I've not been through many of these, so no recommendations except this one that is a reprint from the "Communications of the ACM" from December, 2005: "Agile Project Management: Steering from the Edges"
Frankly, I like it because it is written at a post-college level, something hard to find in the Agile literature, it's a case study focusing on PM, and it has some interesting insights.
One thing I don't agree with is that it equates complex software systems and projects with 'complex adaptive systems' [CAS]. CAS is definitionally about biological systems that have the ability to change and evolve behaviour in response to stimulus. Ant colonies are the classic example. I don't buy that one. Software always works the way it is designed to work, even if it's designers don't understand the design they wrought, and users don't understand how the system reacts to input and control. There's nothing biological about software, and it certainly doesn't learn, WATSON et al not withstanding.
However, there are aspects of CAS that do apply: CAS are non-linear, and software systems can certainly be non-linear, especially when tangible system assets can't meet the demands of the intangible software. CAS are open and dynamic, responding to environment. As repects the project rather than the software deliverables, I agree that many projects that deliverable intangibles have such characteristics.
But I digress: read the paper. There are things to be learned
