I've seen that type of uptick from web crawlers like search engines or archive snapshots, and from people doing mass captures or mirrors of the site.
Usually Google and the big players are careful to rate-limit their requests, it is generally the people doing dumps and mirrors that cause the issues.
There are several good Apache modules to help. Two immediately come to miund. First, mod_ratelimit is supported by Apache, and limits transfer rates per connection. A third party mod_limitipconn limits the number of simultaneous connections by a single address, which can also help. Between the two, most of the bulk pulls can have their impact softened. Few will use distributed tools, sometimes they'll be pulling through AWS but even then it is uncommon to see more than 5-10 IP addresses pulling at once.