I think the important parameter in life tables is transition between life stages. So, if you can justify combining cohorts as being at the same life stage, there is no reason not to.

Another important consideration when combining data is independence. If the cohorts can be considered independent, that is, properties of one cohort do not affect the others, then you can combine them. If cohorts influence each other, then considering them as one sample is problematic.

Thank you David! This is a really good criterion. Therefore, for example, if the age of the sexual maturity is three years, then I can try to combine up to three consecutive cohorts.

No hard rules for not considering the consecutive cohort for life table analysis. If  consecutive cohort  are from same stock, it is easy to combine both cohort for the purpose.

Я не думаю, что выборки разных лет можно объединять в одну последовательность. Условия каждого года могу по разному формировать когорты. 

А как Вы думаете, можно ли количественно оценить эти различия или их отсутствие? Точнее, знаете ли Вы работы на эту тему?

Существуют стандартные статистические процедуры, способные оценить однородность выборки.

Во-первых, мой вопрос об объединении когорт, а не выборок разных лет. А во-вторых, у меня не может быть однородных выборок, т.к. они структурированы по возрасту от 0 до 12 лет.

Dear Pavel, I combined three cohorts in the same survival and recapture rate analyses in the following work (Ferreira et al 2013 Life history of a neotropical marsupial. Evaluating potential contributions of survival and reproduction to population growth rate). I first tested if there was a difference between age classes in survival rates. I found no difference between the tree consecutive years.

Good luck!

Thank you Mariana! This is exactly I am looking for. How did you test these differences?