Dear MPA News:
In the lead article in the October 2002 MPA News ("Measuring the Effects of Marine Reserves on Fisheries: The Dilemmas of Experimental Programs"), Ray Hilborn raised concern about the design of appropriate experiments to evaluate the impact of marine reserves. He noted that most studies to date "have not involved sufficiently rigorous experimental design," and that current research on reserve effects suffers from two internal biases: (1) current protected areas were "almost certainly selected for protection because of their higher productivity," and (2) effort excluded from the reserve area will be "redirected to the unprotected areas," thereby decreasing measures of biological response (density, etc.) in the control site and inflating the perceived effect of reserve protection.
These are valid concerns, and we can address them using an existing database developed by Benjamin Halpern to synthesize results from studies that assessed the biological impact of marine reserves from around the world (Halpern, in press). Are reserves preferentially placed in areas of high productivity? There are many examples of reserves that were created for non-biological reasons, such as a location's proximity to a university (research), military base (buffer zone), or natural feature (aesthetic value for a park). Certainly some reserve locations were chosen because the sites were particularly productive, but given the current difficulty in creating reserves around any location deemed productive, and therefore good fishing grounds, it seems reasonable to assume that many reserves were placed haphazardly at best and in non-productive locations at worst. Furthermore, some case studies (e.g., Roberts et al., Science, 2001) show that despite the displacement of fishing effort due to reserve creation, catch size and value actually increased in fished areas adjacent to the reserve. In many cases, then, the spillover of fish to control sites may actually diminish the perceived enhancement of populations within reserves, even though fishing effort is displaced to areas outside the reserve.
One can test these ideas more rigorously, however, by analyzing data from studies of marine reserves that collected data in both a control and a treatment site, before and after the treatment (a before-after/control-impact, or BACI, experimental design). Of the studies included in the database mentioned above, eight were BACI designs. If reserves were placed in highly productive locations, then they should have higher initial values compared to control sites. The results from the studies are revealing, and do not indicate the biases suggested by Hilborn. Density, biomass, and average size of organisms were not significantly different in the reserve site compared to the control site before reserve creation. (In four studies, density in the control site was higher than, or not different from, the reserve site; in three cases, the reserve-site density was higher.) The same studies can also be used to assess whether declines occur in control sites due to redirected effort by comparing conditions in the control site before and after the reserve was created. Here the overall results are actually contrary to Hilborn's expectation, with density, etc. increasing in control sites after reserve protection. (For density, five studies showed higher values in the control site after reserve protection compared to before, while only one study showed the opposite trend.)
These results do not invalidate concerns about inadequate experimental design - most studies of reserves have only compared conditions inside and outside a reserve at one point in time. But the results from these better-designed studies suggest such concerns are probably overstated, and even suggest a bias against seeing a reserve effect.
As a final note, one of the greatest difficulties in designing adequate monitoring programs for assessing the impact of marine reserve networks stems from the very criteria used to design the networks. Individual reserves within a network need to be connected to each other so that the network is truly a network, but this connection between individual reserves means that all areas between the reserves will also benefit from the increased production, etc. coming from within the reserves. How, then, does one find a control site that is independent of the treatment so that the real reserve effect may be deduced? This is a serious concern, because the actual reserve effect is likely to be underestimated if control sites benefit from nearby reserves. Hilborn is correct that we need to be careful about how we design programs to evaluate reserve performance, but we believe the bias in such programs is more likely to be against than in favor of seeing a reserve effect.
E-mail: halpern [at] lifesci.ucsb.edu
E-mail: warner [at] lifesci.ucsb.edu
E-mail: gaines [at] lifesci.ucsb.edu
The letter writers are all from the Department of Ecology, Evolution and Marine Biology at the University of California at Santa Barbara (USA).
Editor's note: For a copy of Halpern's synthesis paper on the biological impacts of marine reserves, currently in press, please e-mail him at halpern [at] lifesci.ucsb.edu.