dna sequencing - Why does one combine PCR and cloning as ways for amplification of sequences?

Why does one combine PCR and cloning as ways for amplification of sequences? Don't they produce the same result? I was reading the paper https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1864885/ and got confused by the following passage:

Cloning and Bisulfite Genomic Sequencing

Each bisulfite-converted DNA sample was subjected to PCR by primers that did not discriminate between methylated and unmethylated sequences, using a GeneAmp PCR core reagent kit (Applied Biosystems). The primer sequences are listed in Table 1, and reaction conditions are listed in Supplemental Tables 1C and 1D (available online at http://ajp.amjpathol.org). Each subsequent PCR product was TA-cloned into pGEM-Teasy vector25 (Promega) for transformation into Escherichia coli strain JM109, according to the manufacturer’s instructions. Clones were picked randomly and colony PCR was then performed using vector primers T7 and SP6 to amplify the cloned inserts. Cycle sequencing was performed using BigDye version 1.1 (Applied Biosystems) and an automated capillary DNA sequencer Genetic Analyzer 3100 (Applied Biosystems). The sequences obtained were aligned and compared using SeqScape software (Applied Biosystems). The completeness of bisulfite conversion was first confirmed before scoring. The CpG sites sequenced as cytosine or thymine residues were scored as methylated or unmethylated, respectively. The methylated site frequency was calculated for each sample by dividing the total number of methylated sites over all cloned CpG sites.

Can someone comment on what purpose the cloning serves when the amplification is already done by PCR?

Answer

There are several reasons to clone a PCR product but the main reasons include expressing the product and maintaining the library inexpensively.

In the study you mentioned, the cloning and transformation is apparently done for obtaining single clones (from single colonies).

Clones were picked randomly and colony PCR was then performed using vector primers T7 and SP6 to amplify the cloned inserts. Cycle sequencing was performed using BigDye version 1.1 (Applied Biosystems) and an automated capillary DNA sequencer Genetic Analyzer 3100 (Applied Biosystems).

Since the original PCR would have a mix of different sequences, cloning is done to separate them out for the ease of studying them by sequencing. If they don't separate the sequences, they would have to use NGS (or more expensive methods like PacBio for longer reads). They have used Sanger sequencing. This article was published in 2007. At that time NGS was not as cheap as it is today and was hence not used by everyone. Also, the NGS machines of that time had even smaller read lengths (~40 for Solexa/Illumina GA and ~75 for GA-II). Considering these, Sanger sequencing may have been preferable to the authors of this paper. However, as I mentioned before, Sanger's method cannot be used for sequencing pooled samples; you need pure templates or else there will be mixed signals and base calling would be inaccurate.

(See the example chromatogram below: look at the position with the Y which has mixed T and C signals. It is not from a pooled sample, though).

Therefore cloning, transformation and colony picking was necessary.

Moreover, in many cases picking individual clones would be necessary for performing functional assays to associate phenotype with genotype.