MARK 4.2 | now available...

announcements (new versions, changes, bugs, installation problems...) related to program MARK

MARK 4.2 | now available...

Postby egc » Wed Apr 27, 2005 2:06 pm

Greetings -

Gary has just posted the newest version of MARK (now 4.2). You can grab the setup.exe from his server, or here from this mirror directly:

http://www.phidot.org/software/mark/download/

Remember, to install a new version of MARK, it is highly recommended you uninstall the old version first.

This new buid is more than a 'cosmetic tweak' - here are the major changes, fixes and enhancements:

April, 2005 -- Version 4.2 -- Female Lark Bunting

108. The models for estimation of closed populations when there is
mis-identification of animals developed by Paul Lukacs, Colorado State
University, have been added to MARK. These models were developed for use with DNA marking, when there is a "small" probability of the genotype of the animal being incorrectly determined because of low quantity or poor quality DNA. The existing 6 closed captures models in MARK were all extended to include the alpha parameter (probability that the animal was correctly identified, with 1 - alpha the probability that the animal was incorrectly identified). Thus there are 12 closed captures models now in MARK. Because the population size is no longer included in the likelihood with the mis-identification models, all of the closed captures models now also report the population estimate as a derived parameter, so that model averaging and variance components analyses can be conducted across these different data types. That is, model averaging is appropriate for the closed captures models with and without the mis-identification parameter is appropriate.

However, as before, the Huggins models and the models that include the
population size (N) or more technically, the number of animals never
encountered (f0), do not have comparable likelihood values, so you shouldnot try to perform model selection when both of these models are in the same results browser window.

109. The 6 robust design data types were extended to incorporate the
mis-identification parameter, so that there are now 12 different robust
design data types.

110. The 6 multi-strata robust design data types were extended to
incorporate the mis-identification parameter, so that there are now 12
different multi-strata robust design data types. Note that the Barker
robust design data types have NOT been extended to incorporate the
mis-identification parameter because the most likely place in these models where the mis-identification would take place is the resightings during an interval (coded as a "2"), and the theory has not been worked out to handle mis-identification with this type of encounter.

111. A problem with retrieving the muti-strata models when the psi
parameters had been redefined was fixed. Previously, the redefinitions of
the psi parameters was lost when a model was retrieved, so that the default of obtaining the probability of staying in the current stratum was obtained by subtraction was always the case. Now, the redefinitions are maintained when the model is retrieved from the results browser.

112. A problem with retrieving models with design matrices that used the POWER function have now been fixed.
egc
Site Admin
 
Posts: 201
Joined: Thu May 15, 2003 3:25 pm

Return to software problems/news

Who is online

Users browsing this forum: No registered users and 1 guest