Medoff et al. 2022, a paper published in November 2022 in Science, claims that the Papahānaumokuākea marine national monument, the largest marine protected area (MPA) in the United States, caused a spillover effect in tuna—the holy grail of protected area advocacy. Spillover happens when species become so abundant in a protected area that they spill over into surrounding, unprotected areas benefiting the overall population. If true, Medoff et al. 2022’s claim would be significant evidence supporting the 30×30 movement and the creation of large, open ocean MPAs.
However, their science doesn’t stand up to careful analysis.
The statistical analysis used by Medoff et al. 2022 was flawed in two significant ways:
- The largest effect was observed in yellowfin tuna, but the MPA’s creation corresponded with a significant surge in yellowfin tuna population across the western Pacific. Furthermore, before the MPA was established, the catch per hook (CPUE, a measure of fish abundance) was substantially higher in proximity to the MPA than in farther locations, suggesting that habitat closer to the MPA was better than farther away.
When tuna catch changes, we expect distribution to change proportionally to habitat quality, but the Medoff et al. analysis used absolute changes in CPUE as opposed to relative or proportional changes; see Figure 3 below. Using this approach, Medoff et al. expected the CPUE in areas with very low CPUE to increase just as much as areas with high CPUE prior to the MPA establishment, about 1 yellowfin per 1000 hooks.
We reran Medoff et al.’s analysis with proportional measures instead of absolute ones and found no evidence of spillover.
- The Medoff paper also failed to mention or discuss the effect of the largest El Niño event in history that is likely responsible for the major change in yellowfin tuna abundance and CPUE.
In this post, we explain the history of tuna in the Papahānaumokuākea MPA and break down the methodology of Medoff et al. to show what led to inaccurate results.
What is a spillover effect?
Spillover theory suggests that when an MPA is established, the abundance of fish within the MPA increases significantly and eventually overflows into neighboring regions, benefiting both the fish population and the fishermen who lost access to their previous fishing locations. This theoretical “win-win” is a crucial argument in support of MPAs.
However, the real-world implementation of MPAs often results in the displacement of fishing pressure outside areas. Fishermen fish more aggressively outside the MPA to catch the same quantity of fish. Consequently, while the protected area may have more fish, the overall fish population remains unaffected. From a fisherman’s perspective, MPAs make fishing harder as more fishing effort is needed to achieve the same catch.
Medoff et al. claimed a spillover effect by analyzing tuna data from the Papahānaumokuākea MPA 3.5 years after it was expanded in 2016. They concluded that closing the area to yellowfin and bigeye tuna fishing led to higher CPUE (in this case, fish per 1000 hooks) outside the MPA. They observed that fishing closer to the MPA boundary saw bigger increases in CPUE than fishing farther away, concluding that the MPA was responsible.
What evidence is necessary to demonstrate causality?
Between 2016 and 2019, the CPUE of yellowfin tuna in the waters around Hawaii did increase. The question is, did it increase closer to the MPA than farther away? What drove the abundance and CPUE increase? Was it due to the expansion of the MPA? Or was it something else, like natural fluctuations in the tuna population or unusual climate conditions?
Before the MPA expansion in 2016, only about 1700 yellowfin tuna were caught each year in the area that would later be closed to fishing. Medoff et al. essentially claims that protecting those 1700 fish (out of an estimated population of ~48,000 in the MPA and 1-2 million in the surrounding area) directly led to a 60% increase in yellowfin CPUE near the MPA in just three years.
That is extremely unlikely. Putting aside the fact that those fish only represent ~3% of the population in the area, Yellowfin mature in 2-3 years. Those 1700 fish would have had just one or two year-classes of spawn mature enough to be caught. Even if all of those 1700 fish had stayed in the MPA, they would have increased the population inside the MPA by just 9%.
The increase in catch rate is probably better explained by the 2014-2016 El Niño event. Historically, El Niño events lead to much more tuna around Hawaii (more on this later).
What does the tuna abundance data show?
There is no doubt that tuna populations in the region increased dramatically during the study period, but the increase began in 2014, not after the MPA expansion in 2016. The scientific stock assessment of yellowfin tuna shows that both catch and CPUE around Hawaii have continued their upward trajectory since the closure in 2016.
The abundance of the yellowfin tuna population around Hawaii rose from 46,000 MT in 2014 to 74,000 MT in 2018, a 60% increase both near and far from the MPA boundary. See the chart below from 2014-2018.
Another flawed MPA model
How, then, did Medoff et al. conclude that the MPA itself caused an increase in the abundance of tuna around the MPA? Medoff et al. used confidential fishery observer data to model CPUE as a function of distance from the MPA boundary, but they built the model to be additive, not proportional. This means the model expected CPUE increases to rise by the same absolute amount—not by a proportionate amount.
For example, before the MPA expanded, bigeye tuna CPUE in the area close to the MPA boundary was ~1 (measured by tuna per 1000 hooks). Five hundred miles away, the CPUE was ~0.33. Medoff et al. 2022’s model controls for the possibility that both areas would see the same absolute change in CPUE: If the area near the MPA increased from 1 to 2 (an increase of 1), the far area would have to increase by 1 to be considered equal (from .33 to 1.33). The near area would double while the distant area would have to quadruple. See Figure 3.
In the real world, we expect fish populations to increase proportionally to what the local habitat allows. If the near area doubles from 1 to 2, we expect the far area to double from 0.33 to 0.66. By modeling in absolutes, Medoff et al. distorts the data in favor of areas close to the MPA where CPUE was already high and habitat was historically better. The increase in tuna in specific locations reflects an increase in proportion to its habitat quality.
Because the observer data is confidential, we asked the staff of the Western Pacific Regional Fisheries Management Council (who have access to the data) to rerun the yellowfin model using a proportional measure of CPUE changes instead of an additive measure.
We found no evidence for spillover:
In the figure above, the proportional model in open circles shows no sign of declining change in CPUE with distance from MPA; a pattern observed by the absolute model (used by Medoff et al. 2022) in black circles that shows a 1.3 CPUE increase closer to the MPA with a gradual decrease to 0.5 CPUE farther from the MPA.
The El Niño in the room
According to Dr. John Hampton, chief scientist at the Secretariat of Pacific Communities (SPC) and former head of the SPC Oceanic Fisheries Program—the intergovernmental organization tasked with monitoring tuna populations in the Pacific, the increase in tuna abundance and CPUE around Hawaii from 2014-2018 was probably due to the El Niño event.
Most of the tuna population in the Pacific lives in the Western Pacific (closer to Asia). But during El Niño years, warm water from the Western Pacific moves eastward towards Hawaii, bringing tuna with it.
This effect was first published in Nature in 1997. During the 2014-2016 El Niño, millions of tons of tuna moved closer to the Hawaiian Islands; the measured abundance of yellowfin tuna increased by 60% from 2014-2018 around Hawaii—quite a coincidence with the 2016 MPA expansion.
The framework of Medoff et al.’s model (called a difference-in-differences) is, in theory, supposed to control for coincidental environmental factors like an El Niño, but it cannot control for the reality of how a climate shock affects the spatial distribution of tuna.
Their model would have been able to control for it if the El Niño event had resulted in the same increase in the number of tuna both near and far from the MPA. However, given the strong gradient in historic tuna abundance in this region, it is more likely that more tuna arrived closer to the MPA in proportion to the better habitat there.
The scientific evidence for fish benefits of 30x30 remains weak
As the 30×30 movement grows in political popularity, research with inventive post hoc analysis (like Medoff et al. 2022) will be needed. Unfortunately for advocates of the movement, real-life evidence of spillover remains hard to find.
The Phoenix Islands Protected Area (PIPA), a similar MPA to Papahānaumokuākea in size and habitat just 1500 miles Southwest of Hawaii, also showed no effect of closing its waters to tuna fishing in 2015.
Before it was closed, fishermen caught an average of 22,000 metric tons (MT) of skipjack tuna and 1,900 MT of bigeye tuna in the area that would become PIPA. That’s ~500 times the amount of yellowfin tuna caught in the area that would become the Papahānaumokuākea marine national monument.
Hampton et al. 2023, a paper by John Hampton and the SPC tuna team, found no increase in tuna populations due to the MPA. After losing millions of dollars of fishing revenue over the past eight years, the Kiribati government announced they would reopen the PIPA to tuna fishing last year.
Another recent study, this of the largest network of MPAs in the U.S., off the coast of California, showed no increase in regional or overall populations of several fish species.
The implementation of MPAs frequently results in the displacement of fishing pressure rather than its reduction. Fishermen compensate for lost catch by intensifying their fishing activities outside of the MPA, particularly in the case of mobile fish like tuna that frequently move in and out of the MPA.
While MPAs can effectively protect marine life inside their boundaries if adequately enforced, there is insufficient evidence to suggest that they are effective at increasing large-scale populations outside of MPA boundaries.