From Landau & Lifshitz, Classical Mechanics, the number of integrals of independent integrals of motion for a system of $s$ degrees of freedom is $2s-1$.
I am considering a spherical pendulum in absence of gravity. For this problem $s=2$, so I expect 3 independent integrals of motion. The first two are the angular momentum in the $\theta$ and $\phi$ direction. The energy is also a constant of motion, but in the absence of motion it is a function of the angular momentum since $$L=T=\frac{m\ell^2}{2}(\dot{\theta}^2+\dot{\phi}^2\sin^2(\theta)).$$
I have 3 integrals of motion but 2 are independent. Did I miss one, or have I misunderstood the statement from Landau? The linear momentum should not be conserved, so should I include integrals of motion of ignorable coordinates?