Make sure the vast majority of your content is textual.
Text hasn't been safe from scraping since the beginning of scrapers. No. The only way you can "watermark" text is by filling it with references to yourself and/or your brand, and even then it's not perfect. No, I would modify this by saying, make sure the vast majority of your content is hard to scrape, like tools.
For something like recipes, there are special rules in copyright law. I mean, how many apple pie recipes are there? So it's something that's likely to be a target for scraping anyway. However, if you were to build a tool that helped suggest ideal recipes for people based on their dietary needs and preferences, that would be genuinely useful, and impossible to replicate without a ton of work, because the source code would be hidden. I don't think anyone has built something like that yet, but it would be incredibly link-worthy if they did.
Code, not content, is where added value can be both created and protected. We're swimming in content, but we're really short of ways to intelligently filter that content for different niches.