Good titles are clear, succinct, and memorable without being disruptive. Bad titles are misleading, vague, and/or overly wordy.
Here are some publications with various titles.
The first one “A distinct lineage of giant viruses brings a rhodopsin photosystem to unicellular marine predators” was used as a counter argument when it was said in a discussion that titles were not that useful when scanning through journal articles. It succinctly presents the important/interesting points of the paper.
The next two “Reinforcement of stickleback mate preferences: sympatry breeds contempt” and “Nice snake, shame about the legs” are quite memorable. And when talking about reinforcement in the process of speciation “sympatry breeds contempt” is an elegant way to summarize a central idea.
“Reading the entrails of chickens: molecular timescales of evolution and the illusion of precision” is memorable but a bit hyperbolic. “Molecular dating: ape bones agree with chicken entrails” seems to springboard off of this and have fun with the symbolism.
“The Matthew effect in science, II: Cumulative advantage and the symbolism of intellectual property” may seem like an odd title to include here. It is not a stellar title. However, the author spends the first part of the introduction talking about the title and concludes “An obscure title can also have a latent function: to keep one from assuming that the title truly speaks for itself, and thus to make it necessary to elucidate one’s intent.” However, it is not advisable to purposefully engineer an obscure title.
Avoid a title that is overly specific as well. Including the scientific name of the organism, the geographic location of study, or method details in the title makes it sound overly specific and not relevant to the reader. “RFLP variation and null alleles in Acipenser baerii sturgeons of the Yenisei River basin” will not get as much readership or interest as “Gene flow between fragmented populations along a disrupted freshwater corridor”.