NEW PAPER ARGUES FOR REDUCED ATLANTIC MENHADEN NATURAL MORTALITY RATE

 

A recently-released paper, “Investigation of Atlantic menhaden mortality rates,” authored by Jerold S. Ault and Jiangang Luo, which appeared in the November 2026 edition of the journal Fisheries Research, has been drawing the attention of menhaden advocates.

The paper argues that the natural mortality rate for Atlantic menhaden used in the most recent benchmark stock assessment, M=1.17, was much too high, and that a rate of M=0.50 was probably a more realistic estimate, noting that

“This estimate was substantially lower than the value currently adopted in the Atlantic menhaden stock assessment and was consistent with previous assessment results, species life-history characteristics, and observed longevity.”

If the paper’s conclusions are correct, they would have very serious implications for Atlantic menhaden management, as a sharply reduced natural mortality rate would, once input to the population model, result in a significant reduction in estimates of Atlantic menhaden fecundity (the overall number of eggs produced, which the menhaden assessment uses in lieu of spawning stock biomass to gauge menhaden abundance) and a corresponding increase in the fishing mortality rate.  Stock status would almost certainly change from the current “not overfished” and “not experiencing overfishing” to “overfished” and “experiencing overfishing,” and that change would justify the Atlantic States Marine Fisheries Commission’s Atlantic Menhaden Management Board slashing menhaden landings and placing significant new restrictions on the menhaden fishery.

Thus, it’s not at all surprising that the various menhaden advocacy groups are calling attention to the paper, which effectively supports their claims that the Atlantic menhaden population is in trouble.  We see a group that calls itself “The Forage Fish Campaign” alleging that

“Atlantic menhaden are likely collapsing after years of overharvest,”

and other advocacy groups making similar assertions.  For them, the Ault paper is pretty much a dream come true.

So what does the Ault paper actually say?

Its primary approach deals with a coastwide mark-recapture study—that is, a tagging study—undertaken by the National Marine Fisheries Service during the years 1966 through 1971, which placed magnetic tags in more than one million menhaden, and the efficiency of the magnets used to detect those tags in fish caught by the menhaden reduction industry. 

Inefficient magnets, which failed to detect many of the tags present in recaptured menhaden, would in theory lead to the false conclusion that a larger proportion of the tagged fish succumbed to natural mortality, either eaten by a predator or, perhaps, being a part of one of the mass die-offs that occur from time to time, when too many menhaden crowd into a harbor or back bay, then die when a falling tide, and perhaps elevated water temperatures, leave them with too little oxygen to survive.

Such false conclusion would then lead to an inaccurate stock assessment for, as the paper notes,

“Accurate mortality estimates are essential because stock assessment outcomes, including estimates of biomass, fishing mortality, biological reference points and sustainable harvest levels, are highly sensitive to assumptions regarding natural mortality.”

When the benchmark stock assessment adopted the natural mortality estimate of M=1.17 in 2020, a level that, the Ault paper states,

“represents a 2.3-fold increase relative to the value of M=0.50 yr^-1 used in the previous [2015] benchmark assessment and was approximately 14 standard deviations above the mean of more than a dozen previously published estimates, as well as natural mortality values used in recent U.S. and international stock assessments [citations omitted]”

that high estimate of natural mortality

“contributed to the [benchmark assessment’s] conclusion that Atlantic menhaden spawning biomass exceeded 80% of unfished levels.”

It also led to the benchmark assessment’s conclusion that Atlantic menhaden fecundity was above its target level, while fishing mortality was below its target, which is about as favorable a conclusion as a stock assessment can reach.

Unfortunately, it was almost certainly the wrong conclusion, as we learned when the 2025 Atlantic Menhaden Stock Assessment Update was released about one year ago.  In presenting its findings, it stated that

“The only new change for this update assessment is inclusion of a new vector of natural mortality based on a revised analysis of the historical tagging data that was completed by the M Working Group.”

But that change was important.  After taking another look at the magnet efficiency question, the biologists performing the assessment settled on a lower estimate of natural mortality, M=0.92.  That, in turn, was enough to substantially reduce the estimate of fecundity, and increase the estimated fishing mortality rate.  As a result, the status of the stock changed “not overfished” and “not experiencing overfishing,” with fecundity above the target level and fishing mortality below target, to the same “not overfished” and “not experiencing overfishing,” but with both fecundity and the fishing mortality rate somewhere between their target and threshold levels.  In the case of fecundity, that level was only about 5% above the threshold that defines an overfished stock.

Now, the Ault paper argues that the stock assessment update still significantly overestimated the natural mortality rate, saying

“the revised analysis continued to rely on aggregated, landings-weighted estimates of magnet efficiency derived from plant-test experiments.  Magnet efficiencies varied substantially among plants and over time, ranging from 0% to 100%, indicating highly heterogeneous, strongly non-normal distributions of tag-detection efficiency that are not adequately characterized by simple averages.  Because tag-recovery probabilities directly influence estimation of natural mortality, fishing mortality, and catchability, the use of aggregated efficiency values may obscure important spatiotemporal variation and contribute to parameter confounding.  These observations suggested that treating magnet efficiency as a fixed external input could contribute more uncertainty to mortality estimation than the movement or catchability components of the model.  Consequently, despite reducing the estimate of natural mortality from M=1.17 to M=0.92 yr^-1, the revised estimate remained higher than values suggested by Atlantic menhaden longevity, life-history characteristics, and previous stock assessments.”

With those issues in mind, the researchers Ault and Luo began recalculating the natural mortality rate, using two different approaches, one based on the menhaden’s life history, one based on a reexamination of the tagging data.

The issue of magnet efficiency was a large part of the latter reexamination.  The paper explains that

“Prior to the mark-recapture study, high-strength magnets were placed throughout the processing lines of participating reduction plants to remove metal debris from the fish meal and prevent damage to processing equipment.  The tagging study assumed that tagged and untagged fish were fully mixed within the population prior to capture.  Following harvest by commercial purse-seine vessels, fish landings were delivered to reduction plants for processing, where fish bearing implanted ferromagnetic tags could be detected and recovered by the plant magnets.  Consequently, the probability that a tag was recovered and reported depended directly on the efficiency of the plant magnet system.  Because magnet efficiencies varied among plants among plants and through time, recovery probabilities were imperfect and heterogeneous.  Accurate estimation of magnet efficiency was therefore essential because recovery probabilities directly influence estimates of survival, fishing mortality, and natural mortality in mark-recovery analyses.  [citations omitted]”

Much of the paper was dedicated to the models used to determine magnet efficiency and thus a better estimate of natural mortality; they were, as one might expect, highly reliant on statistical analysis and other elements of mathematics that probably need not be discussed in detail here.

However, the discussion of the two analytical approaches taken by the researchers is worth examining. 

“Model-based estimates of natural mortality should therefore be evaluated in the context of independent biological evidence, including observed longevity and age structure.  Liljestrand et al (2019) did not reconcile their mortality estimate with the documented age structure of Atlantic menhaden.  Even after corrections reduced the estimate to M=0.92 yr^-1, the implied maximum lifespan remained only approximately 5-6 years under natural mortality alone.  This prediction is inconsistent with historical observations of Atlantic menhaden attaining ages of 11-12 years in fishery and survey samples.  Discrepancies between model-derived mortality estimates and independently observed longevity warrant careful evaluation, as natural mortality estimates should be both biologically plausible and statistically supported.

“Our reanalysis produced a natural mortality estimate of M=0.50…, less than half the original estimate reported by Liljestrand et al (2019).  After correcting the input data, the principal source of this difference was not the fishing-effort data themselves but the methodological treatment of tag-recovery efficiency.  The resulting estimate is consistent with observed longevity, life-history characteristics, and previous stock assessment results for Atlantic menhaden, making it more biologically plausible than the substantially higher estimates adopted in recent assessments.  Given the strong influence of natural mortality on stock assessment outcomes, these findings have important implications for future stock assessments and the biological reference points, stock status determinations, and fisheries management decisions that depend on them.  [citations omitted]”

It is certainly a compelling argument, and as noted at the start of this essay, the menhaden advocacy community has been making some effort to have the Ault paper deemed the definitive analysis—what in a federal context might be considered the “best scientific information available”—with respect to the Atlantic menhaden’s natural mortality rate.

But in doing so, the advocacy community might be falling victim to a collective sort of confirmation bias, given that the Ault paper supports their long-held contentions about the health of the menhaden stock.  For while, in the Ault paper, we have a peer-reviewed study that makes an argument for a natural mortality rate of M=0.50, we also have a peer-reviewed stock assessment that rejects the essential premises and findings of the Ault paper, and sets the natural mortality rate at M=0.92.

At first glance, that might seem unlikely, given that the Ault paper was only recently published.  But if we look at the 2025 stock assessment update, we will find that those premises and general findings had already been established in 2023, although not in their final form.

Thus, the assessment update tells us,

“Ault et al (2023) submitted a working paper to the Atlantic menhaden [Stock Assessment Subcommittee] and the Ecological Reference Points Work Group (ERP WG) that re-analyzed the historical tagging data and produced an estimate of M=0.56, significantly lower than the M=1.17 reported by Liljestrand et al (2019).

“However, Ault et al (2023) had used a different subset of the data and a different approach to handling the parameters, which made direct comparisons with Liljestrand et al (2019) difficult.  The [Stock Assessment Subcommittee] formed a working group to review the datasets and methods in consultation with the primary authors to determine the best estimate of M for use in the Atlantic menhaden stock assessment.  The M [Working Group] and the [Stock Assessment Subcommittee] determined that the main cause of the difference in the M estimate was the handling of the magnet efficiency parameter, which was equivalent to the tag reporting rate in conventional tagging models.  The M WG and SAS found that Liljestrand et al (2019) had overestimated the magnet efficiency rate in their analysis, but did not agree with the stepwise estimation approach proposed by Ault et al (2023) to estimate this parameter.  In the end, the M WG and SAS recommended a revised estimate of M=0.92 from the tagging study, based on the corrected magnet efficiency rate and updated effort and landings datasets, which was lower than the value used in the 2020 benchmark, but higher than the value estimated from Ault et al’s (2023) method.  [emphasis added]”

Thus, it appears that the conclusions reached in the Ault paper do not represent a completely new approach to estimating Atlantic menhaden natural mortality, bur rather an approach that was considered and, after consultations with Dr. Ault, rejected by the Stock Assessment Subcommittee that prepared the assessment update.

That context matters, because such past consideration suggests that the Ault paper does not reflect a new revelation that necessarily resets the natural mortality estimate, but rather an approach that was considered and ultimately rejected before the 2025 assessment update was finalized.

That doesn’t mean that the ultimate conclusions in the Ault paper are wrong, but it does mean that Ault and Luo are offering an alternative estimate of M which does not automatically trump the M=0.92 estimate that was used in the 2025 assessment update; Ault and Luo must still convince the Stock Assessment Subcommittee that their estimate is the correct one.

Having said that, an argument can be made that using the Ault-Luo estimate of M=0.50 would be an appropriately precautionary action until further research either confirms or disproves its accuracy.  However, deeming the M=0.50 estimate to be the best scientific information available would be premature.

So where does that leave us?

The Ault paper was a comprehensive reanalysis of the data used to calculate natural mortality for Atlantic menhaden.  The arguments made in the paper are robust, and make a logical case for the lower natural mortality estimate. 

However, the Ault paper should not be taken as Gospel.

Its core concepts have been considered and ultimately rejected by the Stock Assessment Subcommittee, for reasons stated in the 2025 assessment update.  Unless and until its authors can successfully refute the Stock Assessment Subcommittee’s criticisms, the Ault paper should not be used as a primary driver of menhaden management measures, although its findings may well inform the discussion and debate about menhaden management issues.

It is a paper that seeks to shed light on the natural mortality question, but it does not represent the final word on the subject.  More research still needs to be done.