1) You have assumed that R0 is constant, which it is not, and therefore the rise in case numbers is not exponential (actually it is 'worse' than exponential at first, but becomes 'better' than exponential very quickly and tends toward linear before reaching a peak). The shape of your own graph from March 2020 should tell you that at a glance. ....
R0 doesn't (or shouldn't) change for a particular variant of a particular virus - since it is a measure of transmission of the virus in a totally susceptible (non-immune) population who are 'mixing totally freely', with no restrictions. It is the 'effective' reproduction number (R, Re or Rt) which will change as a result of changes in the susceptible (non-immune) population, formal 'NPIs' or 'informal' changes in behaviour of the population.
Since I have received a couple of questions off-list about this, I thought I would attempt to explain things, as I understand them, in case anyone else is confused and/or interested ...
As above, essentially by definition, R0 (the 'basic reproduction number') for a particular variant of a particulate virus cannot change within a given population (it may
differ, but still not ever change, in other countries with different ethnic/whatever characteristics) - since it represents the average number of people who will be infected by one infected person
if there is totally free unrestricted mixing within the a totally susceptible population (i.e. no-one with any immunity).
What
does change, for various reasons, is the 'effective reproduction number' (R, Re or Rt). This will reduce (to below R0) if any steps are taken to reduce transmission (e.g. formal NPIs, personal behaviour changes etc.) or if the population ceases to be 100% susceptible" (i.e. if some people develop immunity).
There is a lot of talk about 'exponential growth' of the number of new cases although, strictly speaking, this is never possible. Exponential growth implies the same
proportionate change over equal time periods (e.g. 50% increase every day, 'doubling every 7 days' etc.) - which would lead in not only the number of new cases increasing every day, but with the rate of rise also increasing every day (since, each day, the rise would be X% of a number greater than the previous day), reflected in the classic shape of an 'exponential curve'.
However, in the context we are discussing, such exponential growth would/could only occur if ('effective') R remained constant (so, exactly the same percentage day-on-day increase every day). However, every time a new patient becomes infected, the proportion of the population who are 'susceptible' decreases slightly (since some are assumed to become immune due to the infection). R will therefore reduce slightly every day, hence the 'rate of rise' (of new cases) will be slightly less that with true 'exponential growth'.
However, having said that, in practical terms that effect is fairly trivial (unless one allows the virus to spread 'totally out of control'). Primarily because they have responded to impending 'waves' by implementing NPIs, no country has had a 'wave' of infection that involved an appreciable proportion of the country's population, so the effect of this progressive decrease in R (hence slowing of 'rate of rise') has, in practice, never been particularly significant.
Even in India, who are just coming out of a wave lasting around 4 months with a peak of just over 4 million reported cases per day (on 1st May), the total number of reported cases between the first day it rose above 40,000 per day (20th March) to the first day it eventually fell back to below 40,000 per day (5th July) was 'only' just over 19 million - a massive number, but only about 1.4% of the country's population, so only resulting in a very small decrease in R (over that entire ~4 month period), hence 'rate of rise' (i.e. deviation from exponential growth). However, small though it is in most countries, this effect
is taken into account in many/most of the models used.
That leads to a question. Echoes is correct in saying that, in a good few countries (including the UK), growth of new cases has been far from exponential - although relatively close to exponential at the start of the rise, most of the subsequent rise appears to be closer to linear than exponential (i.e. a much lower, maybe approaching zero, 'rate of increase of the increases' than would occur with exponential growth). So, if, as above, the effect increasing degree of population immunity on R is far too small to account for this, then what
is the explanation for this appreciable deviation from exponential growth?
Mathematically, the answer is simple. The only thing that could cause this is a progressive reduction in R for other reasons - but what are those other reasons? No-one knows the answer for certain, but what seems the most credible guess is that when people see (the daily reported numbers of) cases, hospitalisations and deaths rising rapidly, they become increasing compliant with imposed NPIs and, in a good few cases, go beyond the minimum requirements of those NPIs (e.g. keeping 3 or 4 metres away from people, rather than 2 metres, or not meeting people indoors, or going to non-essential shops, even if the 'rules' allow them to). If that's what's been happening, it could explain the progressive reduction in R as the rising phase of a 'wave' worsens - in other words, sensible 'negative feedback' self-imposed by the population.
Contrary to what echoes seems to be suggesting, most modellers do
not ignore this and hence just assume 'exponential growth'. Most attempt to include this effect in their models but, of course, they cannot do much more than guess how behaviour will change as people see 'rising numbers' being reported. It's getting a bit easier for them - in that, being now in our third major 'wave' of infection, they can get some handle on that changing behaviour during a 'rising phase' by looking at what happened during the first two waves.
For those who advocate strategies which rely on trying to achieve 'herd immunity' in most (or all) of the population, this is presumably a frustration to them, since human nature being what it is, a substantial proportion of people seeing the numbers 'rise rapidly into the clouds' will probably respond by self-imposing 'protective measures' (even if no mandated ones are in place), thereby potentially greatly extending the period of time required to achieve the desired 'herd immunity' (by infection), even if that is attainable.
I hope that the above makes sense and may be of some help and/or interest to some on my readers (if there are any whio have got this far!).
Kind Regards, John