Open Source development is closely tied to the Unix philosophy, which C# doesn't really subscribe to. C# shines in complex, multi-tiered OOP architectures, where, combined with a good comfortable IDE, it can help handle enormous amounts of complexity. If you have to build a giant monolith that does everything including invoicing, taxes, printing subscription forms, suggesting related products, and watering the plants, then C# is your first choice.
However, the open source world seldom needs this; open source projects are usually much smaller, and more modular by themselves. The KISS principle, exercised to the max, often leads to designs for which even C is perfectly adequate, and consequently, OOP is seldom called for (a few notable exceptions notwithstanding).
Another reason might be that many Unix programmers prefer a lean-and-mean textual 'environment' (a terminal, a selected set of command-line tools, and a good text editor like vi or emacs) over a full-blown IDE. Such a setup is great for C, Perl, Python, etc. (or at least I think so), but C# practically needs an IDE to compensate for its verbosity and deeply nested library namespaces - either way, coding C# using only vi is not fun (again, speaking from my own experience). Open source is about freedom, and this freedom includes the freedom to choose whichever tools suit you best. A language for which there is only one IDE doesn't work well in such a world.