I had not heard of a transformation for kurtosis. Other researchers also address this problem as  Leptokurtic distribution. High kurtosis data has heavy tails. I would suggest apply power transformation. You can take a look at:

https://www.utexas.edu/lbj/sites/default/files/file/news/normalization.doc

Hope this helps

Try robust estimation of your analysis. This will give less weight to outliers. There are several methods with Robust Analysis that you could try to achieve reasonable estimates of your biomass decomposition function. The thick tails might be an indication of a more complex process involved that creates such outliers.