You are mixing base R with tidyverse
syntax here, when working with tidyverse
functions there is no need for JS_CU$
before the variable name. I can't test this because you are not providing a REPRoducible EXample (reprex) but I think your code should be like this:
JS_CU %>%
group_by(time_int, survey) %>%
summarise(mass_mean = mean(mass, na.rm = TRUE))