That's a very interesting question in general. I rotate amps and speakers and I am always a bit amazed at the differences I hear, even on supposedly equally competent (based on specs and published data) amplifiers. Leaving room issues aside, yes, the speaker is, by far, the most important factor and you won't get clean 30Hz bass from a speaker that, by design, falls sharply at 80Hz. You can correct that a bit, as explained here
https://www.electronicdesign.com/systems/how-get-big-sounds-small-speakers (ironically for TI based class D amplifiers), or even, in today's world, after you have measured your speakers (REW + umik for example) and applied a DSP correction in roon.
Now, whether that correction is somewhat effective depends a lot on the amount of energy your amplifier can deliver and how the speaker behaves in that frequency range. I couldn't find the Atom 7 measurements but I did find the Atom 3 (
https://www.stereophile.com/content/paradigm-atom-v3-loudspeaker-measurements) - the speaker is intrinsically limited by its size and design, doesn't seem too hard to drive, but you can see how it falls off the cliff below 100Hz. There's probably no way you are going to be able to compensate for 10, 15 or 20 dB by applying a DSP filter.
As far as why the Marantz sounds better, remember that to produce bass, you need a lot more energy than for mids or highs. Here is a very good educational resource on the basics principles of loudspeaker design
http://hyperphysics.phy-astr.gsu.edu/hbase/Audio/spk.html and more specifically on the more power necessary for bass
http://hyperphysics.phy-astr.gsu.edu/hbase/Audio/spk.html#c5.
Cheap class D amplifiers running on anemic power supplies can sound good but are intrinsically limited in the power they can deliver. For fun some people have experimented running them on large batteries and seem happy with the short term result in terms of bass expansion at higher volume (I have my doubts on the reliability due to increased heat dissipation, but that is another story). One of the possible explanations (and the most likely imho, but would love to be educated in details by people more competent than I am) is that the Marantz, being build like a tank in terms of power supply and capacitors (at least compared to the cheapish class D amplifier) is able to deliver more energy to the speakers in that frequency range. Another possibility, but one that would still require the ability to deliver more energy in that frequency range - you can't cheat physics - is that it has an adjusted response curve, which would have been well suited to the typical speakers of the day.
Anecdotally, comparing a FX Audio D802 on the supplied power brick to a Marantz HD-AMP1 on KEF LS50 (which have a better bass response than the Atom 3), you could at first think the D802 sounds as good as the 10 times more expensive HD-AMP1. But as you raise the volume to "punch" level, you literally hear the D802 bass disappear... The KEF LS 50 wireless - amazing speakers for the size - use class D amplification as well, but they use dual amplifiers, per spec 200W in the LF range and 30W in the HF range. That at least gives us a rough idea on what KEF engineers though the energy delta required was.
That issue is something I keep running into. It even rears its ugly head with higher end speakers in the 20000 eur/usd + range where I have a pair that I would describe as hard to drive (impedance going as low as 2 Ohms) and that only seems to deliver taunt and punchy bass with what I see as oversized amplifiers.
I really would love to see tests in the area of speaker/amp matching - I feel sensitivity, SNR, separation and response curves only give part of the picture. But then, maybe it would be pointless as it seems we are heading for integrated active speakers...