A fascinating DNA analysis of a huge variety of citrus species and its relatives, and an attempt at the ascent of the various species.

www.plosone.org/article/fetchObject.action?uri=info%3Adoi%2F10.1371%2Fjournal.pone.0062574&representation=PDF

takeaways:

Poncirus and Fortunella are true citrus, and closer than the Australian citrus are; both are about as related as medica is. Traditional citrus basically breaks down into Citron/Pommelo/Mandarin/Poncirus/Kumquat groups.

Clymenia is true citrus, and is as closely-related to the citron/pommelo/mandarin cluster as eremo/microcitrus is.

Eremocitrus glauca is a descendant of microcitrus australis.

So it appears that citrus learned to be deciduous (Poncirus) fairly recently, and may have learned xerophytia (eremocitrus) even more recently.

The drawback of such analysis is that it uses the sequences from chloroplasts that inherited almost exclusively from seed , mother plant. Since most of the citruses are complex hybrids it could be misleading.
Several papers are now in preparation that use the total sequence of hundreds citrus accessions.