data .......................................................................................................................................... 3.................................... 5 PRINTING AND EXPORTING YOUR RESULTS ...Contents I... 2............. G...................................................................... B.................................................. 2.............................. B............ 9 FOURIER MODEL FOR FULL DISTANCE LINE TRANSECT DATA ......................................................... L.......... A.................................................................. G. 10 Viewing the histogram of distances................. 2 DATA ENTRY... K......... 6 POINT AND LINE SURVEY THEORY......................................................................... 14 ............................................................................... 1 INSTALLATION AND GENERAL INSTRUCTIONS .................. 3.................. 12 FOURIER MODEL FOR NBELT LINE TRANSECT DATA...................................................................................... 3..................................................... 1......................................................... 13 Viewing the results ..................................................... 6 LINE TRANSECT METHODS ................................... 5 ZOOMING AND PANNING ON GRAPHS ............................. 13 Viewing the histogram of distances..................................................... 1 SYSTEM REQUIREMENTS .................................. 6 Line transect methods: the Fourier series estimator......................................................................................... C............... 5 EDITING EXISTING DATA .. 13 Goodness of fit.................................. I........ II........................................................................................................... 1 GENERAL INSTRUCTIONS ............................... 11 HALFNORMAL MODEL FOR FULL DISTANCE LINE TRANSECT DATA ................................................. J.............................................. 14 EXPONENTIAL DETECTION MODEL FOR 2BELT LINE TRANSECT DATA ............................................................................................................... 2 DEMONSTRATION DATA SETS... C................................................................ 12 Viewing the results .............................................................. 13 Viewing the results ........................................................................ 1....................................................................................................... 5 MAXIMUM SIZE OF THE DATA SET ...... 13 LINEAR DETECTION MODEL FOR 2BELT LINE TRANSECT DATA ........................................................................................................................ INTRODUCTION............................ F.................. 12 Viewing the histogram of distances............................................................................ 1 INSTALLATION ...................................................... 1 OBTAINING HELP ....................... E......... 2 SETTING AND CHANGING UNITS ......................... 1.. 14 Viewing the results .............................. 1.............................................................................................................................................................. 12 ALL OBJECTS DETECTED DURING LINE TRANSECT SURVEYS ................. 2 OPENING A DATA SET .............................................. 1................................ 10 Truncating the data .............. III....................................................................... 10 Viewing the results ............................... 6 Point transects ............. 2........... D............................................................ A..... E.............. 2......................... 1............................. F...... H..........................................................
.............. 14 POINT TRANSECT ...................................... 18 DATA SIMULATION .................................. 1.... 1956) .....H........................................................................ 14 Viewing the results ............ 1957) . C...............................2BELT MODEL .................... 16 KENDALLMORAN (KENDALL AND MORAN.................................. 1957) .............................. B.............................. 18 Line transect simulation . I................................................................. G.... 1957) ........................................... 17 ORDERED DISTANCE SECOND CLOSEST INDIVIDUAL (MORISTA................... 20 IMPORTANT REFERENCES...................................... 18 Plotless density estimation.......................................................................... 1............... F....................... 2................... 1..................................... A.... 15 PLOTLESS DENSITY ESTIMATORS ......... 18 ANGLEORDER SECOND NEAREST INDIVIDUAL (MORISTA...................................................................... 1949) ................... 16 ORDERED DISTANCE CLOSEST INDIVIDUAL (MORISTA.................................................................................................... POINT TRANSECT ................................................. IV.................................................. 17 ORDERED DISTANCE THIRD CLOSEST INDIVIDUAL (MORISTA... I.................. D............................ 1957).................................. 17 ANGLEORDER NEAREST INDIVIDUAL (STEARNS...... V.................... 15 NEAREST NEIGHBOUR (COTTAM & CURTIS.............. H........................... E. 1963) ................ 1953)............................ 15 CLOSEST INDIVIDUAL (COTTAM ET AL......HALF NORMAL MODEL..... 21 .... 14 Viewing the results ...................................
About: Details about the program version. General instructions Start DfD in the normal Windows fashion either by clicking on the program icon or from the start button. The simulated data can be the basis of a classroom practical of the merits of the various methods.I. and doubleclick Setup. Edit: The standard Windows menu. The installation program should start automatically. Place the DfD CD in the CD drive. A PC running Windows® 95 or later operating system. File: To open. For example. They are particularly appropriate for the estimation of population density for plants or large animals living at low density in difficult to traverse habitat. These work in the same way as standard Windows programs. The animals are counted while walking along specially cut forest trails. 2. II. Installation 1. export and save data sets. A. Pisces Conservation Ltd 1 . When installation is complete there will be a DfD entry on your Start: Programs menu and a folder. Help: to enter the Help system. DfD has been designed to be easy to use and is particularly suitable for ecological teaching because it allows students to quickly enter or simulate data and explore a range of methods within a familiar Windows setting. Minimum of 2 MB spare hard disk space. If the CD does not autoplay. initiate. The survey methods used in Density from Distances aim to estimate density using observations on the distance between objects or from a selected line or point to the animals or plants and thus do not require the worker to accurately map out or define the sampling area. Options: Allows the setting of defaults. 3. An important feature of the program is the range of simulations that can be undertaken. distance sampling using a Fourier series is frequently the method of choice for estimating population size of primates in neotropical forest. C. Installation and General instructions System requirements 1. follow the onscreen instructions. browse the CD in Windows Explorer or My Computer. These are useful for both students who wish to learn about the various methods and researchers who need to appreciate the level of accuracy they are likely to obtain. Simulations: To simulate plotless and transect data. The instruction manual (in Adobe Acrobat PDF format) is also available from the Start: Programs: DfD menu.exe in the root directory. Researchers in other fields may also find these methods useful for estimating density. 2. Introduction DfD (Density from Distances) is a Windows® program that offers a range of analytical techniques commonly used by ecologists to estimate animal and plant density. print. An uninstall program will also be created to remove the program if you wish. window views and units of measurement. B. Distance methods are particularly appropriate for static objects such as plants and nests. The high density of the forest and the low density and mobility of the monkeys would make it futile to try to census a number of predefined quadrates. Along the top bar are a number of pulldown menus. For bird surveys the difficulty of accurately estimating distance results in the need for simple point or line transect methods where distance is divided into two or more categories. C:\Program Files\DfD on your hard disk that holds the program files. Such methods are available in DfD.
csv as the data is stored in simple comma delimited form.00 to 17.When the program is started. Transect. See Setting and Changing Units to select the units used to measure distances. Data entry Data is entered in the Data window that is opened by clicking on the Data tab. Help and frequently asked questions are available on our web site at http://www. Opening a data set Use FileOpen to start the file dialogue to select an existing data file for analysis.csv A closest neighbour data set from a population with a density of 1 per square metre. To create a new data set select File New from the pulldown menu. (1980). To enter new data select the type of data you wish to enter from the dropdown menu above the data grid and fill in the grid with appropriate data.csv An example of a line transect with distance belts from page 68 of Burnham et al. Demonstration data sets DfD is supplied with a number of demonstration data sets as follows: Stakes. Pisces Conservation Ltd 2 . KM. clicking on the help button or selecting the help drop down menu.csv The stakes demonstration data set for a line transect from page 62 of Burnham et al (1980).demon.csv A data set of distances from a random point to its closest object and from this object to its nearest neighbour for a population with a density of 1 object per square metre. The data grid that will be formed is shown below. E.00).com If you have problems using the program or entering data which you cannot solve then contact Pisces Conservation Ltd by email pisces@irchouse. From the File menu choose Open to load an existing data set and New to start entering new data.csv Multiple point data for a simulated data set with an actual density of 0. D. Because the different methods require different data structures you must then select the method you want to apply from the drop down menu in the top panel. F.co. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. The data simulator will automatically organise the data correctly. p68_belt. To see the type of data that is required run simulations from the Simulation dropdown menu and view the data grid. Closest. Obtaining help For most active windows context sensitive help can be obtained by pressing F1. you will be presented with a blank data grid. Multi_point. G.piscesconservation. Perpendicular Distance Select this option if you want to estimate density for line transect data and you have a series of measurements of perpendicular distance from the transect line. The default file extension is . The different options and the data they require are described below.uk or by phone to England 44 (0)1590 676622 during office hours (09.05 objects per square metre.
To add another transect click on the Add New Transect button. Transect. To add another transect click on the Add New Transect button. All the data for a transect can be removed by double clicking its title cell and selecting Delete Column. To enter data click on a cell and type. The data grid that will be formed is shown below. To add another transect click on the Add New Transect button.To enter data click on a cell and type. Point transect Select this option if you want to estimate density for point transect data and you have a series of measurements of the distance to the objects. Cells holding perpendicular distances will be coloured blue. To move down a row press the return key or move the mouse to the cell and click. The data grid that will be formed is shown below. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. Distance and Angle Select this option if you want to estimate density for line transect data and you have a series of measurements of the distance to the objects and their angle from the transect line. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel.belted Select this option if you want to estimate density for line transect data and you have a series of the number of counts of objects in set belts of distance away from the transect line. Cells holding perpendicular distances will be coloured blue. All the data Pisces Conservation Ltd 3 . To move down a row press the return key or move the mouse to the cell and click. This dialogue box also allows the transect title to be changed. In the outer limit column place the distance to the outer edge of each band. To move down a row press the return key or move the mouse to the cell and click. All the data for a transect can be removed by double clicking its title cell and selecting Delete Column. Perpendicular Distance . Cells holding perpendicular distances will be coloured blue. This dialogue box also allows the transect title to be changed. This dialogue box also allows the transect title to be changed. The data grid that will be formed is shown below. To enter data click on a cell and type. Cells holding perpendicular distances will be coloured blue. To add another transect click on the Add New Transect button. To move down a row press the return key or move the mouse to the cell and click. To enter data click on a cell and type. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. All the data for a transect can be removed by double clicking its title cell and selecting Delete Column. Transect.
To enter data. The data grid that will be formed is shown below. The data grid that will be formed is shown below. To enter data click on a cell and type. All the data for a transect can be removed by double clicking its title cell and selecting Delete Column. Nearest Neighbour/Closest Individual Select this option if you want to estimate density using a plotless technique such as nearest neighbour. To move down a row press the return key or move the mouse to the cell and click. Cells holding perpendicular distances will be coloured blue. All the data for a transect can be removed by double Pisces Conservation Ltd 4 . To add another transect click on the Add New Transect button. closest individual or an ordered distance method. Point transect – with belts Select this option if you want to estimate density for point transect data and you have a series of the number of counts of objects in set belts of distance away from the observation point. click on a cell and type. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. To enter data click on a cell and type.for a transect can be removed by double clicking its title cell and selecting Delete Column. To add another transect click on the Add New Transect button. This dialogue box also allows the transect title to be changed. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. Cells holding perpendicular distances will be coloured blue. To move down a row press the return key or move the mouse to the cell and click. All the data for a transect can be removed by double clicking its title cell and selecting Delete Column. To move down a row press the return key or move the mouse to the cell and click. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. This dialogue box also allows the transect title to be changed. To add another transect click on the Add New Transect button. AngleOrder Select this option if you want to estimate density using the angleorder plotless method. Cells holding perpendicular distances will be coloured blue. The data grid that will be formed is shown below. This dialogue box also allows the transect title to be changed.
K.clicking its title cell and selecting Delete Column. The data grid that will be formed is shown below. Maximum size of the data set The maximum number of observations in any column is 500. colour etc. feet. miles and nautical miles. For windows users. Units available are millimetres. You can convert between units at any time even after the density has been calculated. Editing existing data The raw data grid can be edited by using the mouse to click on a cell to select it and typing in a new value. These changes will not alter a saved file until FileSave is undertaken. nautical miles. metres. yards. kilometres. All the data for a transect can be removed by double clicking its title cell and selecting Delete Column. You must also select the units of measurement in the dropdown menu on the righthand side of the top panel. H. This dialogue box also allows the transect title to be changed. The units of measurement for the calculated density estimates are chosen from OptionsSelect density units of measurement. Setting and Changing Units The units of measurement are selected using the Select units dropdown dialogue box in the top panel of the data tab sheet. can be specified. To enter data. margins. Pisces Conservation Ltd 5 . who wish to include a graph in another document. kilometres. centimetres. miles. J. Copies can be made to the Windows clipboard or to a file from FileExport clicking on the Copy to clipboard or Copy to file buttons in the dialogue box. Images of graphs can be stored as bitmaps. metafiles or enhanced metafiles. When print is selected a print preview window is opened in which size. I. This dialogue box also allows the transect title to be changed. The units available are square metres. To add another transect click on the Add New Transect button. Printing and exporting your results Any active window can be printed using FilePrint and data or graphs can be copied to the clipboard using edit copy in the normal Windows fashion. the metafile or enhanced metafile formats are to be preferred as they will allow the image to be changed in size without loss of resolution of the text. Kendal Moran Select this option if you want to estimate density using the KendalMoran plotless method. To move down a row press the return key or move the mouse to the cell and click. yards and feet. Cells holding perpendicular distances will be coloured blue. Use FileSave as to save your data under another name. click on a cell and type.
along which the observer moves in a straight line. Zooming and panning on graphs To zoom in on an area. One problem. large grassland mammals. This simple approach is often difficult to undertake for two reasons. whales and large. To pan the graph hold down the right hand mouse button and move the mouse. While it may be possible to count animals from a suitable vantage point or while moving along a transect. 1. for example a bird census may be based on bird song or an electric fish survey on the detection of the distinctive electrical signals. If it is possible to count all of the individuals. To return to the original view. LeResche & Rausch (1974) from a study of bias during aerial surveys of moose Alces alces concluded that this bias was sufficient to invalidate the method as a means of absolute population estimation. firstly. corals and molluscs. not all of the animals present may have been spotted. L. within a known area. An enlarged view of the selected area will be displayed. is simply: ˆ n D= a Counting often requires the observer to move over the census area and thus favours the use of strip transects (long. Distancing sampling methods have been developed to allow for these problems by assuming that the likelihood that an individual will be observed will decline in a mathematically definable way with distance. n. 2w. The methods available within DfD are also useful for sessile or slow moving organisms such as barnacles. Objects on the line are always detected. A. active insects such as butterflies.L. the count can only be converted to a density estimate if the area scanned can be estimated. D. ships and remote operated vehicles. Dodd & Murphy (1995) made a comparison of census methods for bird nest estimation. then this is termed a census and the estimated density. (1993). The counts obtained suffer from error. a. it may not be possible to estimate sufficiently accurately the area scanned and secondly. III. but counts can also be undertaken from motor vehicles. Point and line survey theory Line transect methods: the Fourier series estimator Line transect methods have been developed for situations when it is not possible to count all the animals within a strip transect. A survey of 17 studies on large mammals by Caughly (1974) found that the proportion of the population counted varied from 23 to 89%. as can be the case with a visual survey. Pisces Conservation Ltd 6 . The methods are based on the idea that only animals lying on the centre line of the of the strip transect along which the observer moves will be certain to be detected and that the probability of detection will fall with perpendicular distance from this line. and width. The techniques presented here have been reviewed in detail by Buckland et al. which varies with the counting rate and bias because of the tendency of observers to undercount. They can be applied to data collected nonvisually if distance can be estimated. For many animals the observer may walk the transect. move to the top left corner of the area to be enlarged then hold the left hand mouse button and drag to the lower right hand corner and release the button. This is particularly the case for large or easily seen animals such as birds. is to ensure that the same individual is only counted once. Line transect methods Counting the number of sightings forms the basis for estimating density for many animal groups. For these methods it is assumed that: 1. thin quadrats) of length. hold down the lefthand mouse button and move upwards and to the left and release.
P.2. and the sighting angle. The observer does not influence the recorded positions. of the animals present are detected then the equation becomes: ) D= n 2 wLP . which is the angle of the object from the transect line and calculate x rather than measuring x directly. If a proportion. is assumed to equal 1. Line transect methods use the distribution of the perpendicular detection distances to estimate P. and randomly placed in P erpendicular distance the study may be used. θ. For mobile animals. arranged as a Angle regular grid. r. The probability of detecting an object within a strip of area 2wL. which is the distance from the observer to the object. L. 2w are counted then the estimated density is: ) D = n 2 wL . g(0).. but slow movement in relation to the observer creates little inaccuracy. the probability of detecting an object lying on the line. It is assumed that the number of observations will decline with perpendicular distance from the line and a detection function. Given assumption (1) above. P. In practice a number of lines. L. it is Sighting distance usually easier to record the sighting distance. g(x) which describes this reduction is fitted to the data. 1993). and width. 3. If all n objects in a strip of length. The objects are correctly identified The basic field procedure is for the transect route to be a straight line of length. which on substitution gives ˆ D= n ˆ 2L ∫ g ( x )dx w 0 Pisces Conservation Ltd 7 . Further. randomly placed with respect to the animals or objects to be counted and the perpendicular distance to each detected object of interest. Distances and/or angles are measured accurately. recorded. x. the position must be that prior to any response to the presence of the observer: The theory has been developed under the assumption that the objects are immobile. Methods of population estimation based on r and θ were reviewed by Hayes & Buckland (1983) but are not available in DfD as they are considered inferior to those based on the perpendicular distance (Buckland et al. is: P = ∫ w 0 g ( x ) dx w . 4.
(1979) and is a general model that has been shown to give good results for a wide variety of data. n the number of observations and m the number of cosine terms determined by the stopping rule. Density is expressed as numbers per square unit of length where length of the transect and distance to the objects are expressed in the same length units such as metres or kilometres. the probability density function (pdf) evaluated at x = 0 is f (0) = 1 ∫ w 0 g ( x ) dx and thus the general estimator of density is often expressed as: n fˆ ( 0 ) ˆ D = 2L . First calculate the variance and covariances of the parameters ak ˆ var(a k ) = 1 1 2 2 w* a2k + w* − ak n −1 . Experience suggests that m should be < 7. xi the perpendicular distance of the ith animal. without truncation this is the largest perpendicular distance observed. that you choose the first value of m such that 1 2 w* n + 1 1 2 ˆ ≥ a m +1 . When perpendicular distance has been measured as a continuous variable and not grouped into size classes the Fourier series f(0) is given by: m 1 ˆ ˆ f (0) = * + ∑ a k w k =1 . The sampling variance of D requires considerable computation. Having calculated f(o) D is estimated.As it is assumed g(0) =1. where 2 n ) kπ x ak = cos * i * ∑ nw i = 1 w w* is the transect halfwidth. This is the Fourier model of Crain et al. A flexible function with the desired properties to use as a detection function is a cosine series. Pisces Conservation Ltd 8 .
var(n) is a Poisson variable so that it is equal to n. From the above Fourier model it can be seen that the key computational issue is the estimation of g(x) or f(0). (1980) suggest that a sounder approach is to undertake a number of replicate transects. These methods are almost only used for bird surveys where the patchy suitability of the habitat to the birds may make line transects inappropriate because they cut across a number of habitat types. 510.ˆ ˆ cov( a k . Burnham et al. The Fourier series method can also be applied to grouped data. Point transects are often easier to undertake because the observer needs only estimate distance and markers may be placed in advance to aid the estimation of distance. As a rough rule. to individual animals. a j ) = 1 1 w * (a k + j + a k − j ) − a k a j n −1 . For the case of a half normal detection function the maximum likelihood estimator for density has the particularly simple form: ˆ D = 2π k n 2 n ∑ . (1993) argue that extreme observations of perpendicular distance should be removed from the data set prior to calculating population density. Then sum the variancecovariance matrix to give 1 var = ˆ a ∑ ∑ m m j =1 k =1 ˆ ˆ cov( a j . calculate D for each and then find the variance of these replicates. Point transects Instead of traversing a transect the observer may move to a number of fixed points and record the distance. Finally. i =1 ri 2 Pisces Conservation Ltd 9 . 2. It may also be advantageous to group data when bias in favour of certain distances is suspected.1015 and 1520 metres. It is normally assumed that the variance of the number of animals counted. As this need not be so. k the number of point transects undertaken and h(0) the slope of the probability density function of detection distances evaluated at zero distance. a k ) . If possible grouping data should be avoided as it reduces accuracy (Southwell & Weaver. A description of these methods is beyond the scope of the present text. Buckland et al. Population density is given by: ˆ n h (0 ) ˆ D = 2π k Where n is the number of animals observed. r. Such data are generated when it is impossible to assign observations to accurately measured distances so they are allocated to distance bands such as 05. they suggest that 5% of the data should be truncated. In similar fashion to what was described for the line transect method the central problem is to estimate h(0). the estimated variance is calculated using var 1 ˆ a ˆ = D 2 var( n ) + ˆ var D n 1 2 ˆ a ( ) ( ) ( ) . 1993).
To obtain a density estimate either click on the F button at the top of the window or click on the Select Model tab and click on the Fourier model button 1. Fourier model for full distance line transect data To apply this method you will require data of either the perpendicular distances from the transect line to the objects or the distance to the objects plus their angle from the line. When multiple transects have been entered into the data grid the transect to display is selected from the Plot transect dropdown menu. Viewing the histogram of distances Select the Histogram tab to view the distribution of distances observed.B. The number of bins (frequency classes) for the distance data is selected using the Select bin number dropdown. To select a particular cut level select Custom and use the edit box to select the Pisces Conservation Ltd 10 . The percentage of observations to remove is then selected using the radio buttons at the top of the graph. The units of measurement for both the distances to objects and the length of the transects must be in the same. If a line transect model has been run then the detection function with be plotted on the graph as a green line. Perpendicular Distance from the drop down menu on the data form. Distance. If you wish to enter distance and angle select Transect. angle. The length of each transect must be entered at the head of the first column for each transect and the units of measurement selected from the dropdown menu. All the calculations are undertaken using the perpendicular distance which is calculated by DfD when distance/angle data is supplied. Truncating the data It is generally advisable to truncate distance data to remove some of the largest distances observed. Click on the truncation tab to view the histogram selected in the Histogram tab. The distance observations for each transect are entered in columns and additional transects are added using the Add New Transect button on the data form. If you have perpendicular distances for each transect then select Transect. 2.
The parameters calculated for the Fourier model are viewed using the Model Parameters tab. Once a cut level has been applied this will be used for future calculations when a model is selected. DfD determines automatically using a standard stopping rule the appropriate number of terms. Pisces Conservation Ltd 11 . If only 2 terms were used for the calculation then only a(0) and a(1) would be used to form the detection function. 3. In the next column is given the number of terms of the Fourier series that were used for the calculation. See Setting and Changing Units above to change the units of measurement. The observations that will be omitted from the calculation are shown in green. Viewing the results Once an analysis has been undertaken the density estimates for each transect are viewed using the Density Estimates tab. The results for each transect are arranged in a grid. Column 2 gives the estimate for f(0) which is used to calculate the estimated density. The parameters for each transect are presented on separate rows.percentage required. Truncation is not applied to the half normal point transect model. The first six terms for the Fourier series are given in columns 4 to 9. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects.
Perpendicular Distance from the drop down menu on the data form. 2. Truncating the data See Section B. The results for each transect are arranged in a grid. A plot of the detection function used for the calculation can be viewed by selecting the Histogram tab. The results for each transect are arranged in a grid. If you wish to enter distance and angle select Transect. All objects detected during line transect surveys This option is included so that you can see what the density estimate would be if all the objects were detected. Halfnormal model for full distance line transect data To apply this method you will require data of either the perpendicular distances from the transect line to the objects or the distance to the objects plus their angle from the line. C. The program assumes that the counts are for a strip transect with a width equal to twice the maximum perpendicular distance observed. D. The distance observations for each transect are entered in columns and additional transects are added using the Add New Transect button on the data form. 3. In Pisces Conservation Ltd 12 . The length of each transect must be entered at the head of the first column for each transect and the units of measurement selected from the dropdown menu. Distance. To obtain a density estimate either click on the All button at the top of the window or click on the Select Model tab and click on the All objects detected button.A plot of the detection function used for the calculation can be viewed by selecting the Histogram tab. Viewing the results Once an analysis has been undertaken the density estimates for each transect are viewed using the Density Estimates tab. angle. Viewing the histogram of distances See Section B. If truncation is applied then the maximum distance for the truncated data set is used. The units of measurement for both the distances to objects and the length of the transects must be in the same. If you have perpendicular distances for each transect then select Transect. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. Once an analysis has been undertaken the density estimates for each transect are viewed using the Density Estimates tab. To obtain a density estimate either click on the N button at the top of the window or click on the Select Model tab and click on the Half normal button 1. The parameters calculated for the model are viewed using the Model Parameters tab. All the calculations are undertaken using the perpendicular distance which is calculated by DfD when distance/angle data is supplied.
Linear detection model for 2belt line transect data Simple two belt models are often used for animals such as birds where it is difficult to measure distance accurately. The units of measurement for both the distances to objects and the length of the transects must be in the same. Column 2 gives the estimate for f(0) which is used to calculate the estimated density. The first six terms for the Fourier series are given in columns 4 to 9. The counts simply comprise the numbers observed within and beyond a set distance from the transect line. 2. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. The length of each transect must be entered at the head of the first column for each transect and the units of measurement selected from the dropdown menu. The parameters calculated for the Fourier model are viewed using the Model Parameters tab. Viewing the results Once an analysis has been undertaken the density estimates for each transect are viewed using the Density Estimates tab. DfD determines automatically using a standard stopping rule the appropriate number of terms. 3. E. A plot of the detection function used for the calculation can be viewed by selecting the Histogram tab. Goodness of fit The goodness of fit tab shows the observed and expected frequencies in each belt and a chisquared test for the goodness of fit. The expected values are calculated using the calculated Fourier parameters. Fourier model for nbelt line transect data To apply this method you will require data of the number of objects counted in defined belts away from the transect line. N the total objects observed. F.addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. In the next column is given the number of terms of the Fourier series that were used for the calculation. Pisces Conservation Ltd 13 . The observations for each transect are entered in pairs of columns. The density of the objects is calculated as N1 1 − 1 − N D = 10 N wL where w is the distance to the outer edge of the inner belt in metres. Viewing the histogram of distances See Section B. The parameters for each transect are presented on separate rows. To obtain a density estimate either click on the F button at the top of the window or click on the Select Model tab and click on the Fourier model button 1. N1 the number of objects within the inner belt and L the length of the transect in kilometres. The results for each transect are arranged in a grid. If only 2 terms were used for the calculation then only a(0) and a(1) would be used to form the detection function. In the first column is entered the outer limit of each belt and in the second the number of objects observed within this belt.
The detection function is modelled as a half normal function and the density is estimated using: D= n2 π ∑ d i2 i =1 n where n is the number of objects observed and di the distance to the ith object. the density estimates for each transect are viewed using the Density Estimates tab. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. I. The density of the objects is calculated as D = ln N N N 2 πr 2 where N is the total number of objects counted and N2 the number beyond the fixed radius r. the density estimates for each transect are viewed using the Density Estimates tab. G. Pisces Conservation Ltd 14 . The results for each transect are arranged in a grid. The results for each transect are arranged in a grid. N1 the number of objects within the inner belt and L the length of the transect in kilometres. Viewing the results Once an analysis has been undertaken.1. Exponential detection model for 2belt line transect data Simple two belt models are often used for animals such as birds where it is difficult to measure distance accurately. Point transect . The counts simply comprise the numbers observed within and beyond a set distance from the transect line. 1.half normal model This method can be used when point transects have been used to measure the distances to each object detected. N the total objects observed. The density of the objects is calculated as N − ln1 − 1 N D = 5N wL where w is the distance to the outer edge of the inner belt in metres. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. Point transect . Viewing the results Once an analysis has been undertaken. the density estimates for each transect are viewed using the Density Estimates tab. H. The results for each transect are arranged in a grid. Given these limitations a simple two belt method is appropriate. 1.2belt model Point transects are most frequently used for bird surveys where it is usually impossible to accurately record the distance to each sighting and line transect methods are inappropriate because the birds are localised into particular parts of the habitat. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. Viewing the results Once an analysis has been undertaken.
4. If these "basic distance" methods are to be applied tests of randomness must be carried out. Engeman et al 1994 suggest that angleorder methods are unlikely to reward the user for their increased cost by producing greatly superior estimates. You should have some idea of the spatial distribution of the objects under study . Seber (1982) gives other test procedures. This is because the spatial pattern of the study objects can change over the sampling area. However. with a mean greater than one. Plotless density estimators These methods are used to estimate the density of stationary objects in the field.1. It is therefore often essential to place sufficient sampling points to give good coverage. Generally. 1. If the distribution is clumped (as is often the case) then the best methods offered by DfD are Angleordersecond nearestindividual.can they be assumed to be random. To help you with this task DfD offers a data simulation facility. The results for each transect are arranged in a grid. there is no clearly superior plotless density estimator as the methods differ in their ability to handle different types of spatial distribution. However. the distances are entered as a single column. A. In practice. IV. their study used simulated data and the situation in the field can be even more adverse than even they assumed. Methods that are more robust could be developed. You must have some idea of the effort required to collect measurements in the field. Viewing the results Once an analysis has been undertaken. Although most obviously applicable to plants they can also be used for many static or slow moving animals or their signs such as ant hills and nests. Closest individual (Cottam et al. regularly distributed. Engeman et al 1994 reviewed the quality of the density estimates obtained using the various estimators available and DfD implements all of the less costly methods that they found best. For each study area. If you can assume that the objects are randomly distributed then you can use either closestindividual or nearestneighbour estimators which are the simplest and least costly of methods to use. 5. it seems better to use more sampling points rather than investing a large effort at a small number of points. clumped or highly clumped? 3. ordereddistancethird nearest individual and the KendallMoran estimators. while the accuracy of the density estimate obtained by quadrate sampling is relatively insensitive to the spatial distribution of the object under study this is not the case with plotless methods. There can be no substitute for a pilot study in which a known density of the objects of interest is estimated by a variety of plotless estimates to determine the most costeffective method. In addition to the density the grid gives the variance and the standard error of the estimates and gives the average density for all the transects. but they would be too costly to use. The Angleorder estimators are more costly than the other methods to implement so it is wise to use simulations to decide if ordereddistance or KendalMoran estimators might be acceptable. 1953) To use this estimator you will require a series of measurements of the distance from a randomly selected point to the closest individual. the Poisson Index of Dispersion is a satisfactory test of randomness. Unfortunately. it is important to select the method that offers the easiest and quickest field measurements commensurate with the accuracy required. They are especially useful in situations where the use of quadrate sampling would be difficult or too costly. The following points should be considered when considering a survey using plotless methods.. the density estimates for each transect are viewed using the Density Estimates tab. Pisces Conservation Ltd 15 . 2. 6. For populations sampled in quadrates.
1 H i 2 4 N Density.Density. N the number of nearest neighbours and Bi the combined search area from the ith sampling point for the closest individual and its nearest neighbour. DfD can simulate data collected using this method so that you can appreciate the likely errors. D. For each study area. DfD can simulate data collected using this method so that you can appreciate the likely errors. the distances are entered as over two columns. The program calculates the area searched which is the sum of the circular areas within the radii of the closest individual and its nearest neighbour minus the overlap. 1956) To use this estimator you will require a series of measurements of the distance from a randomly selected individual to its closest neighbour. the distances are entered as a single column. Nearest neighbour (Cottam & Curtis. is calculated using: D = P + N −1 ∑B i where P is the number of closest individuals. DfD can simulate data collected using this method so that you can appreciate the likely errors. The first column holds the distance to the closest individual and the second column the distance to its nearest neighbour. is calculated using: D= where Hi is the distance from the ith individual to its closest neighbour and N the number of measurements made. Density. B. The problem is that it is not really possible to randomly select the individuals to measure from in the field. Pisces Conservation Ltd 16 . D. C. For each study area. D. is calculated using: D= 1 Ri 2 4 N where Ri is the distance from the ith random point to its closest individual and N the number of random sampling points. KendallMoran (Kendall and Moran. 1963) For this method you will require a series of pairs of measurements of the distance from a randomly selected point to the closest individual and from this individual to its nearest neighbour.
D. Density. the distances are entered as a single column. Density.D. DfD can simulate data collected using this method so that you can appreciate the likely errors. Ordered distance second closest individual (Morista. Density. 1957) To use this estimator you will require a series of measurements of the distance from a randomly selected point to the third closest individual. 1957) To use this estimator you will require a series of measurements of the distance from a randomly selected point to the second closest individual. For each study area. is calculated using: D = (2 N − 1) 2 π ∑ ( Ri ) where RI is the distance from the ith random point to its closest individual and N the number of random sample point. is calculated using: D = ( N − 1) 2 π ∑ ( Ri ) where Ri is the distance from the ith random point to its closest individual and N the number of random sample point. For each study area. the distances are entered as a single column. 1957) To use this estimator you will require a series of measurements of the distance from a randomly selected point to the closest individual. Ordered distance third closest individual (Morista. D. DfD can simulate data collected using this method so that you can appreciate the likely errors.D. the distances are entered as a single column. Seber (1982) as gives the variance of D as: 2 var[ˆ ] = D ( D ˆ 2N − 2) . Ordered distance closest individual (Morista. E. F. Seber (1982) as gives the variance of D as: Pisces Conservation Ltd 17 . For each study area. is calculated using: D = (3 N − 1) π ∑ ( Ri ) 2 where Ri is the distance from the ith random point to its closest individual and N the number of random sample point. Seber (1982) as gives the variance of D: ˆ ˆ var[ D ] = D 2 (N − 2) .
is calculated using: D = 28 N π ∑ 1 / Rij 2 where Rij is the distance from the ith random point to the second closest individual in the jth quarter and N the number of random sample points. Angleorder second nearest individual (Morista. DfD can simulate data collected using this method so that you can appreciate the likely errors. G. Simulated data will be added to data grid and can be viewed and saved in the normal fashion. D. For each randomly selected point the area around it is divided into equal quarters and the distance to the nearest individual in each quarter recorded. These simulators can be used to explore the efficiency of the different methods under a variety of assumptions. Density. D. 1. DfD can simulate data collected using this method so that you can appreciate the likely errors. 1949) This is a 'pointcentredquarter' method. following selection. H. Data Simulation Select Simulation and either Transect Data or Plotless Density Estimate from the top menu to simulate transect sampling observations or plotless density estimation protocols. DfD can simulate data collected using this method so that you can appreciate the likely errors. Plotless density estimation The dialogue for this simulator is shown below. In both cases. Angleorder nearest individual (Stearns. I.ˆ ˆ2 var[D] = D (3N − 2) . Pisces Conservation Ltd 18 . a dialogue window opens in which simulation parameters are selected. is calculated using: D = 12 π ∑ 1 / Rij 2 where Rij is the distance from the ith random point to its closest individual in the jth quarter and N the number of random sample points. For each randomly selected point the area around it is divided into equal quarters and the distance to the nearest individual in each quarter recorded. Within the Method panel on the righthand side select the type of data that you wish to simulate. 1957) This is a 'pointcentredquarter' method. Density.
The length and width of the sampling area is chosen using the Dimension edit box. Select Regular to arrange the objects in a regular grid pattern and Clumped to produce a contiguous distribution. If Random Seed is set to 0 (zero) the program selects the random number seed using the date/time function of the computer. Place in the Density edit box the required density of the objects in the study area. If Regular is selected then it is possible to select the degree of regularity by the numerical value placed in the Noise edit box. Select Random to position the objects at random with the study area. For example. The smaller the value the tighter the clusters. Thus. if closest individual is selected this will select the number of distances measured from random points to their nearest object. The value entered into the Number of Samples edit box gives the number of samples that will be simulated. This is simulated by using a random number generator to give x and y coordinates. The objects are randomly placed within each cluster and the Tightness parameter can take any value from 0. The distribution of the objects and the results of the simulation are shown graphically (next page). The default value of 1 is good starting point as it produces a reasonable number of objects. Values for tightness above 1 will produce ever more randomlike distributions. A value of 0 (zero) will be perfectly regular: as the value increases a potentially larger random displacement from the regular distribution is applied to each object. every simulation will be different. This is useful as methods differ in their ability to handle nonrandom data. Pisces Conservation Ltd 19 . For most studies a 10 by 10 area. which is the default. To be able to generate exactly the same simulation a number of times enter a number in Random Seed and remember it! Once the simulation parameters have been entered select Ok to generate the simulated data or Cancel to leave without data generation. If clumped is selected the Number of Clumps can be altered to give the number of clusters of objects within the sample area. will be found suitable.The distribution of the objects that will be sampled is selected within the Distribution Type panel.
Inner belt distance is the distance to the outside limit of the inner belt in metres. The density is assumed to be measured in objects per square metre. Any positive integer can be used although large numbers may result in an overload error if the number of objects detected becomes too large. number can be chosen. 2. However. The detection function (the decline in the likelihood of an object being spotted with distance) is selected from the Detection Function panel. the selected point or individual as a yellow cross and the distance calculated by the program as a green line. Line transect simulation The dialogue for this simulator for both point and line transect data is shown below. Depending on the function chosen other parameters may have to be entered in the parameter list on the righthand side. This parameter is required when simulating a 2belt line or point transect. Within the Method panel on the lefthand side select the type of data that you wish to simulate. Maximum detection distance is the maximum distance at which an object will be detected in metres. Actual density is the density of the objects in the simulated study area. positive. Transect length is the length of the line transect in metres.The individual objects are shown as red squares. It is used by the simulator to fix the scale of the sighting experiment you are envisaging. Pisces Conservation Ltd 20 . large numbers may result in an error if the number of objects detected becomes too large. Any real.
P. T. & A. Nearestneighbour distance measurements for the estimation of animal populations. On a method of estimating biological populations in the field. Ecology 37:451460. Hale. J.. For line transects the detected objects are shown as red squares and the perpendicular distance as green lines. C. R. Biometrics 38:127135. F. & Laake. (1958). (1976). Buckland. Wildlife Management 59(4). Distance to nearest neighbor as a measure of spatial relationships in populations.. Anderson. Burnham. The larger the value the more rapidly the probability of detection declines with distance. R. & T. 1986. Cottam. V. B. Robust density estimation using distance methods. P. & Laake. 667673. 1975. For point transects detected objects and the sighting distance is shown in red or green and undetected objects in blue. S. Biometrical Journal 21. (1980). Curtis. Delince. Byth. D. L.. L. On the variable circular plot method of estimating animal density. Distance sampling: estimating abundance of biological populations. 1987. 1986. & Anderson. 1982. J. D. K. Biometrics 42:753767.. Chapman and Hall. 1953. T. J. T. Undetected objects are shown as blue squares. San Diego. M. The larger the value the more objects will be detected at extreme distances. G. E. Bulletin of the Ecological Society of America 28:60. & Hill. K. J. T. Some sampling characteristics of a series of aggregated populations. Diggle. The use of distance measures in phytosociological sampling. D. J. (1993). Ecology 35. T. For both point and line transects the results of the simulations are shown graphically. Curtis. Bird census techniques. P. Robust density estimation through distance measurements. Decay constant is the negative exponential detection function parameter. Clark. A point method for making rapid surveys of woodlands. Nonparametric estimation of population density for line transect sampling using Fourier series. R. Crain. Curtis.. 1956. Important references Bibby. 325336. M. C. Some sampling characteristics of a population of randomly dispersed individuals. A. Ecology 38:610621. J. L. S. C. D. (1992). Biometrics 32. Ecology 67:15761581. 21618. Cottam. P. Pisces Conservation Ltd 21 .J. Blackith. London. (1995). Cox. Biometrika 62:3948. Biometrics 43:363384. P. K. Wildlife Monograph 72. 731748. dev. T. Craig. G. (1953).. of detection function is the standard deviation used for a halfnormal detection function. Anderson. W. & B.. J. It is a real. Academic Press. Ecology 39. On robust distancebased intensity estimators. Burnham. & Murphy. 14750. R. K. R. This should be a real.. Cottam. (1979). & J. G. Biometrika 40. 1957. I. C. Accuracy and precision of techniques for counting great blue heron nests. N. J. Burnham. D. G. positive value. Ecology 34: 741757. positive number. Burnham. Buckland. K. G. D. Cottam. 44553. R. Burgess. P. Anderson. Mathematical models for nonparametric inferences from line transect data. & Evans. G. (1954). Dodd. For two belt surveys the colour of the detected object and the sighting line is green if the point is within the inner belt and red if it is the outer belt. J.Std. Clayton.. 257 pp. Some robust density estimators for spatial point processes. F.. 1947. & Laake. Estimation of density from line transect data. Catana.
Morisita.C. Ensign. 113123. B. 1979. M. Kovner. Properties of estimators of wildlife population density for the line transect method. M. R.. N. Pennsylvania. E. L. Persson. Burnham. & K.. Patil. A. H. G. England. E. M. (1982). Series E. M. 52(1). Estimation of population density by spacing method. W. S. S. Over. L. Burnham & J. & Smith. S. Pollard. Evaluation of analytical procedures for density estimation from linetransect data: Data grouping. Sen. U S National Marine Fisheries Service Fish. Volume 2. C.. The robustness of estimating density by distance measurements. (1996). A. 24(1). K. S.. T. & Weaver. Journal of Wildlife Management 43:484492. Biology 1:187197. Skellam. Washington. F. M. Geometrical probability. Moran. P. C. W. Kovner. Kovner. Waters. L. A. London. Tourigny. M. D. 1975. Pank. Tissot. Biometrics 30. Physiology and Ecology 7:134144 (in Japanese). 1971. E. Piciou. Southwell. R. A. Lewis. Ecology . & Patil. Wildlife Soc. Johnson. W. P. London. L. C. (1995). 213222. (1974). Biometrics 30. Optimum nonparametric estimation of population density based on ordered distances. G. Biometrics 25. D. Biometrics 35:597604. (1969).. Bull. Sci. Stearns. J. & Barss. & Dolloff. Marshall. M. G. University Park. Hixon. J. Bull. I. 1954. R.. On distance estimators of density in randomly distributed forests. A. Pisces Conservation Ltd 22 . Density estimation by variable area transect. E. Nonparametric estimation of plant density by the distance method. USDA Forest Service. C. H.. 30:350358. E. 317328. Seber. P. Angermeier.. C. Kendall. R. 385400. Griffin. C. R. 3236. 1971. 94(1). (1993).. Morisita. Line transect method of estimating grouse population densities.. J.Engeman.. J. E. L. Statistica Neerlandica 17. Biometrics 38:243248. D. 13545. H. & P. (1996). An Introduction to Mathematical ecology. Griffin. The distance method for estimating densities.. O. Robust estimation of density for a twodimensional point process.P. T. & Dusenberry. A. 329340. USA. 1982. Wiley_Interscience. New York and London. A. (1968). Keuls. data truncation and the unit of analysis. The mathematical foundations underlying the use of line transects in animal ecology. A. H. E. Fish. Statistical ecology. J. Starr. 1957. Ecology 75(6). (1974). Ninety years' change in a northern hardwood forest in Wisconsin. Jr. Simulation study of estimators for the line transect sampling method. Use of line transect methods of estimate abundance of benthic stream fishes. A new method for the estimation of density by spacing method applicable to nonrandomly distributed populations. Can. Wildlife Research 20(4). & De Wit. C. & Olson. (1958). Gates. Gates. 1979. 7191. Aqu. M. Pielou. (1969). Pages 175187 in G. 225230. G. Comparison of submersiblesurvey and hydroacousticsurvey estimates of fish density on a rocky bank. Patil. Biometrics 27:9911002. E. (1994). R. Biometrika 62:519521. [Available as Forest Service translation number 11116. K.. G. and W. P. 1963. F. Sugihara. The estimation of animal abundance and related parameters. (1963). K. Kelley. F. W 1949. Patil.] Parker. 17691779.. Fox. S. Biometrics 24. Linetransect sampling for estimating breeding wood duck density in forested wetlands. P. Memoirs of the Faculty of Science Kyushu University. 433444. W. J. I. editors. Pennsylvania State University Press. Biometrics 14. USA. S. E. On the line transect sampling method. A comparison of plotless density estimators using Monte Carlo simulation.
R. Size and dispersion of a Louisiana population of the cricket frog. G. J. H. The growth and distribution of the mounds of Lasius flavus (Fabricius) (Hym. Ecol. & Stortenbere. E. W. 54. (1962). Distribution of distance to nth neighbour in a population of randomly distributed individuals. R. Dean. J. The assessment of the size of populations of adults of the red locust. 549569. F. 421437. W. P. 258268. (1956). N. (1956). 31. 93104. M. The theory of line transects. Turner. 391394. & Blackith. (1963). Res.. Bull.: Formicidae) in Silwood Park. Bird Study 3. Nomadacris septemfasciata (Serville). B. Ecology 41. ent. Acris gryllus. R. Ecology 37. Yapp. Pisces Conservation Ltd 23 . Anim. C. (1960). Thompson. in an outbreak area. Waloff. W. B. Berkshire.Symmons.
This action might not be possible to undo. Are you sure you want to continue?