Most accounts of the K-T extinction state that no purely terrestrial animal larger than a cat survived (crocs and champsosaurs are / were semi-aquatic). Cat-/chicken-sized non-flying predatory dinos are known (most notably Compsognathus).
_No_ purely terrestrial animal larger than a cat survived, and _almost no_ purely terrestrial animal smaller than a cat survived.
David Marjanovic suggests the whole terrestrial ecosystem was
devastated too baldly to sustain viable populations of predatory dinos.
We have some direct evidence for this, in the form of paleosols, the fern spike, the fungal spike, and...
But this is too simple:
* Mammals survived.
As I just said: many mammals did _not_ survive. Let's take the metatherian example: Before the boundary, we had "pediomyids", stagodontids, peradectids, and maybe still deltatheridians. After, we had peradectids and notometatherians*, which is apparently descended from some "pediomyid" -- it is AFAIK enough to suppose that _two species_ of metatherians survived. The eutherian example is little different: before the boundary, we had true and false zhelestids, cimolestans, probably** asioryctitheres, *Deccanolestes*, probably** zalambdalestids, and leptictidans. After it, we had cimolestans, leptictids, and the clade composed of *Purgatorius*, *Protungulatum*, *Oxyprimus*, and the (other?) placentals in the strict sense, which is the sister-group of Leptictida; it is enough to suppose that _three species_ of eutherians survived. I know little about multituberculate phylogeny, but not long before the boundary there were djadochtatherian**, taeniolabidoid and ptilodontoid multis, of which only the latter two groups are known from after the boundary, and the known Cretaceous diversity of taeniolabidoids was very small, so it is likely that very few species of multis survived.
I also remember reading a little book from 1981 (by Archibald and someone else) that argued that 30 % of the North American lizard species died out at the boundary. Notably, the herbivorous polyglyphanodontids bit the dust.
* Contains the marsupials in the strict sense and the now extinct borhyaenoids. ** Known only from Mongolia, where the top of the Maastrichtian is missing.
Why could small small non-flying predatory dinos not
survive in similar ecological niches?
Because they, or most of them at least, did not have similar ecological niches. The surviving mammals and lizards mentioned above were for the most part granivores and insectivores; of the only dinosaurs in these niches, some survived -- all of them happened to be neornithean birds (and lithornithids, if those don't belong to Neornithes).
Incidentally, no flightless theropod the size of *Compsognathus* or smaller is currently known from the Maastrichtian, AFAIK.
* The greatest devastation of plants was in N America, less in other parts of the N hemisphere, and even less in the S hemisphere.
Are you sure? New Zealand burnt down as well. That's where the fungal spike (world going mouldy) was found for the first time.
He also bases his response solely on the impact hypothesis. I've seen suggestions that the N and S hemispheres had separate wind systems, as they do now, so most of the fallout from Chixculub ( well N of the tropics) would have affected only the N hemisphere, except for the CO2 emissions.
No. The gas emissions went all the way up to the mesosphere; the solid ejecta exited the atmosphere and reentered all over the globe. This is inevitable from an explosion of that size.
If another factor (e.g. Deccan Traps) was responsible for most of the extinction in the S hemisphere,
(The Deccan traps are out of the question anyway because the main episode of eruptions ended 100,000 years before the boundary. It had made the global average temperature rise by 3 °C, and when it was over, the temperature came back to normal. And that was it.)