I'm not familiar with MATLAB's PSO toolbox, but you could save *every* fitness evaluation from within fitness function. In the postprocessing, the "fitness of each iteration" is a minimal/maximal fitness of set of $n$ entries ($n$ is swarm size).
where $f$ is vector of fitness for all evaluated particles, $n$ is swarm size and $i$ is iteration index. If you want "overall best fitness up to $i$-th iteration", just additional do $F_i = min(F_i, F_{i-1})$.
You have to code the PSO in matlab and then save the value of the fitness function by yourself. In stead of starting from scratch, you can try my code (attached). If you have any clarification in the code, do not hesitate to contact me.