This is version 3.0 of PEST. Here is a brief history of PEST versions: PEST 2.0 (Zujko-Miller and Miller, 2003) is an application written in Microsoft .NET developed to assess congruence between tree files for Continuous Jackknife Function analysis as described in Miller (2003). Version 2.1 (Zujko-Miller and Miller, 2003) added the capability of parsing tree files in nexus format. Version 2.2 (Zujko-Miller and Miller, 2003) added the capability of counting conflicts between trees. Version 2.3 added the capability of counting unresolved nodes and also added an option for discounting the most inclusive resolved node. This latter modification means that users no longer have to modify the reference tree by collapsing the basal node to form a trichotomy. Version 3.0 has an entirely new interface with the capability to spawn tree searches in NONA (Goloboff, 1993). Miller (2003) used PEST 1.0, written in JAVA, to describe Continuous Jackknife Function analysis. PEST 2.3 is faster, easier to use, and more feature rich than PEST 1.0. However, PEST 2.0 and later versions abandon the cross platform potential of the JAVA language. PEST versions 2.0 and later require instillation of the .NET Framework. The original Pest 1.0 is available from this link; PEST 2.3 is available from this link. Pest 3.0 has been tested in Windows 2000 and XP.

The PEST tab.
Tree search in NONA. This check box determines whether PEST will be used to launch a tree search pattern in NONA.
Assess congruence. This check box determines whether PEST will assess congruence between trees. Three options for assessing congruence are available: matches, conflicts, and unresolved. Matches counts the number of clades with identical composition (although not necessarily identical internal arrangement) in a comparison of two trees; conflicts counts the number of nodes in conflict between two trees; unresolved counts polytomies in rarefied trees that are resolved in the reference tree. The sum of matches, conflicts, and unresolved nodes will be equal to the number of resolved nodes in the unrooted reference tree.
Reference tree. The user can specify an xread tree file to serve as the reference tree or can allow NONA to determine the reference tree. Under the default setting, the strict consensus of the most parsimonious trees found in NONA under the options selected in the Tree search tab will serve as the reference tree.
Collapse basal node? PEST scores are calculated on an unrooted network. Networks contain one less resolved node than rooted trees. NONA saves tree files with this additional node, not a basal trichotomy. It is recommended that unmodified trees saved using NONA have the basal node collapsed.
Output file. Select the name of the output file and use the browse button to select its location. The output file will be a tab delimited text file. When PEST is finished running, import this file into Excel and graph the results.
Run. Begin analysis. Make sure the proper settings under the Tree search tab have been selected first.

Tree search tab.
Path to NONA. Use the Browse button to indicate the path to NONA.
Path to matrix. Use the Browse button to indicate the path to the data matrix.
Replicates for each batch of rarefied characters. This is the number of times the data matrix will be subjected to each level of character removal.
Interval between batches. This will determine how fine grained the analysis is. If 2 is selected, NONA will first run a series of analyses in which probability of any character being removed from the matrix is 2%, then 4%, until the probability of removal for each character is 98%. If 5 is selected, the probability of character removal will first be 5%, then 10%, continuing until 95%. It is recommended that the interval between batches selected divide evenly into 100.
Number of trees to keep after each search. This is the HOLD/ command in NONA. This limits the number of trees kept in each replication of "MULT." Keeping this number low speeds up individual replicate searches, but can result in missing most parsimonious trees.
Number of random taxon addition replicates. This is the MULT* command in NONA. MULT* performs TBR branch swapping on a Wagner tree generated using random taxon addition. The number selected indicates the number of MULT* replicates for each rarefied matrix. High numbers can dramatically increase analysis time, but low numbers can lead to inaccurate results.
Branch swapping. This is the MAX* command in NONA. This can increase the accuracy of results, but will increase analysis time. Some branch swapping already occurs under the MULT* search.
amb= / amb-. The amb= setting allows NONA to count a node as resolved if its only and uncontradicted support comes from an ambiguously optimized character that can be alternatively optimized to support some other node; the amb- setting would consider that node collapsed.
Tree files. Use the Browse button to select a location where tree files will be saved.
Tree file prefix. Give a name to identify the tree files. One tree file will be created for each percent probability of data removal. A number indicating the percent probability data removal will be appended to the root of the file name.
Save the batch file. This is an option to save the series of commands given to NONA. Use the Browse button to select the path and file name for the batch file. Save the batch file with current settings using the Write Batch File button.

Congruence tab.
This tab is useful for comparing trees found using other programs (including NEXUS programs like PAUP*), or comparing trees found using PEST and/or NONA at an earlier date. Options under this tab will only be available when Tree search in NONA (under the PEST tab window) is unchecked and Assess congruence (under the PEST tab window) is checked. A reference tree must also be selected under the Pest tab window.
Comparison trees. Use the browse button to select the folder containing the reference trees.
Filter trees. This is the file extension used to identify tree files. PEST will compare all files with this extension found in the selected folder to the reference tree selected in the Pest tab window.
Tree format. Select xread or NEXUS file format. Both the reference and comparison trees must be in the same file format.

Typical Use of PEST.
Under the PEST tab, the following options should be selected: Tree search in NONA, Assess congruence, Reference tree: use default reference tree from NONA search. Under Output data, select Matches. Choose whether to also include Conflicts and/or Unresolved. Enter the name and directory path for the output file. The output file will be a tab delimied text file that can be imported into Excel.
Under the Tree search tab, use the Path to NONA browse button to indicate the path to NONA. Use the Path to matrix browse button to indicate the path to the xread data matrix that is being analyzed. Select the number of replicate matrices to be created for each level of character removal. Select the interval between batches. Select the number of trees to hold for each MULT* search. Select the number of MULT* search replicates for each rarefied matrix. Choose whether to use MAX* branch swapping. Choose whether to use the amb= or amb- criterion for dealing with ambiguous character support for nodes. Select a folder where output trees will be saved. Indicate the file name prefix for tree files. Choose whether to generate a batch file containing the commands that will be sent to NONA.
Return to the PEST tab and press the Run button.

When the analysis is complete, open the output file in Excel. The content of this file will depend on the Output data selections made (matches, conflicts, unresolved). If matches were selected, the output file will contain congruence, which is the average raw number of matching clades for each percent probability character removal, and scaled congruence score, which is the congruence divided by the number of nodes in the reference tree. If conflicts were selected, the output file will contain conflicts, which is the average raw number of conflicting clades for each percent probability character removal, and scaled conflicts score, which is the conflicts divided by the number of nodes in the reference tree. If unresolved was selected, the output file will contain unresolved, which is the average raw number of unresolved clades for each percent probability character removal, and scaled unresolved score, which is the unresolved divided by the number of nodes in the reference tree. All of these results are given for each value of percent probability character removal (remember, the higher the number, the more data is removed). Create an XY plot in Excel with percent probability character removal as the X-axis, and the scaled results (e.g., scaled congruence score) as the Y-axis.

Nixon, K. C. 2002. WinClada. Version 1.00.08. Program and documentation. Available from http://www.cladistics.com. Published by the author, Ithaca, NY, USA.

Miller, J. A. 2003. Assessing progress in systematics with continuous jackknife function analysis. Systematic Biology, 52:55-65.

Zujko-Miller, C. AND J. A. Miller. 2002, 2003, 2004. PEST. Version 1.0, 2.0, 2.1, 2.2, and 3.0. Program and documentation. Available online at http://www.gwu.edu/~spiders/pest.htm. Published by the authors, Washington, D.C., USA.