Let $X\subset \mathbb{P}^n$ be a smooth complex projective variety, and consider a non-trivial action of $\mathbb{C}^*$ on $X$. For any connected fixed component $Y$ of the fixed point locus, we may define the plus and minus Białynicki-Birula cells $X^{\pm}(Y)$ as $$X^{\pm}(Y)=\{x\in X\mid \lim_{t\to 0} t^{\pm 1} t\cdot x \in Y \}.$$ It is a famous result by Białynicki-Birula that $X^{\pm}(Y)$ are locally closed, and they provide two decompositions of $X$.
My questions is: can it happen that $X^+(Y)=X^+(M_1)\sqcup X^+(M_2)$, with $M_i$ two subvarieties of $Y$?
I've always thought of those cells as "tubes" joining two connected fixed components $Y$ and $Y_1$, but maybe they could join also another component $Y_2$. Intuitively, can the Białynicki-Birula cells appear as "disjoint pants"?
The disjoint union is forced since we are considering orbits, but I cannot come up with a proof or a counterexample.