I'm guessing that the raw data for bycatch are counts (cf. http://en.wikipedia.org/wiki/Bycatch) so a purist might say that ANOVA may not be the way to go, since that best suited for testing differences in a continuous variable (eg. length, mass, etc.) between multiple groups. ANOVA on count data can be done, but caution is required when interpreting the output (cf. http://sci.tech-archive.net/Archive/sci.stat.math/2004-12/0076.html).
After you've had a look at the data by plotting (including the regression as @PeterFlom suggests), you might use a goodness-of-fit test (with equal expected frequencies) to test your null hypothesis that there is no difference in bycatch between the gear categories.
A popular goodness-of-fit test is the chi-square test which is well-documented for minitab, eg. http://www.minitab.com/en-US/support/answers/answer.aspx?ID=1255 and
Some of the important limitations of the chi-square test are that you can't input percentage data, only raw counts, all counts should be greater than five, and that the sample should come from a simple random distribution (cf. http://www.minitab.com/support/documentation/Answers/Chi-Square%20Test%20Assumptions.pdf).
If your data don't satisfy the assumptions of the chi-square test, you can use an exact test or a randomization test. A quick look suggests these are available in minitab, but I'd suggest getting familiar with R, since it has a very large community of active users to seek advice from and substantial amounts of user-contributed tutorials on all of these methods (here is one of the best: http://www.statmethods.net/). R is also free to use and gives easy access to more exotic tests that you might consider, such as Bayesian contingency table analysis and likelihood tests.
That said, if ANOVA is the convention for count data in your field, or you convert the counts to a ratio (would bycatch per person per hour be sensible?), here is a nicely documented example of how to do it with R: http://ww2.coastal.edu/kingw/statistics/R-tutorials/oneway.html