We argue that pseudoreplication is unlikely to be associated with confounded effects as 1) we have accounted for such issues in our statistical analysis, by using an appropriate multilevel model (GLMM) [ 61 ]; 2) we have maintained very high standards of water quality, which made sure there were no other differences in the seawater between treatments, other than carbonate chemistry, that could be responsible for the differences observed; and 3) our results are highly significant, and consistent with other studies [ 38 , 39 , 40 ], which provides a degree of replication on its own.
1
—
—