I'm not sure there is a single "best test" to analyze growth data as preferences change over time.  However, the questions you are asking and the method and types of data you are collecting to acquire your measurements of growth can help point you in the general direction.  For example,  if your rate of growth is acquired through recapturing individuals, then a repeated measures analysis would likely be appropriate.  If it is based on otolith increments, then perhaps a MANOVA would be more suitable.  What kind of data (and how was it collected) are you using for your analysis?

Bensahla, to compare growth of fish one must account for differences in fish size (allometry), temperature, and the growth period (time), because these factors are known to effect fish growth. If you can't account for these factors, you can't really say much. Assuming your fish is a heterothermic ectotherm. To account for fish size a scaling parameter (often called Omega) is used. This can vary among species. For example, see the paper by Perry et al. on my research gate page titled, "Using a Laboratory-Based Growth Model to Estimate Mass- and Temperature-Dependent Growth Parameters Across Populations of Juvenile Chinook Salmon" - this paper and its citations should lead you well into the world of fish growth.

The ideal case is when your fish have the same initial starting weights, are reared under constant and known temperatures for a known amount of time. Then size, temperature, and time can be factored out, and one can compare fish growth using R.J.'s suggestions above.

Also, If you know the quantity of food eaten, and the energy density of the food and fish, then perhaps bioenergetics models based on mass-balance relationships may be used.

Enjoy,

JP

what is the mode and the characters used to measure that growth? if the measurements made on different individuals with sampling different populations you can compare averages by stuend-fisher test for unrelated samples; if not you can compare your average compared to a Reference witness

Hi Bensahla,

I don't know if it is the best option, but you can usually find the Pauly and Munro (1984) formula:

φ’=2*log(Linf)+log(k)            Being Linf and k the VBGP

Pauly, D. and J.L. Munro. 1984. Once more on the comparison of growth in fish and invertebrates. Fishbyte. Vol. 2 (1): 21. (online available)

Regards!

Alba

Conducting ANOVA on long term Condition Index values

thank you all for your valuable answers: i use lenght frequencieswith Von Bertalannffy equation so i have for entry Linf, K,t0 and lenght frequencies i read about hotelling T2 test (copare growth between sexes  or different populations...etc) in several articles but didn't find how to use it?! does anybody have an idea ?

that test won't control for any of the environmental variables discussed above.  But it can be done relatively easily in MS Excel if you lack more powerful software and if your datasets aren't too large:

http://www.real-statistics.com/multivariate-statistics/hotellings-t-square-statistic/hotellings-t-square-independent-samples/

Von Bertalanffy equation describes growth like exponential curve. I dont know how compare exponential curves. But I think possible simplify a problem. If you transform primary data (eg find the logarithm), it you get regression lines which possible compare by adapted t-test. This procedure (compared of two regression lines) described by Glantz "Primer of biostatistics".

I also think you may try use factorial ANOVA test. Before the ANOVA tests, dont forget about data transformation, to get normal distribution.

I hadn't used Hotelling's T2 test, but how I have understand, it has Student's t-test in base too.

Hi Bensahla,

you could logarithmise your growth curves (maybe you need to double-log on x- and y-axis) to transform them into  straight lines and then perform an analysis of covariance (ANCOVA), which allows you to compare the slopes and intercepts of two lines and test whether these are significantly different from each other. You can find software for this in R using the package Rcmdr.

Best,

Nik

Hi Bensahla,

It's been a long time since I have done any age and growth work.  We used a likelihood ratio test to compare von Bert growth curves.  The technique is described in Kimura, D. K., 1980.  Likelihood methods for the von Bertalanffy growth curve.  Fishery Bulletin.  77:765-776.  You can download the paper for free from the NOAA's Fishery Bulletin webpage.