MAC PROGRAMS FOR ANIMAL BEHAVIOR

by J. Bradbury & S. Vehrencamp

This page describes and provides access to three Macintosh programs that may be of use to researchers in animal behavior or field ecology: ANTELOPE, THE KERNEL, and SINGIT! These are available for free to interested users. The authors have sporadic time to devote to their modification and upgrading, so free use comes at the expense of little support. All are offered under copyright by the Regents of the University of California. In addition, some MATLAB ROUTINES for home range computations (including overlap measures) and for temporal cross correlation, correcting for within series autocorrelation, are provided below.

ANTELOPE (version 1.3)

FUNCTION OF PROGRAM:
ANTELOPE performs a number of spatial statistics tasks that we have found useful in our teaching and research at UCSD. It only analyzes x,y data (not x,y,z) but it accepts text files with additional columns which might be used to focus on subsets of the data (e.g. sex, time, age, etc). Such a set of locations might result from mapping individuals in a population or from following a radio-tracked individual over time. Data are input as text (ASCII) files with columns separated by tabs. The first two columns must be the x and y location respectively.

Features include:

SUPPORT AND AVAILABLE VERSIONS:
This is an unsupported program: the authors have only limited time available and users utilize the program at their own risk. ANTELOPE was written in Symantec Pascal.There are several versions of the program which may be downloaded from this home page:

Future versions of ANTELOPE will add lacunarity computations to the Fractal module. We have begun adding a Regions Menu which will allow for computations of overlap of "patches" defined by grouping algorithms, home ranges, or external criteria. The program will currently save and reload "region files"; we have not yet implemented the overlap calculation box (but it is coming along). We are also setting up to convert the ugly box plots in the Fourier and Harmonic home range plots into contour plots (next version, we hope). The next version is likely to be written entirely in C++.

MEMORY REQUIREMENTS:
Current versions of ANTELOPE have the requested memory set to 1.5 MB (2.5 MB for PowerMac version). The minimum storage just to start up the program is about 1.2 MB, and the program gobbles more as data are read in and analyses proceed. For most data sets, we find that we need at least 2 MB and for bigger sets, 3 MB. To adjust memory for the program, open the Get Info box and change the memory request allocation in the lower right corner.

HELP AND DOCUMENTATION:
The program has extensive HELP boxes available under the Apple Menu. This is provided instead of an external manual. At the end of each help section, some relevant references are listed. The program assumes that users know why they should perform a particular analysis and how to interpret the results. There is no tutorial provided. If users do not find analyses in ANTELOPE that they need, they may wish to examine the of Legendre and Casgrain which covers a much broader set of analyses.

CITATION:
ANTELOPE has no manual or other published documentation. If you require a citation for publication purposes, please cite the web address for this page:

http://www.nbb.cornell.edu/neurobio/jbsv_downloads/programs.html

THE KERNEL (version 1.0)

FUNCTION OF PROGRAM:
THE KERNEL is a program that accepts a file of x,y locational data and uses the kernel method to estimate home ranges. This module will eventually be incorporated into ANTELOPE when and if the authors get time to do so. For the moment, we are making this free-standing module available to anyone who could use it. Kernel methods have largely replaced harmonic mean estimates of home range in the last few years. They also have an advantage over Fourier methods in that they are not as sensitive to the size of the world in which the range is estimated.
SUPPORT AND AVAILABLE VERSIONS:
As with ANTELOPE and SINGIT!, this is an unsupported program. Use at your own risk. There is only one version available: this has been compiled to run on any recent Mac including PowerMacs. To download a bin-hexed version of this program, click here on The Kernel.
HELP AND DOCUMENTATION:
Double click the icon to start the program. There is a help box under the Apple Menu that explains the use of the program, the format required for data input, references for kernel home range methods, etc. You may need to increase memory for very large data sets using Get Info under your File Menu before starting the program up.
CITATION:
THE KERNEL has no manual or other published documentation. If you require a citation for publication purposes, please cite the web address for this page:

http://www.nbb.cornell.edu/neurobio/jbsv_downloads/programs.html

SINGIT! (version 1.7)

FUNCTION OF PROGRAM:
This is a program for interactive playback of digitized songs using a Macintosh. It is much faster and cleaner than alternatives (e.g. Hypercard), but it takes a little prep work to set up the sound files. You will need your own copy of the Apple program ResEdit . If you do not have this program, and wish to download version 2.1.3, then here is your chance: ResEdit .You will also need a program and hardware with which to digitize your sounds (e.g., SoundEdit, Canary, etc.). These you must buy yourself. Sounds, labels, and a mnemonic image are stored in resource files on your hard disk. SINGIT! can load any such resource file, let you select which songs you want to play back to some poor unsuspecting organism, and then let you play songs back selectively by either clicking on an image or hitting an assigned key. Sounds may be played through the onboard speaker or more usually fed out the Mac speaker port to an external amplifier/speaker. The program includes a clock dialog box which can be brought to the front of the desktop to allow for normal delays between songs. The program also acts as a simple event recorder should you want to keep track of when you and your wild opponent each produced a sound.

AVAILABLE VERSIONS:
There is currently one available version of SINGIT! for all Macs (v1.7). This can be downloaded as a folder within which are the program, a sample sound resource file, and a detailed TEXT file with instructions. If you want to download this now, click here on Singit! v 1.7. File is bin-hexed for transfer. After transfer, double click on antelope.sea icon to open up compacted folder containing necessary files.

MEMORY REQUIREMENTS:
SINGIT! itself requires only minimal memory. However, we usually allocate 2.5 MB to the program to accomodate sizable resource files. Use the GetInfo box before starting the program to increase memory for very large resource files.
CITATION:
For citation, please use:
 
Bradbury, J. and S. Vehrencamp. 1994. Singit!: a program for interactive playback on the Macintosh. Bioacoustics 5:308-310

 

MATLAB FILES

SOURCE:

We oversaw a graduate course in spatial and temporal statistics within the Neurobiology and Behavior program at Cornell University in Fall 2000. The course reviewed the relevant statistical methods and worked up routines for them, wherever possible, using Matlab. Several topics overlap with those covered in the stand-alone programs above, so we make them available here. Each week we covered a different topic. This involved presentation of a general review of that topic, a set of Matlab M-files that were useful, and a homework assignment that helped participants understand how to use the methods and the Mfiles. Below we provide the Mfiles and homework assignments for several of the topics. As with the above programs, we accept no responsibility for errors or difficulties you may have with these routines. They work on the most recent versions of Matlab for PC's and on the last MAC version that was commercially available (v. 5.1). Student versions of Matlab may not have all the routines that our Mfiles require.

HOME RANGE TOOLS:

This package deals with the computation of home range area and overlap given one or more files of x,y coordinates. Although a number of methods are available, (minimum convex polygons, elliptical ranges, Fourier smoothing, harmonic mean ranges, and kernel home ranges), we only provide Mfiles for the first and last of these alternatives. These two were the ones judged most robust to boundary conditions, assumptions, and local artifacts. The entire package of Mfiles, homework data, and the assigment (with instructions on how to use the new Mfiles) can be downloaded either as a MAC package or a PC package. The focal Mfiles are:

convexpoly: function A = convexpoly(XY,CXY). This routine accepts either a single two-columned matrix XY containing x,y locations, or a single matrix XY and a cell matrix CXY containing any number of two-columned matrices. In either case, it plots the minimum convex polygon for each submitted matrix and computes its area in the units of the file. The computations are exact (not estimates). If multiple files are submitted, the user can also compute the area of overlap for any subset of the submitted files, and the fraction of each file's polygon occupied by that overlap area. Written by Rulon Clark, Martin Schlaepfer, and Jack Bradbury. Cornell University, December 2000. This unit requires access to Mfiles area, interval, intsecl, iscross, isinpoly, isintpl, linechk, polybool, and polyints written by Kirill K. Pankratov, Massachusetts Institute of Technology, 1995.

kernel: function XYZ = kernel(XY, CXY). This routine computes kernel home range for one (XYZ) or more (XY and CXY) two column matrices. CXY must be a n x 1 cell matrix with a single cell for each submitted matrix. The routine also computes both area and volume overlap estimates for any subset of the submitted matrices. Output is a matrix with the first two columns containing the x,y coordinates of grid points within the sample world and each subsequent column containing the probability of finding the animal in the submitted file at each grid point. This routine requires access to Mfiles changemaps, findh, and overlap. Written by Rulon Clark, Martin Schlaepfer, and Jack Bradbury. Cornell University, December 2000.

TEMPORAL CROSS CORRELATION:

This packet includes tools for extracting correlations from TWO time series, correcting for autocorrelations within each before-hand or as part of the process. There are two ways to do temporal correlations. Lag sequential analysis treats the time series as sequences of discrete events. An excellent program for this can be downloaded for free (GSEQ)The second method assumes continuous variables, (if you have event results, convert the streams to rates) and for this, we have a very sophisticated MFile set. For the entire package of files and data, select MAC or PC format.

macra: function macra(X). This is a shell for running a bidirectional multivariate auto- and cross-regressive analysis on two simultaneous, continuous variable time series. X is a N by 2 rectangular matrix. It plots auto and partial correlations with pacorr.m and then calls the stepwise GIU stpwise.m. Procedure computes the significance of the difference between a regression model with cross regressive plus autoregressive terms and a model without the cross regressive terms, first with one column as the dependent variable and then with the other column as the dependent variable. Written by S. Vehrencamp, Nov. 2000.

pacorr: pacorr(x, n, cor, name). This calculates and plots the serial auocorrelation function and the partial autocorrelation function of x, a column vector, for up to n lags. n should be < N/4, where N is the length of x. The unbiased estimate of r and the progressive Bonferroni correction of alpha are used. name is a string variable with the name of the column. Written by Gerry Middleton, November 1996 & modified by J. Bradbury Sept 2000 and S. Vehrencamp Nov 2000.

stpwise: function stpwise(X,y,inmodel,alpha). An interactive tool for stepwise regression that fits a regression model of Y on the columns of X specified in the vector INMODEL. ALPHA is the significance for testing each term in the model. By default, ALPHA = 1 - (1 - 0.025).^(1/p) where p is the number of columns in X. This translates to plotted 95% simultaneous confidence intervals (Bonferroni) for all the coefficients. The least squares coefficient is plotted with a green filled circle. A coefficient is not significantly different from zero if its confidence interval crosses the white zero line. Significant model terms are plotted using solid lines. Terms not significantly different from zero are plotted with dotted lines. Click on the confidence interval lines to toggle the state of the model coefficients. If the confidence interval line is green the term is in the model. If the confidence interval line is red the term is not in the model. Use the pop-up menu, Export, to move variables to the base workspace. Modified by J. Bradbury Nov. 2000.

Additional m-files called during these routines that you may need to add to your folder (included in your packet): chi2cdf.m, diff.m, fcdf.m, mean.m, std.m, tinv.m

For questions and feedback on these programs, send email to

jwb25@cornell.edu

Last updated: 22 January 2001