Please bear with me for this technically convoluted piece of information below.

To my knowledge, the ecommerce pagination poses a lot of challenges for pagerankto flow through from a top level page to its subsidiary pages. Also, the sorting of product listings by price, color or size (for example) creates multiple versions of the same set of pages (through query string parameters appended to the base URL) thus creating more duplicate content. In these cases, the rel=canonical tag pointing to the top level page on each of the pages of product listings was intended to make search engines understand that the top level page is the most representative page.

But the rel=canonical tag really does not help in the indexation of all unique pages because the representative page at the top of the chain is given so much importance. This has been solved to a great extent by rel=prev and rel=next tags.

To show that all the products are listed in sequence and that each of them is important as a distinct unique product in a given category, the rel=prev and the rel=next indicate to the search engines that it is a series and that all links to individual pages should be consolidated as a whole than spread it around individual pages like abc.com/products?page=1 and abc.com/products?page=2 etc

In the case of content, as in the case of products above, if you have the URL for page 2 of the article as abc.com/article?page=2, you can have rel=canonical tag pointing to <link rel'"canonical" href="abc.com/article?page=2"> and
<link rel="prev" href="abc.com/article">
<link rel="next" href="abc.com/article?page=3">
for the same purpose of showing the search engines it is to be considered as one series though it appears to be many and that all links and indexing is to be consolidated. There is no top level page getting superiority in this case as each page is in itself a representative page.

Hope this helps.