Being in Digital Ecosystems research myself, I would say the author is missing one important point: Absence of single points of control and failure. Natural ecosystems are rarely if ever dependent on singletons (single species or single organism) to be sustained. Their adaptivity and resilience comes from that one fact. All the examples the author states are tied to a single corporate entity, often designed so, and therefore cannot exceed the lifespan of the entity that created them. A great example of that is napster. Once the central entity was shut down by way of courts, the ecosystem evaporated.

In a sense the Internet and the Web are much closer to the true Digital Ecosystem, even though purists such as myself can still find things that can be improved, single points of control/failure etc. but at least they are not as pronounced as the ones present in the star-topology twiter/amazon/apple ecosystems.

To a certain extent I agree with that.

But ecosystems are not only biological they are also physical and the sun is a pretty good candidate for a singleton then isns't?

To me the defining factor is the flow and exchange of energy throughout the system. In digital ecosystems the data is what needs to flow.

The internet is obviously the mother of all digital ecosystems but it is also restrictive (firewalls etc) which keeps data from flowing.

The data (and metadata) is definitely important. And yes the Sun is a singleton in our solar system. Closer to home, if anything goes wrong with the earth, (asteroid, LHC, some unknown weapons programme) we're all doomed, including all ecosystems on the planet (depending of course on the magnitude of the disaster). The field of global existential risks is devoted to asking such questions.

That being said, I would rather my work depends on the sun's or earth's continued existence, rather than on twitter's or facebook's or amazon's uptime/finances/goodwill. We should aim to remove SPOFs/SPOCs to the extent possible.