The flow and distribution of nutrients in the ocean follow weak gradients from their underlying sources in a kind of thermodynamic equilibrium. Some areas will have nutrient excesses based on geography, geochemistry, and limitations on consumption rate due to dependencies on other nutrients.

If iron is the rate limiting ingredient, then when you seed an area with iron a bunch of other nutrients are consumed in the process that currently are not being consumed. This changes the chemical equilibrium driving those other nutrient flows in the ocean and may stop critical nutrients flows into areas that rely on them. Any major local change to nutrient balance changes the equilibrium and thermodynamic gradients of the entire system.

In hindsight this is kind of obvious. There are similar equilibrium problems in large chemical reactors too and the ocean is just a giant reactor vessel to a first approximation. I think the original assumption was that the ocean is so big that no one would notice but long distance effects on local nutrient balances were observed such that increased sequestration productivity in one area was at least partially offset by losses of productivity in other areas due to new nutrient bottlenecks.

In principle modeling the entire system would allow one to inject the right nutrients at the right handful of spots to maximize aggregate sequestration performance with minimal risk. Building such models is still very much beyond us.