Over the past decade, marine scientists published a series of studies warning that humanity’s burgeoning carbon dioxide (CO2) emissions could cause yet another devastating problem. They reported that seawater acidified by rising CO2—already known to threaten organisms with carbonate shells and skeletons, such as corals—could also cause profound, alarming changes in the behavior of fish on tropical reefs. The studies, some of which made headlines, found that acidification can disorient fish, make them hyperactive or bolder, alter their vision, and lead them to become attracted to, rather than repelled by, the smell of predators. Such changes, researchers noted, could cause populations to plummet.
But in a Nature paper published today, researchers from Australia, Canada, Norway, and Sweden challenge a number of those findings. In a major, 3-year effort that studied six fish species, they could not replicate three widely reported behavioral effects of ocean acidification. The replication team notes that many of the original studies came from the same relatively small group of researchers and involved small sample sizes. That and other “methodological or analytical weaknesses” may have led the original studies astray, they argue.
“It’s an exceptionally thorough replication effort,” says Tim Parker, a biologist and an advocate for replication studies at Whitman College in Walla Walla, Washington. Marine scientist Andrew Esbaugh of the University of Texas, Austin, agrees that it’s “excellent, excellent work.”
But marine biologist Philip Munday of James Cook University, Townsville, in Australia, a co-author of most of the papers the Nature study tried to replicate, says there are “fundamental methodological differences” between the original and replication studies. “Replication of results in science is critically important, but this means doing things in the same way, not in vastly different ways,” he wrote in an email.
Munday helped launch research on the behavioral impacts of ocean acidification together with Danielle Dixson, now at the University of Delaware. In 2009, their paper in the Proceedings of the National Academy of Sciences showed that orange clownfish (Amphiprion percula) reared in seawater with elevated CO2 levels no longer recognized the chemical cues that could help them find a suitable habitat on the reef. (“Losing Nemo” was a popular headline for stories about the paper.) That study was followed by dozens of others showing similarly striking, and often large, behavioral effects in clownfish and other species, mostly from tropical waters.
Timothy Clark, the first author on the Nature paper and a marine scientist at Deakin University, Geelong, in Australia, says he initially set out to probe the physiological mechanisms behind those behavior changes. But after he failed to reproduce the changes—let alone explain them—he invited other scientists to set up a systematic replication attempt. It focused on three reported effects of acidified waters: making reef fish prone to swim toward their predators’ chemical cues rather than fleeing them, increasing their activity, and altering the fish’s tendency to favor either their left or right sides in some behaviors. The researchers didn’t seek to repeat each previous experiment one for one, but Clark estimates the entire effort covers the research reported in at least 20 studies.
Overall, the group reports, exposing fish to seawater with acidification levels predicted for the end of the century had “negligible” effects on all three behaviors. The Nature paper also reports the results of a statistical analysis called a bootstrapping simulation, designed to calculate the probability that Munday and co-authors could have found the striking data on chemical signal preference presented in seven papers. The authors say the odds are exceedingly low: “0 in 10,000,” as they put it.
Clark declined to elaborate on the implications of the bootstrap finding, but says he “would encourage any other avenues of investigation to find out what has caused the stark differences between our findings and theirs.” Esbaugh calls the bootstrap analysis “a little concerning,” but he objects to the “somewhat nefarious undercurrent” in the Nature paper. “I know both of these research groups,” he says, “and they’re both very, very good.”
Munday stands by his papers and plans to detail many “critically important” differences in the designs of the two sets of experiments in a response to the Nature paper. For instance, he notes the replication group didn’t study clownfish, used different water volumes and experiment durations, and used a different setup to study chemical cue avoidance. Dixson—who presented her findings at a 2015 White House meeting—also says methodological differences make a direct comparison between the studies “inappropriate.” But the Nature authors say some methods had to be adapted because they didn’t work as described in the original papers. They add that they could not catch enough clownfish, so used six other species also used in the previous studies.
Replication studies often cause quibbles about methods, Parker says. But, he argues, “If the original finding is reasonably robust,” then researchers using even somewhat different methods should be able to replicate it. And he notes that the replication team went to great lengths to be transparent. Unlike the original authors, the team released video of each experiment, for example, as well as the bootstrap analysis code. “That level of transparency certainly increases my confidence in this replication,” Parker says.
Researchers say the Nature paper allays one fear about the impact of ocean acidification. But Josefin Sundin of Uppsala University in Sweden, the Nature paper’s last author, stresses that climate change still poses a serious threat to sea life. “If the oceans were as acidic as we have been testing, it would also be much warmer, and that’s a huge issue,” she says.
Although replication efforts have blossomed in psychology, biomedicine, and other fields, they’re still rare in ecology, says biologist Shinichi Nakagawa of the University of New South Wales in Sydney. The new paper “sets a great example,” says Nakagawa, who hopes it “will instigate and inspire more replication studies—not to prove previous results wrong but to make our science more robust and trustworthy.”