Minor feedback (typos):
"Early in the ecological network litterature" should be "literature"
"In part, this success is to be attributed to the fact that connectance relates to early deﬁ-
nitions of network complexity (Pimm 1982), and to the fact that connectance predicts
reasonnably well some key dynamical properties of ecological networks" I would rewrite as
"In part, this success can be attributed to the relationship of connectance to early definitions of network complexity (Pimm 1982) and also because connectance reasonably predicts key dymanical properties of ecological networks."
I would rewrite "For this reason, and given the importance that degree distribution took in the
recent years, it is important that we clearly understand how constrained degree distribution actually is, in relation to connectance. "
"For this reason, and given the rising importance of degree distribution, it is important for ecologists to understand constraints upon degree distribution in relation to connectance."
I would change Figure 2 to label network size in the upper-right corner.
I think connectance is a very visual topic and it may help model-shy ecologists to see diagrams of different networks so that we can better understand the kinds of networks discussed in this paper. It would be very helpful if, when discussing each instance of how connectance imposes limits on the shape of the degree distribution in Part 3 Simulations, you also used two very simple example network with only a handful of species (nodes) involved, make one densely connected and one sparsely connected, and explicitly discussed how the variance, coefficient of variation, kurtosis, and skewness would vary in both of these networks. I know that you are already simulating simple networks and choosing the number of nodes, but I think the paper would benefit from a more bare-bones network with species at each node to help non-theoretical ecologists understand your punchline.
I did not get a clear idea of how to test whether a network has constraints on interactions. How do you find the expected degree distribution of a network? Is it possible to estimate connectance if not all connections within a network are known?