As suggested by his name, the guy is a legend. One of the most cited authors in ecology, I have him in (almost) the same league with James H. Brown, sir Robert M. May or Stephen P. Hubbell.
Legendre is not famous for creating a revolutionary ecological theory and he does not stand out as a storyteller. Rather, he is appreciated for the multivariate analytical methods that he co-created and popularized. And he is a master teacher. His papers and books (e.g. Numerical Ecology) are an example of an exquisite didactic writing.
Last week I saw Legendre speaking at the Spatial Statistics conference in Columbus, Ohio. He spoke mostly about spatial eigenvector mapping, it was more of a retrospective of the last 15 years of research in that area. He was exactly how one would expect: precise, methodical and didactic in the strictest sense. Specifically, Legendre speaks to his audience and he implicitly (and correctly) assumes that most people do not have a clue about anything. Other statisticians bogged their audience in equations, assumptions and busy Beamer slides. In contrast, Legendre took his time to explain every symbol, every term and method. And he still fitted into his time window, he transmitted the idea and everybody got it. Wonderful.
Legendre is also a classical frequentist. Nowadays the main statistical hype is all about maximum likelihood, MCMC and Bayesian hierarchical models. All of the other talks at the conference were about that. Legendre does not go there - he still dwells in the area of permutation tests, null hypotheses, analytical solutions, normal assumptions and linear algebra. But he is so good at it, he explains it so well, and he provides a ready-to-use software so that everybody will cite and use his methods anyway.
Oh, and just a funny observation: Legendre's voice and body language reminded me of one particular TV character: Walter White from Breaking Bad (before he breaks bad).