As to hardsuit's claims, DMR has all of the feature sets he described in one form or another but it will be heavily dependent on the manufacturer. Motorola, Hytera and Simoco have the most featured rich DMR products on the market right now.
One of the major advantages to Motorola and Hytera DMR systems is the ability to enable pseudo-trunking. While these are vendor proprietary, it allows low budget trunking systems to be built (but there are limitations like the lack of ID databases) but access can still be restricted if planned correctly. The real advantage is for single channel trunking (Simoco also offers this for a single channel) allowing for two voice channels on a single frequency (this should've been part of the DMR Tier 2 spec IMO).
Then there is the true control based trunking systems such as Motorola's (Trident's) now EOL Connect Plus and the systems that follow the Tier 3 standard such as Motorola's Capacity Max (which I'm not fond of), Hytera's T3 solution, Tait/Harris' T3 solution and Simoco's Xd T3 solution (which is my favorite as it's is in a lower price-point than Motorola's multi-site pseudo-trunked solution, Linked Capacity Plus). All of these use authentication as a standard feature in a manner where the SU's serial number is used to validate the SID during the registration process (note, P25 also offers this but majority of the systems out there don't use it).
NXDN on the other hand, has better audio and typically has lower cost equipment. While I've never had the opportunity to physically check, I've also heard reports of better low-signal capabilities compared to DMR. Primary disadvantages to NXDN is a lack of vendors with Icom, Kenwood (who now has a DMR solution as well), and Alinco and the physical channel spacing capabilities. What I mean with that, both Kenwood and Icom advertised the ability to fit two 6.25 kHz channels in 12.5 kHz spectrum (where DMR is 2 times slots on a single 12.5 kHz channel) but what they failed to inform customers about was the fact this physically can't be done on a single cavity combiner and will either require two cavity combiners with physical separation of antennas or a hybrid combiner. So a choice has to be made, do you split your combiner and physically separate your transmit antennas (which may have adverse coverage consequences) or do you take the extra 5 dB loss with the hybrid? This is before even addressing the additional hardware costs for site infrastructure if you need equivalent capacity to a DMR system.
Now like mmckenna said, he went from a Smartnet system (Type II) in a 5 channel configuration to a NXDN system in a 5 channel configuration as he didn't need the extra 5 voice paths a DMR control based trunking system would have offered for the same 5 carriers. Personally, I wouldn't go with either for a large scale trunking system but that's just me.