Case Information
United States District Court
Middle District of Florida
Jacksonville Division
C ENTER FOR B IOLOGICAL
D IVERSITY , AND N OKUSE
E DUCATION , I NC .,
Plaintiffs, v. N O . 3:23-cv-936-WWB-LLL U.S. F ISH AND W ILDLIFE SERVICE ,
M ARTHA W ILLIAMS , IN HER OFFICIAL
CAPACITY AS D IRECTOR OF THE U.S.
F ISH AND W ILDLIFE S ERVICE , AND
D EB H AALAND , IN HER OFFICIAL CAPACITY
AS S ECRETARY OF THE U.S. D EPARTMENT
OF THE I NTERIOR ,
Defendants. ________________________________________________________________________
Order
Before the Court is Plaintiffs Center for Biological Diversity and Nokuse Education, Inc.’s Motion [to] Supplement the Administrative Record and Memorandum in Support, doc. 20; defendants’ response in opposition, doc. 22; and plaintiffs’ reply, doc. 23. Plaintiffs move to supplement the administrative record with a sworn declaration (the declaration) by Dr. Kevin Shoemaker. For the reasons discussed below, the motion is denied.
Background Plaintiffs initiated this action by filing their complaint in August 2023 against defendants U.S. Fish and Wildlife Service, Martha Williams, and Deb Haaland. Doc. 1. Plaintiff Center for Biological Diversity (CBD) is a national, nonprofit conservation organization that works to protect endangered species and their habitats. ¶ 12. CBD is incorporated in California and headquartered in Tucson, Arizona. Plaintiff Nokuse Education, Inc. is a nonprofit education and conservation organization, which operates the E.O. Wilson Biophilia Center in Walton County, Florida. ¶ 13. The complaint alleges that defendants wrongfully denied the gopher tortoise endangered or threatened species status, doc. 1 ¶¶ 105-112, and requests declaratory and injunctive relief, doc. 1 at 35.
The gopher tortoise is a terrestrial turtle living in the southeast United States from the eastern tip of Louisiana to the southern tip of South Carolina and through much of Florida. Id . ¶ 2. The gopher tortoise is dark brown to grayish black, has a domed shell, elephantine hind feet, and shovel-like forelimbs evolved for digging burrows; it is the only native tortoise east of the Mississippi River. ¶ 42. The gopher tortoise lives an average of 50-80 years, although plaintiffs allege gopher tortoises are highly vulnerable to threats because of “their long lifespans, late age of reproductive maturity, and low reproductive output, along with a high hatchling and juvenile mortality rate.” Id. ¶ 46. Because adults often survive threats better than hatchlings, plaintiffs allege that the “enduring presence of adults can mask declines and imminent population extirpations related to poor or no reproduction.” Id . The gopher tortoise’s habitats include upland forest, savanna, grassland, and coastal dunes. Id . ¶ 1. Plaintiffs seek judicial review of the U.S. Fish and Wildlife Service’s (FWS) finding that listing the gopher tortoise as endangered or threatened under the Endangered Species Act was not warranted.
Defendants answered the complaint in October 2023, doc. 9, the Court then issued a Case Management and Scheduling Order (CSMO) in November 2023, doc. 11. At the parties’ request, the Court stayed the CMSO so the parties could work out the contents of the administrative record. Docs. 15, 17. The parties represented that they “reached agreement regarding the inclusion of 51 records in an amended Administrative Record . . . [but] [t]he parties were unable to reach agreement regarding supplementation of the Administrative Record . . . .” Doc. 18 ¶¶ 17, 18. Plaintiffs now move to supplement the administrative record with the declaration by Dr. Kevin Shoemaker that “explains technical terms and complex subject matter in this case relating to the Service’s population viability modeling for the gopher tortoise.” Doc. 20 at 7.
Authority Judicial review of an agency action is governed by the Administrative Procedure Act (APA). 5 U.S.C. § 706. Section 706 of the APA requires a district court to review the administrative record. Preserve Endangered Areas of Cobb's History, Inc. v. U.S. Army Corps of Eng'rs , 87 F.3d 1242, 1246 (11th Cir. 1996) (“The focal point for judicial review of an administrative agency's action should be the administrative record.”). Because an agency presumably knows the content of the record the agency considered, an agency's certification of the completeness of the administrative record receives a measure of presumed correction. Alabama-Tombigbee Rivers Coal. v. Kempthorne , 477 F.3d 1250, 1262 (11th Cir. 2007). Although—as discussed more fully below—the applicability of exceptions permitting extra-record evidence in the Eleventh Circuit is not settled, the Ninth Circuit allows a reviewing court to “go beyond” the administrative record if “(1) an agency's failure to explain its actions effectively frustrates judicial review; (2) it appears the agency relied on materials or information not included in the administrative record; (3) technical terms or complex subjects need to be explained; or (4) there is a strong showing of agency bad faith or improper behavior.” Cobb's History , 87 F.3d at 1247 n.1 (citing Animal Def. Council v. Hodel , 840 F.2d 1432, 1437 (9th Cir. 1988)). “[Al]though certain circumstances may justify the district court going beyond the administrative record, it is not generally empowered to do so.” Ala.-Tombigbee Rivers , 477 F.3d at 1262.
Analysis Plaintiffs move to admit Dr. Shoemaker’s declaration as extra-record evidence. [1] They argue that the declaration explains the complex subject of population modeling and associated technical terms, relying on the “technical terms or complex subjects need to be explained” exception to the general rule that judicial review should be of the administrative record. Doc. 20 at 11-12; Cobb's History , 87 F.3d at 1247 n.1. Defendants counter that plaintiffs have failed to establish a legal basis for the Court to look beyond the administrative record and that the administrative record contains all the information needed for the Court to review FSW’s decision. Doc. 22 at 8-15.
A. The Declaration and its Findings Plaintiffs retained Kevin T. Shoemaker, Ph.D., to “explain technical terms and complex subject matters in this case relating to population viability modeling for the gopher tortoise.” Doc. 20-1 ¶ 1. Dr. Shoemaker is an Associate Professor of Wildlife Population Ecology in the Department of Natural Resources and Environmental Science at the University of Nevada, Reno. ¶ 4. He has a bachelor’s degree in biology from Haverford College, and an M.S. and Ph.D. in Conservation Biology from State University of New York, College of Environmental Science and Forestry. ¶ 3. Dr. Shoemaker has served as principal investigator for 18 funded research grants, 5 of which directly focused on tortoise population ecology. ¶ 8. Additionally, Dr. Shoemaker has published more than 50 peer-reviewed publications, 10 of which are complex, technical aspects of the Service’s PVA model.” Doc. 20 at 10. Thus, I construe plaintiffs’ motion as one to admit extra-record evidence and apply the corresponding standard. See SOSS2, Inc. v. U.S. Army Corps or Eng’rs, 403 F. Supp.3d 1233, 1237 (M.D. Fla. 2019).
focused on turtle and tortoise ecology. Id. Further qualifications are listed in his curriculum vitae. See doc. 20-1 at 19-46.
In the declaration, [2] Dr. Shoemaker represents that the scientific grounds for the FWS’s decision, which was documented in a Species Status Assessment (SSA), were largely derived from a rangewide demographic simulation model that was published in the scientific journal Global Ecology and Conservation (Folt et al. 2022). AR 003442– 73, 003523–59 (SSA); AR 011116–36 (Folt et al. 2022). Id ¶ 13. Dr. Shoemaker declares that while the published version of the SSA model had minor differences from the model used in the FWS’s SSA document, they are mostly identical, and he refers to both versions in the declaration as the “SSA model.” Id. FWS used the SSA model to predict population growth and extinction risk for the gopher tortoise. ¶ 14. The SSA model forecasts annual gopher tortoise population size across the range of species for 80 years, 2021 to 2100, under different threat scenarios. FWS included 656 distinct local populations in its SSA model; FWS defines local populations as aggregations of tortoises readily accessible to one another for the purposes of reproduction. The local populations were then grouped into 253 landscape populations, which represent sets of local populations situated close enough to one another so that occasional movement of individuals among local populations was possible.
The SSA model used by FWS analyzed four threats: 1) climate warming, 2) sea level rise, 3) urbanization’s effects on tortoise immigration and habitat management, and 4) climate-change effects on habitat management. Id. ¶ 15. FWS then described six scenarios, which represent various combinations of the threats in addition to varying management responses and alternative plausible representation of gopher tortoise life history. Id. For each scenario, the SSA model was run 50 times. Id. Finally, the conservation outlook for the gopher tortoise under each scenario was summarized by computing statistics such as extinction risk and the average range-wide abundance and number of extant populations at year 2100. Id.
Dr. Shoemaker explains that the SSA model’s output depends on its specification and implementation, in addition to informational inputs. ¶ 16. Dr. Shoemaker evaluated the SSA model and “identified inconsistencies between how the simulation model works and USFWS’s own findings about gopher tortoise ecology and life history.”
The SSA model predicted considerable population declines and local extinctions over an 80-year period across the four threat scenarios. ¶ 17. The model also predicted that a subset of populations would remain large and self-sustaining over this time period, which led FWS to conclude that “the future condition of the species with relatively large numbers of individuals and populations suggests resiliency to withstand stochastic environmental and demographic change, and redundancy to buffer from future catastrophic events.” Dr. Shoemaker opines that FWS’s SSA model erred in evaluating two factors: 1) immigration and 2) the maturation rate between juvenile and adult tortoises. See id. ¶¶ 18-32.
Immigration
In the declaration, Dr. Shoemaker opines that “the way the SSA model handled immigration (movement of tortoises within the ‘landscape populations’ defined in the SSA) is inconsistent with common modeling practices and introduces a strong positive feedback process that results in unrealistic runaway population growth for a substantial subset of populations.” Id. ¶ 18. The SSA defines immigration as the movement of tortoises into local populations from other local populations within a landscape population. Id . Dr. Shoemaker further opines that the “method used for modeling immigration in the SSA model results in the “immigration” process having a much stronger positive impact on local and regional population viability than USFWS or the modeling team likely intended[.]” Id. He refers to such unintentional consequences as “artifacts” in his declaration. Id.
Dr. Shoemaker noted that population growth in the SSA model was sensitive to the immigration rate. ¶ 19. When the immigration rate was set to 4%, the total range wide population size grew on average 11%; when the immigration rate was set to zero, range wide population size decreased by 98%. Dr. Shoemaker surmises that this result is an artifact of the method used to account for immigration in the SSA model. He represents that in models like the SSA, immigration should be a zero- sum process. ¶ 20. That is, individuals added to one local population should be removed from a different local population within the landscape population, which would result in no net loss or gain of individuals within the landscape population. Id.
According to Dr. Shoemaker, immigrants in the SSA model are not derived from other populations within the landscape population, but rather from theoretical structures, which he refers to as “dummy” landscape populations. Id. The dummy landscape populations are intended to represent the available pool of immigrants into a local population but are “often orders of magnitude larger than the true number of potential immigrants.” Id. Dr. Shoemaker therefore opines that “the immigration process inadvertently (and incorrectly) serves as a major source of new individuals to some subpopulations, often exceeding natural reproduction as the primary source of new tortoises.” Id.
The SSA model pairs each local population with a dummy landscape population, which represents the potential number of adults present in neighboring populations. Id. ¶ 21. Each dummy landscape population is assumed to grow or decline in lockstep with its respective local population. Specifically, each dummy population “grows in proportion to the ratio of adults in the local population from one year to the next[.]” Dr. Shoemaker represents that because each dummy landscape and paired local population are modeled independently from all other pairs, the model effectively decouples from one another local populations within the same landscape population. He continues that “this results in the logical and physical impossibility that dummy landscape populations meant to represent the same cluster of populations can be both growing and declining at the same time.”
According to Dr. Shoemaker, this dummy landscape model creates a positive feedback loop that “drastically inflates population growth rates for many local populations.” Id. ¶ 22. These increases in the number of adult tortoises in a population cause a proportional increase in the paired dummy landscape population, which leads to a larger influx of immigrants to the local population; “high immigration in one year causes even higher immigration in the following year, which causes even higher immigration rates in subsequent years[.]” Id. Once this positive feedback loop begins, it can result in “artificially inflated numbers of immigrants arriving each year from the dummy landscape population throughout much of the 80-year simulation.”
Dr. Shoemaker notes that in the SSA model immigration is often the primary driver of local population growth. Id. ¶ 23. He supposes that this positive feedback loop is illogical because landscape populations are closed to immigration; in other words, it does not make sense that a doubling of the local population, driven by immigration, would cause a doubling of the landscape population, which is closed to immigration. He cautions that this feedback loop creates individual dummy landscape populations that greatly exceed the combined abundance of all the populations within the landscape population, which is impossible because dummy landscape populations are intended to represent the pool of potential dispersers within the landscape population. ¶ 24. He notes that some dummy landscape populations grew from thousands to millions within a single tortoise generation, which is impossible for a species that is long-lived, slow-growing, and slow to reach maturity. Dr. Shoemaker represents that because the SSA model put no limit on dummy landscape population size, they sometimes exceed the 2 females per hectare limit found by FWS.
Dr. Shoemaker finds that “[v]irtually all of the populations that the SSA model predicts to be large and stable (or growing) by the end of the century are, in fact, propped up by hyper-inflated immigration from unrealistically large dummy landscape populations, while all other populations are projected to experience moderate to severe declines.” ¶ 25. He cites local population “ID 110” as an example, which begins the simulation with an average of 22 tortoises within a landscape population of approximately 4,200 tortoises:
In the first 20 years of the simulation, the population grows, on average, from 22 to ~1,000 individuals and the landscape population grows, on average, from 4,200 to ~200,000 individuals, a size more than double the starting number of individuals in the entire simulation . . . . This results in a density of approximately 29.7 females/ha across the entire 6727 acres of the landscape population, inconsistent with the maximum of 2 females/ha assumed by the model, AR003446. This population always experiences this positive feedback-driven growth in all runs of the simulation, and therefore is predicted to have an extinction probability of zero; it is thus designated as one of the populations having the highest persistence probability (“extremely likely to persist” per Folt et al. 2022), despite starting with only ~20 individuals. See AR 011128 (defining “extremely likely to persist”). After examining model results, Dr. Shoemaker found that many of the populations
classified as “extremely likely to persist” in the SSA start small and grow exponentially due to this positive feedback loop, “meaning they are in reality much more prone to extinction than indicated by the model.” Id.
Additionally, Dr. Shoemaker modified the SSA model by imposing a 3% cap on annual growth for all dummy landscape populations, which he suggests is the “highest rate of growth that a gopher tortoise landscape population would be able to realistically sustain[.]” Id. ¶ 30. With this modification, “total rangewide abundance declined from ~70,000 to ~2,000 individuals” compared to a final abundance range of ~20,000 to ~47,000 in the SSA model. Id. He represents that “misspecification of immigration rates . . . resulted in a significantly inflated population projection, roughly 10 times larger than what the projected population would be with the errors rectified.” Id.
Maturation Rate Between Juvenile and Adult Tortoises Dr. Shoemaker continues the declaration by noting that the “model specified the maturation rate between juvenile and adult tortoises in a way that does not reflect gopher tortoise life history.” ¶ 26. The SSA model uses the “flat age-within-stage structure” to estimate the percentage of juveniles that are eligible to transition to adulthood each year. Id. ¶ 27. This approach assumes an equal number of individuals in each age within the juvenile stage. In other words, the model assumes there are the same number of 1-year-olds, 2-year-olds, 3-year-olds, and so on for ages within the juvenile stage.
Dr. Shoemaker counters that this is rarely the case is real world populations. In most real cases, “a multi-year juvenile stage class is expected to contain fewer old juveniles than young juveniles because older juveniles face high rates of mortality and have had to survive for more years.” Id. The assumption that juveniles all experience the same annual survival rate overestimates the number of individuals that are old enough to mature into adults, which results in an “erroneously high population growth estimate.” Id. ¶ 28. Dr. Shoemaker suggests that a better approach would be to keep track of the number of individuals at each age until maturity; with this approach the number of maturation-eligible individuals is the number of individuals in the age class that is one year younger than the maturity age. Id. ¶ 29.
Dr. Shoemaker modeled juvenile age classes by keeping track of the number of individuals in each age class in the juvenile stage. Id. ¶ 31. He then allowed only the individuals one year younger than the age of maturity to mature into adults in a given time period. Id. This modification, combined with the immigration modification, “resulted in a further reduction in the predicted number of persisting individuals and populations—to near extinction (approximately 200 individuals across all populations).”
Finally, Dr. Shoemaker notes that the Folt et al. 2022 peer-reviewed article and the SSA model differ “substantially in the predicted number of tortoises persisting after 80 years despite virtually identical model structure[.]” ¶ 32. For example, for the low management and medium threats scenario, the SSA model predicts ~47,000 individuals, while the Folt article predicts ~20,000 individuals. Dr. Shoemaker concludes that “virtually all of the original conclusions derived from the SSA model regarding gopher tortoises’ future status across their range are artifacts of modeling errors and are primarily driven by the unintended feedback process and the maturation rate error.”
B. Extra-Record Evidence Plaintiffs argue that consideration of the declaration is appropriate because it is necessary to assist the Court in understanding “technical aspects” related to the administrative record. Doc. 20 at 5-6. Plaintiffs cite Miccosukee Tribe of Indians of Florida v. U.S. , No. 95-0533-CIV-DAVIS, 1998 WL 1805539 (S.D. Fla. September 14, 1998) and Sierra Club v. United States Forest Service ¸ 535 F.Supp.2d 1268 (N.D. Ga. 2008) to support their position. Doc. 20 at 14-15. In Miccosukee , the Court considered two extra- record depositions in reviewing an Environmental Protection Agency decision because “they provide a better explanation of the EPA's decision, and because they helped clarify complex, technical terms dealing with water quality criteria that were integral to the EPA's conclusions.” 1998 WL 1805539 at *14-15. However, the Court provides little analysis and information regarding the contents of the depositions.
In Sierra Club , the Court considered extra-record testimony in reviewing a U.S. Forest Service decision regarding the meanings of the terms “inventory” and “population inventory information.” 535 F.Supp.2d at 1291. In doing so, the Court cited Cobb’s History and concluded that it was “both necessary and appropriate to consider testimony not contained in the administrative record . . . because of these terms' technical nature.”
Defendants counter that Coin Center v. Yellen , No. 3:22cv20375-TKW-ZCB, 2023 WL 2889736 (N.D. Fla. April 10, 2023) is more analogous to this case. Doc. 22 at 9. In Coin Center , the Court considered a motion to “supplement” the administrative record with the declaration of a purported expert in cryptocurrency technology. 2023 WL 2889736 at *1. The Court discussed the applicability of the technical-terms exception in the Eleventh Circuit and noted “[t]he district court may only go beyond the administrative record where there is initially ‘a strong showing of bad faith or improper behavior’ by the agency.’” Id. (citing Ala.-Tombigbee Rivers , 477 F.3d at 1262) (quoting Citizens to Pres. Overton Park, Inc. v. Volpe , 401 U.S. 402, 416 (1971)). The Coin Center Court denied the motion because 1) plaintiffs failed to show that defendants acted in bad faith or engaged in improper behavior, and 2) even if the exception was recognized in the Eleventh Circuit, the declaration was not “necessary to explain technical terms or complex subject matter.” at *2-4.
Defendants argue that the Coin Center Court correctly concluded that the technical-terms exception has not been recognized in the Eleventh Circuit. Doc. 22 at 9. Although Cobb’s History lists four exceptions recognized by the Ninth Circuit, see supra pg. 4, it specifies that “[w]e need not consider these exceptions as none of them apply in the instant case.” In subsequent decisions, the Eleventh Circuit has only recognized the bad faith exception. See Nat'l Min. Ass'n v. Sec'y, U.S. Dept. of Labor , 812 F.3d 843, 875 (11th Cir. 2016) (“We have acknowledged that various factors could be considered in determining the propriety of reviewing extra-record material on review of an agency rule . . . ; in practice, however, we generally have focused pointedly on whether the petitioners have made a strong showing of bad faith or improper behavior by the agency.”) (internal citation and quotation marks omitted); Ala-Tombigbee Rivers , 477 F.3d at 1262 (explaining that the district court should go beyond the administrative record “only where there is initially ‘a strong showing of bad faith or improper behavior’ by the agency”) (quoting Overton Park v. Volpe , 401 U.S. 402, 420 (1971)); see also Citizens for Smart Growth v. Peters , No. 07-14122-CIV-MARTINEZ-LYNCH, 2008 WL 11331898, at *2 n.3 (S.D. Fla. Sept. 24, 2008) (concluding that in light of Ala- Tombigbee Rivers , the Eleventh Circuit in Cobb’s History did not adopt and has not since recognized the Ninth Circuit's exceptions to the record rule); Marllantas, Inc. v. Rodriguez , 806 F. App'x 864, 867 (11th Cir. 2020) (finding no error in the district court's decision not to allow discovery to supplement the administrative record produced by the agency because the plaintiff did not make “a strong showing of bad faith or improper behavior”).
I find the Coin Center Court’s analysis prudent and the case analogous to the one before this Court. Plaintiffs do not allege, much less make a showing, that FSW acted in bad faith. Instead, plaintiffs solely rely on the technical-terms exception, which has not been explicitly recognized in the Eleventh Circuit. Because plaintiffs fail to allege FSW acted in bad faith, I find plaintiffs lack a legal basis to admit the declaration as extra-record evidence.
Additionally, even if the technical-terms exception was good law in the Eleventh Circuit, the declaration does not fit such an exception. Plaintiffs argue that the declaration “explains technical terms and complex subject matter in this case relating to the [FWS’s] population viability modeling for the gopher tortoise.” Doc. 20 at 7. While the declaration discusses some technical terms, the thrust of the declaration, as detailed above, is an opinion rebutting the FSW’s findings. See doc. 20- 1 (identifying “inconsistencies between how the simulation model works and USFWS’s own findings about gopher tortoise ecology and life history”; that “[v]irtually all of the populations that the SSA model predicts to be large and stable (or growing) by the end of the century are, in fact, propped up by hyper-inflated immigration from unrealistically large dummy landscape populations, while all other populations are projected to experience moderate to severe declines”; and concluding that “this results in the logical and physical impossibility that dummy landscape populations meant to represent the same cluster of populations can be both growing and declining at the same time”). Because the declaration goes well beyond merely explaining technical terms, I find it does not fit the technical-terms exception of Cobb’s History . See Minto v. U.S. Off. of Pers. Mgmt. , 765 F. App'x 779, 783-84 (3d Cir. 2019) (affirming the district court's conclusion that the administrative record was “exceedingly complete” without a declaration “set[ting] forth additional legal arguments” that were offered to explain complex terminology and references within the record).
“[T]he focal point for judicial review should be the administrative record already in existence, not some new record made initially in the reviewing court.” Camp v. Pitts , 411 U.S. 138, 142 (1973). Here, and as discussed above, plaintiffs’ motion must fail. First, plaintiffs fail to show that the technical-terms exception has been recognized by the Eleventh Circuit. Next, even if such an exception were established, plaintiffs fail to show how the declaration falls within the exception. Thus, plaintiffs lack a legal basis to admit the declaration as extra-record evidence and their motion is due to be denied.
It is ordered :
Plaintiffs Center for Biological Diversity of Nokuse Education, Inc.’s Motion [to] Supplement the Administrative Record and Memorandum in Support, doc. 20, is denied .
Ordered in Jacksonville, Florida, on November 22, 2024. c: Elise Pautler Bennet, Esquire Regan Whitlock, Esquire
Taylor Mayhall, Esquire
[1] While plaintiffs title their motion as a “motion to supplement the administrative record,” the Court construes it as a motion to admit extra-record evidence. To supplement the administrative record means to permit review of material that the agency considered but failed to include. Pacific Shores Subdivision, California Water Dist. v. U.S. Army Corps of Eng'rs , 448 F. Supp. 2d 1, 5 (D.D.C. 2006). To admit extra-record evidence means to permit review of material that the agency was not offered to consider. Pacific Shores , 448 F. Supp. 2d at 5. Although titled incorrectly, plaintiffs acknowledge that they are seeking to present “extra- record materials for the Court’s consideration.” Doc. 20 at 5, n.5. Dr. Shoemaker executed the declaration on June 26, 2024, after plaintiffs filed their complaint in this Court. Doc. 1; doc. 20-1 ¶ 33. Additionally, plaintiffs argue the declaration “explains
[2] Dr. Shoemaker worked with a colleague, Dr. Kevin J. Loope, in making his findings. Doc. 20-1 ¶ 12.