"Base network links on evidence"
By default, each link is evaluated with available evidence from all (more than 50) datasets from all species. You can re-define the evidence base of a specific sub-network retrieved from the database by selecting individual species / data types, or groups thereof.
"Show sub-network(s) in"
By default, only one organism is shown - namely, the one in the box above ("Query ... network" above). However, the FunCoup framework is based on orthology and it is thus natural to ask: "What the respective sub-networks look like in model organisms?". Here you can select a species (even several at once) and see orthologs of the query gene(s) and the network(s) around. Every pair of orthologs will be connected with dotted green line. An interesting feature is that if you choose "several organisms (via orthologs)" and the original species is not selected in the list, then (except the query gene as entry point) ONLY MODEL organisms' networks will be shown.
While using this option, you observe links in a number of species. In general links for the query genes in each species were predcted using all evidence from orthologs all species. However that information was irrelevant for seeing eveolutionary conservation of the network. Hence by default, the retrieved sub-network in this mode is "fairly" based on evidence from only the same species. You can restore the full evidence set by "Use evidence from all species".
In addition, there is a (default) option "Prioritize links with orthologous relations". Given there are too many links in the retrieved sub-network, it prefers such links to the query gene(s) that enable richer sub-network(s) between orthologs in other species. The alternative to this (if you uncheck the box) is filtering in the query species only by link confidence. In this case, remaining sub-network does not necessarily correspond to a conserved network in other species.