There's a great deal of popular misunderstanding about the so-called butterfly effect, and unfortunately this article simply repeats the classic trope. The term "butterfly effect" does arise out of the study of chaotic systems but has nothing to do with real butterflies. Chaotic systems are distinguished by the fact they have two separate “phase spaces” and can shift from one to another by means of very small alterations to initial conditions. A graph plot of such a chaotic system is shown below:
As we can see, this superficially resembles the wings of a butterfly, hence the name.
Although one can create computational simulations in which a small perturbation (the beating of a butterfly’s wings) in one variable can ultimately be amplified to result in statistically significant macro-scale effects in the overall model, in practice dampening effects from all the other interacting components of real-world systems makes such an event vanishingly implausible. This is, mathematically, the same reason why we don’t see quantum uncertainty in macro-scale objects.
Thus we do not have to worry about a rogue lepidoptera in Brazil causing a tornado over Houston.