The federal Endangered Species Act instructs the U.S. Fish and Wildlife Service to determine whether to list a species as threatened or endangered solely on the basis of “the best scientific and commercial data available.” The Service has strived to make transparent the bases for those determinations, including through institutionalized Species Status Assessments (SSAs), which are intended to forecast a “species’ response to probable future scenarios of environmental conditions and conservation efforts,” providing “a single source for species’ biological information needed for agency decisions.”
Unfortunately, in its proposed rule to list as endangered the “distinct population segment” of longfin smelt that inhabits the San Francisco Estuary, the Service has abandoned its allegiance to using the best available scientific information, relying on an SSA that mobilizes facially inadequate data sets, ignores salient information gaps, and shrugs off technical peer-review criticisms that go to the heart of the argument for listing and protecting central California’s longfin smelt as endangered.
The Service explains that an “SSA provides decision makers with a scientifically rigorous characterization of a species’ status and the likelihood that the species will sustain populations, along with key uncertainties in that characterization.” But the agency does not do that in the Bay-Delta Longfin Smelt SSA. While the recently released assessment presents a superficially extensive compendium of knowledge about the Bay-Delta longfin smelt, a closer and critical look finds a narrative that employs unreliable data, presents results from analyses that cannot be justified, and makes troubling predictions for the fate of the fish that are built on flawed assumptions akin to a house of cards.
The Service states in the longfin smelt SSA –
“All the best available field surveys for documenting long-term abundance trends indicate Bay-Delta longfin smelt numbers have substantially declined over time, with current relative abundance reflecting small fractions of the species’ historical relative abundance and representing a decline of three to four orders of magnitude over the course of available historical abundance records.”
To be clear and unequivocal, piece-meal data from decades of interagency fish surveys that encounter longfin smelt essentially as bycatch offer no reliable quantitative information on the status and trends in the size of the longfin smelt population in the San Francisco Estuary. The Service relies on long-standing fish surveys to characterize the status of longfin smelt and the role of environmental stressors that put it at risk. Those surveys were not explicitly designed to monitor longfin smelt. As a result, the survey data cannot provide reliable information to policy makers and resource managers.
The survey trawls upon which the SSA and the proposed rule are wholly dependent fail to provide an adequate basis to draw inference regarding longfin smelt abundance for two reasons. The first, is because the fish surveys do not sample across the full Bay and Delta range of the longfin smelt, including the nearshore ocean outside the Golden Gate. A census population estimate can only be generated from a “closed population” — a population that is not subject to immigration or emigration, so that the population’s changes in size occur through births and deaths. If survey samples are taken from an undetermined fraction of a species’ geographic distribution, fish will inevitably be missed and the size of the population underestimated, potentially grossly so. The second, is because the fish surveys do not sample the habitat strata (or habitat types) that are occupied by longfin smelt within its Bay and Delta range. The fish surveys sample large areas of the Bay and Delta not occupied by longfin smelt and miss habitat areas occupied by longfin smelt at the times they are resident there, both during the fish’s nearly two-year rearing period and when they assemble for spawning runs.
For threatened and endangered species, abundance is the key state variable for purposes of conservation planning because of its basic relationship to a species’ persistence probability and its geographic distribution. Abundance ratios from sampling over time are the fundamental metric for estimating population growth rate. Importantly, recovery criteria for species listed under the Endangered Species Act (ESA) routinely are expressed in terms of abundance and distribution. Yet, the Service failed to provide even the roughest estimate of abundance due to the unreliability of survey trawl data; they can’t be used to draw inferences about longfin smelt numbers. Without adequate data on the longfin smelt’s abundance, how can the Service determine whether the fish is endangered and “is in danger of extinction throughout all or a significant portion of its range,” as endangerment is defined in the ESA? The answer is the agency can’t.
A conservation biologist, confronted by an absence of reliable information on a species’ abundance, looks to the distribution of the species across its range to infer whether the population is declining in numbers. It is always the case that a species in dramatic decline will have disappeared from portions of its historical range. But there is no sign of contractions in the distribution of longfin smelt in the Bay and Delta or emerging gaps in its distribution.
In fact, over the decade and a half since the Service received a first petition calling for longfin smelt to be protected under the ESA, the fish has been found in dozens of locations outside of the footprint of the agency’s fish surveys. Longfin smelt have been recorded in and around every tributary to San Francisco Bay that has been sampled and it has been found to spawn in the western Delta during all water year types — that is, years ranging from wet to critically dry. It has even been found reproducing successfully in recently restored salt marshes in the south Bay.
There is every reason to believe the longfin smelt has always been found throughout the San Francisco Estuary and spawns at times in all named freshwater inputs to San Francisco and San Pablo bays and in brackish and freshwater circumstances throughout the western and central Delta. Importantly, no locations in Bay-Delta from which longfin smelt historically were sampled are now known to be unoccupied. The current distribution of longfin smelt belies the contention that it requires protection under the ESA.
The absence of time-series abundance estimates for longfin smelt did not deter the SSA’s authors from trying to forecast “the species’ response to probable future scenarios of environmental conditions and conservation efforts,” as required by the Service. The SSA presents a population viability analysis (PVA), the de rigueur risk-assessment method used in conservation biology for assessing the risk of a population going extinct over a future period under varying environmental-stressor scenarios. Population viability analysis requires population-size estimates, which for longfin smelt are not available. Says the SSA —
“A count-based PVA is classically applied to census data (counts of an entire population), but it is not necessary to count the entire population. A count-based PVA can also be applied to index values, where a population index represents some portion of the total population as long as the proportion of the population that is observed remains relatively constant over time…”
That is not true. Index values cannot be used in PVAs. And, certainly not in the case of longfin smelt in the San Francisco Estuary, where the population index (abundance index value) is derived from a survey scheme that does not sample a knowable proportion of the total population nor collect data from the habitats and habitat strata occupied by the fish. Projections for the future state of longfin smelt from PVA models depend on abundance and distribution being reliably characterized, with acknowledgement of attending uncertainties. This is critical to predicting the future consequences to the species of habitat availability and resource conditions. Absent an estimate of the size of the longfin smelt population one cannot project its persistence under alternative environmental scenarios – including climate change. For PVAs for endangered species, abundance is the key state variable because of its fundamental relationship with survival over time in the face of environmental variation and emerging threats.
The longfin smelt SSA misconstrues the best available science directive in the ESA as allowing for lousy information to serve analytical purposes if it is the only information available. That is unacceptable, of course. The best available science for longfin smelt must be drawn from research, modeling, and especially monitoring, all of which must meet minimum criteria of reliability. Otherwise, it is not “scientific” information at all. The “abundance-index” values from the long-term fish surveys do not meet the most basic criteria for the reliable knowledge that should populate predictive demographic models or otherwise be used to guide resource management decision-making in an adaptive framework.
The Services’ statement in the proposed rule for longfin smelt that the “SSA report represents a compilation of the best scientific and commercial data available concerning the status of the DPS, including the impacts of past, present, and future factors (both detrimental and beneficial) affecting the DPS and its habitat” cannot withstand scrutiny. The longfin smelt SSA for the Bay-Delta distinct population segment of longfin smelt presents assertions and surmise as ecological facts, ignoring or waiving critical uncertainties aside as non-essential, and predicting futures for the fish from false assumptions. In so doing the Service ignores important empirical observations and necessary corrections from an extensive and incisive peer review of the draft longfin smelt SSA by reviewers selected by the agency itself.
Under these circumstances, the only responsible path forward for the Service in its decision-making for longfin smelt in the Bay and Delta leads back to the drawing board.