There is a fascinating article in BMC Biology (doi:10.1186/s12915-014-0066-4, Open Source)which deals with bacterial evolution. It is an eye-opener for people not involved into the subject. Bacterial evolution is so fast via horizontal gene transfer (HGT) that there is almost no time for “normal” base exchange mechanisms. The predominant way of evolution is by gene loss, but occasionally via gene gain which may change the bacteria dramatically. This has also an impact of the charactization of bacteria whose genome  should be regarded as part of a pangenome or even supergenome. 

I wonder whether a given size (which is obviously maintained)  would limited horizontal transfer of larger parts of the genome.

Since this is open source I copy the conclusion entirely:

Conclusions
The reconstruction of short term genome dynamics events shows that microbial genomes exist in a state of perennial flux, gaining, losing, expanding and  contracting gene families. Typically, genome dynamics processes are rapid, with gains and losses of multiple gene families occurring within the time frame of a single nucleotide substitution per gene. Thus, gene flux is the dominant mode in microbial evolution such that microbes primarily differ from each other on the scale from static to highly dynamic. The rates of gene family gain and loss in most microbial groups are approximately an order of magnitude greater than the rates of expansion and contraction of pre-existing families, indicating that HGT is the principal source of new genes in prokaryote evolution. Overall, gene family loss notably prevails over gain, i.e. evolving genomes appear to spend more time contracting than  expanding. It seems most likely that the gradual gene loss is compensated for by episodes of rapid gene gain; most of such bursts are outside the evolutionary scale accessible through ATGCs although a few were detected. The absolute as well as relative rates of genome dynamics events show remarkable variance among bacteria, spanning almost two orders of magnitude, and do not significantly depend on the ATGC-wide dN/dS estimates, the taxonomic affinity of microbes or their life style. Conceivably, genome dynamics is highly sensitive to local ecological
factors the exact nature of which remains to be elucidated. The analysis of genome dynamics allowed us to estimate the size of microbial supergenomes which in the majority of the analyzed microbial groups turned out to be large but closed, exceeding the characteristic genome size by about an order of magnitude, but for a minority of microbes were appeared to be open.

Good to know!