OPINION

DUNCAN, Circuit Judge:

This appeal arises out of a determination made by the Army Corps of Engineers (the "Corps") that it has jurisdiction, under the Clean Water Act ("CWA"), 33 U.S.C. § 1251 et seq., over 4.8 acres of wetlands located on Precon Development Corporation's ("Precon"'s) property, approximately seven miles from the nearest navigable water. The Corps subsequently denied Precon's application for a CWA permit to impact the wetlands through development. Precon appealed these determinations to the United States District Court for the Eastern District of Virginia under the Administrative Procedure Act ("APA"), 5 U.S.C. § 702, and the parties cross-moved for summary judgment. The district court granted summary judgment to the Corps on September 4, 2009, upholding both its jurisdictional determination and its permit denial. On appeal, Precon challenges only the Corps' jurisdictional determination. Because we find the Corps' administrative record inadequate to support its conclusion that it had jurisdiction over Precon's wetlands, we vacate the district court's grant of summary judgment and remand to the district court with instructions to remand to the Corps for reconsideration of its jurisdiction over the wetlands in question.

I.

A.

Precon is the developer of a 658-acre Planned Unit Development known as Edinburgh (the "Edinburgh PUD"), located in Chesapeake, Virginia. The city of Chesapeake is in southeastern Virginia, a region historically comprised of forested wetlands. Many of these wetlands ultimately drain into the Northwest River, which flows south through the region, passing within five to ten miles of the Edinburgh PUD.

The Edinburgh PUD is a mixed-use development that contains both residences and retail establishments. Its construction began in 2001. Precon acquired the Edinburgh PUD from RGM Corporation ("RGM"), the initial developer, in 2003. Since 2003, Precon has pursued the development of several residential areas within the Edinburgh PUD. Between 2004 and 2006, the Corps granted Precon permits to fill 77 acres of wetlands in order to proceed with these developments, based in part on an understanding that this was the totality of the development planned for the Edinburgh PUD.

In 2006, Precon announced a plan to develop ten additional residential lots in the Edinburgh PUD. The original plan for developing these lots required filling 10.7 acres of wetlands. However, after discussions with the Corps in which the Corps expressed its displeasure that Precon was separately pursuing additional residential development, Precon limited its proposed design so that it would only impact 4.8 acres of wetlands (the "Site Wetlands"). Precon further suggested that it did not believe the Corps had jurisdiction over these 4.8 acres. The Corps disagreed.

A detailed explanation of the geography of the Site Wetlands is critical to understanding the parties' dispute. The Site Wetlands are in the southwest quadrant of the Edinburgh PUD and sit adjacent to a man-made drainage ditch approximately 2,500 feet long (the "2,500-foot Ditch").1 The Site Wetlands do not, however, abut the 2,500-foot Ditch, because when the 2,500-foot Ditch was excavated through the surrounding wetlands in 1977, "[m]aterial excavated ... was side-cast on the east bank and therefore creates a berm between the [Site Wetlands] and the ditch."2 J.A. 264.

The 2,500-foot Ditch, which flows seasonally—i.e., from late winter to early spring—joins a larger, perennial drainage ditch, the Saint Brides Ditch, approximately 900 feet downstream of the Site Wetlands. The Saint Brides Ditch runs along the western boundary of the PUD for approximately 3,000 feet before continuing to meet a second perennial tributary about two and one-half to three miles south of the Edinburgh PUD. These merged tributaries flow into the Northwest River approximately three to four miles downstream.

The 4.8-acre Site Wetlands comprise only a small portion of the total wetland acreage within the Edinburgh PUD. There are, in total, 166 acres of wetlands in the PUD that are part of the Northwest River watershed.3 The remainder of the 166 acres are concentrated along the western edge of the PUD and surround the 2,500-foot Ditch and the Saint Brides Ditch.

B.

In 2007, Precon applied to the Corps for a jurisdictional determination as to whether the Site Wetlands were covered by the CWA, such that a permit would be needed before they could be impacted by development. Precon further requested a permit to fill the Site Wetlands if the Corps determined that a permit was required under the CWA.

On May 31, 2007, the Corps determined that it had jurisdiction over the Site Wetlands, on the ground that the wetlands sat adjacent to a ditch which qualified as "waters of the United States." J.A. 202. It subsequently denied Precon's request for a CWA permit. Precon administratively appealed both determinations. Around this same time, the Corps and the Environmental Protection Agency ("EPA") jointly issued new guidance (the "Rapanos Guidance," issued June 5, 2007) on CWA jurisdiction following the Supreme Court's jurisdiction-limiting decision in Rapanos v. United States, 547 U.S. 715, 126 S.Ct. 2208, 165 L.Ed.2d 159 (2006). In light of the Rapanos Guidance, a Corps appeals officer remanded the Corps' jurisdictional determination to the Corps' Norfolk District for reconsideration.

i.

The Rapanos Guidance instructs Corps and EPA personnel on how to make jurisdictional determinations that comply with the new rules for CWA jurisdiction announced by the Supreme Court in Rapanos.4 The Guidance explains that post Rapanos, wetlands, such as the Site Wetlands, which are "adjacent to, but not directly abutting, a relatively permanent tributary (e.g., separated from it by uplands, a berm, dike or similar feature)," are no longer automatically subject to the Corps' jurisdiction. J.A. 484. Pursuant to the Supreme Court's decision, the Rapanos Guidance instructs the Corps to evaluate such wetlands, along with "similarly situated" wetlands in the area, in order to determine whether they have a "significant nexus" with traditional navigable waters. Id.; see also Rapanos, 547 U.S. at 780, 126 S.Ct. 2208.

The first step in evaluating whether a significant nexus exists, according to the Rapanos Guidance, is to determine the region to be evaluated for significance. To do so, the relevant tributary must first be identified. A "tributary" for these purposes is defined as "the entire reach of the stream that is of the same order (i.e., from the point of confluence, where two lower order streams meet to form the tributary, downstream to the point such tributary enters a higher order stream)." J.A. 486. The pertinent section of the relevant tributary is known as the "relevant reach." Id. at 261. The Corps must next identify all wetlands adjacent to the relevant reach. Together, the relevant reach and its adjacent wetlands constitute the area to be evaluated for a significant nexus with a traditional navigable water.

The Rapanos Guidance explains that such evaluation should focus on the flow and functions of the relevant reach and adjacent wetlands. It instructs the Corps to specifically consider "volume, duration, and frequency" of flow in the relevant reach, as well as hydrologic information, physical characteristics, and functions performed by the relevant reach and its wetlands. Id. at 487. The Guidance then instructs the Corps, "after assessing the flow characteristics and functions," to evaluate whether these factors "are likely to have an effect that is more than speculative or insubstantial on the chemical, physical, and biological integrity of a traditional navigable water." J.A. 487. It emphasizes that "[a]s the distance from the tributary to the navigable water increases, it will become increasingly important to document whether the tributary and its adjacent wetlands have a significant nexus rather than a speculative or insubstantial nexus with a traditional navigable water." Id.

ii.

Upon remand and application of this new guidance, the Corps upheld its finding of jurisdiction over the Site Wetlands. Because the Site Wetlands do not abut—but only sit adjacent to—the 2,500-foot Ditch, the Corps did not treat them as automatically subject to jurisdiction, but instead, as instructed by the Rapanos Guidance, attempted to explain its rationale for upholding jurisdiction through a "Significant Nexus Determination."

First, the Corps' Significant Nexus Determination identified the relevant reach as the 2,500-foot Ditch and the Saint Brides Ditch, collectively. The Corps considered these ditches collectively because the Saint Brides Ditch and the 2,500-foot Ditch are, historically, part of the same naturally defined wetland drainage feature—a feature that was manipulated into discrete ditches in the late 1970s. Further, the Corps labeled both the Saint Brides Ditch, which undisputedly has perennial flow, and the 2,500-foot Ditch as "relatively permanent waters." The Corps defines "relatively permanent waters" as tributaries that "typically flow[] year-round or ha[ve] continuous flow at least `seasonally' (e.g., typically 3 months)." Id. at 242. The Corps found the 2,500-foot ditch to be a relatively permanent water because photographs supported the conclusion that the tributary flowed from February through April. Together, these two ditches were labeled as "a man-altered, first-order tributary to the Northwest River." Id. at 259.

The Corps then determined that the relevant reach of this tributary extended to the point, downstream 3.11 miles, where the Saint Brides Ditch joined the Pleasant Grove Swamp. In making this determination, the Corps ignored "[m]ultiple manmade or manipulated drainage ditches" that carried minor flow to the Saint Brides Ditch at various points downstream of the Edinburgh PUD, and instead selected the point where the Saint Brides Ditch converged with another historically natural drainage. Id.

The Corps next took up the task of identifying "similarly situated wetlands." It first identified all 166 acres of wetlands located within the Edinburgh PUD and the Northwest River watershed as similarly situated. Although the 4.8-acre Site Wetlands are separated from the remaining approximately 161 acres of wetlands by a road that is unfinished (filled but not paved), the Corps focused on this larger area because "the 4.8 acres of wetlands function as one with the remainder of the 166 acres of on site wetlands in the Saint Brides Ditch drainage area." Id. at 265. A berm separates these 166 acres of wetlands from the Saint Brides Ditch, but it has several breaks along its eastern edge.

The Corps then identified 282 more acres of "similarly situated wetlands" adjacent to the relevant reach but not on Precon's property. In determining that these 448 acres of wetlands should be evaluated in the aggregate, the Corps explained that the 4.8 acres of Site Wetlands and 166 acres of PUD wetlands are part of a "physical, chemical and biological connection of wetlands and streams" that exists, and "has always existed," in the area. Id. at 267.

Again in accordance with the Rapanos Guidance, the Corps' Significant Nexus Determination analyzed the functions and flow of the Saint Brides Ditch and the 2,500-foot Ditch. With respect to the Saint Brides Ditch, the Corps noted that the channel has a dynamic storage capacity of approximately 1.2 million cubic feet of water, a channel slope of 0.04 percent, and water velocities of approximately 1.3 feet per second—all of which means that it takes a volume of water approximately four hours to move through the relevant reach. Based on these observations, the Corps found that the ditch "greatly moderates the effect of flood flows" on the Northwest River due to its large storage capacity and slow release. Id. at 263. Additionally, it concluded that this low water velocity and extended residence time allowed suspended sediments to settle out of the water. It estimated that approximately 10,540 cubic yards of sediment is stored in the relevant portion of the Saint Brides Ditch rather than downstream in the Northwest River. It explained that this filtered sediment likely includes some quantity of dissolved pollutants that are thus removed from the Northwest River, improving the drinking water and fishing quality of the river.

With respect to the 2,500-foot Ditch, the Corps found that its 93,750 cubic feet of water storage capacity and substantial accumulation of woody debris allowed it to slow water velocities to 1.13 feet per second, providing "significant flood flow benefits to downstream traditionally navigable waters." J.A. 265. Its large woody debris also allowed it to trap at least 2,083 cubic yards of sediment and organic material that would otherwise flow downstream. Decomposed organic matter from these tributaries, the Corps explained, provides a "substantial food source" to fish species in the Northwest River. Id. at 266.

The Corps' Significant Nexus Determination further explained that numerous other ditches, similar to the 2,500-foot Ditch, drain into the Saint Brides Ditch along the relevant reach, each serving similar functions. The Corps concluded that these ditches cumulatively provide significant benefits to the river below, including "retaining a significant amount of flood water/flows, removing large volumes of sediments and pollutants from the system, as well as delivering important food resources to fish and other species living and spawning in the Northwest River." Id. at 267.

The Significant Nexus Determination then analyzed the 448 acres of similarly situated wetlands. Most of these wetlands are mineral flats, which contain unique hydric soils that have large amounts of organics at the surface, allowing them to retain more water than most wetlands. Although berms "have severed the direct surface water connection" among these wetlands in many places, the Corps found that "the berm has a negligible effect on the overall ecological functions that ... all of the adjacent wetlands in the [significant nexus] determination provide to downstream [traditional navigable waters]." Id. at 271. The Corps found the berms to be neither a barrier to wildlife functions nor an inhibitor of wetland functions, and in fact explained that they provide the benefit of "allow[ing] floodwaters to be retained longer within the wetlands prior to being discharged downstream thus moderating and mitigating flood flows." Id. The Corps also found that subsurface flows exist in the Edinburgh PUD wetlands, explaining that they slowly release groundwater into the Saint Brides Ditch.

The Corps then elaborated on the wetlands' role in flood mitigation. The 166 acres of on-site wetlands are capable of storing up to one and one-half feet of water per acre, and receive approximately 1,222,943 gallons of precipitation a year. Blackened leaves observed on-site evidence the wetlands' prolonged water storage capabilities. The Corps further explained that expert testimony from a trial between the Corps and the previous developer, RGM,5 supported the conclusion that the wetlands "slow release of water maintains base flows to the Northwest River and also moderates downstream flooding during extreme precipitation events." Id. at 272. Moreover, one expert testified that "loss of wetlands on the Edinburgh PUD would result in a major change to the timing and routing of water from the site," and that increased water velocities downstream "would cause erosion of sediments, increasing sedimentation and pollution of downstream waters including the Northwest River." Id.

The Corps further observed that such mineral flat wetlands rapidly cycle nutrients, sequester carbon, and help denitrify water, reducing eutrophication.6 The 448 acres of similarly situated wetlands were found to remove an estimated 448 to 9,403 pounds of nitrogen per acre from the water each year, and also remove an unquantified amount of pollutants and particles. Moreover, the 4.8-acre Site Wetlands and similarly situated wetlands serve as habitat for numerous species, including State endangered species, at least some of which can cross the unfinished road separating the Site Wetlands from remaining wetlands. These species use the area as "a corridor for movement between the Northwest River and points to the north and west." Id. at 276.

On the basis of this evidence, the Corps' Significant Nexus Determination concluded that the tributaries and their adjacent wetlands have "a significant nexus that has more than a speculative or insubstantial effect on the Northwest River," and that loss of these wetlands "would have a substantial negative impact on water quality and biological communities of the river's ecosystem." Id. at 277-78. Accordingly, the Corps reaffirmed its previous conclusion that it had jurisdiction over the Site Wetlands, such that Precon would be required to obtain a CWA permit before filling them.

C.

Precon sought judicial review of this determination, along with the Corps' denial of its permit application, in the United States District Court for the Eastern District of Virginia. The parties filed cross-motions for summary judgment, which were referred to a magistrate judge for a Report and Recommendation ("R & R"). The district court adopted this R & R in full on September 4, 2009, granting the Corps' motion for summary judgment and denying Precon's motion for summary judgment.

The district court found that the Corps had permissibly defined the scope of its review area as including 448 acres of similarly situated wetlands, and that the Corps' determination that these wetlands had a significant nexus to the Northwest River was supported by substantial factual findings. The district court also upheld the Corps' denial of a CWA permit. This appeal followed.

II.

On appeal, Precon challenges the district court's finding that the Corps properly asserted jurisdiction over the Site Wetlands under the CWA.7 Precon argues that there are two major flaws in the Corps' jurisdictional determination. First, it contends that the Corps' decision to aggregate 448 acres of surrounding wetlands in determining jurisdiction was impermissible. Second, it argues that even if all 448 acres were appropriately included in the Corps' jurisdictional determination, the Corps did not provide sufficient evidence that the connection between these wetlands and the Northwest River amounted to a significant nexus.

Before reaching Precon's two substantive arguments, we provide an overview of the Supreme Court's jurisprudence addressing the parameters of CWA jurisdiction. We then turn to each of Precon's contentions.

A.

Congress passed the CWA in 1972 "to restore and maintain the chemical, physical, and biological integrity of the Nation's waters." 33 U.S.C. § 1251. To that end, the CWA prohibits the discharge of pollutants into navigable waters. See id. §§ 1311(a), 1362(12)(A). The CWA defines navigable waters as "the waters of the United States, including the territorial seas." 33 U.S.C. § 1362(7). Although the Corps initially construed this definition to cover only waters navigable in fact, "in 1975 the Corps issued interim final regulations redefining `the waters of the United States' to include not only actually navigable waters but also tributaries of such waters" and "`freshwater wetlands' that were adjacent to other covered waters." United States v. Riverside Bayview Homes, Inc., 474 U.S. 121, 123-24, 106 S.Ct. 455, 88 L.Ed.2d 419 (1985).

In Riverside Bayview Homes, the Supreme Court upheld the Corps' determination that it had jurisdiction over wetlands adjacent to navigable waters. Id. at 139, 106 S.Ct. 455. Even though the plain language of the statute did not compel this conclusion, the Court explained that by including a broad definition of "navigable waters" in the CWA, Congress "evidently intended to ... exercise its powers under the Commerce Clause to regulate at least some waters that would not be deemed `navigable' under the classical understanding of that term." Id. at 133, 106 S.Ct. 455. It further reasoned that the Corps' decision to include wetlands within its jurisdiction was a reasonable one, given wetlands' critical importance to the health of adjacent waters. Id. at 133-34, 106 S.Ct. 455.

The Supreme Court again interpreted the CWA term "navigable waters" in Solid Waste Agency of Northern Cook County v. United States Army Corps of Engineers, 531 U.S. 159, 121 S.Ct. 675, 148 L.Ed.2d 576 (2001) ("SWANCC"). In SWANCC, it considered whether "isolated ponds, some only seasonal, wholly located within two Illinois counties, f[e]ll under [the CWA's] definition of `navigable waters' because they serve[d] as habitat for migratory birds." Id. at 171-72, 121 S.Ct. 675. The Court held that these waters were simply too far removed from any navigable waters to be included within that term. Id. To distinguish these isolated ponds from the wetlands it considered in Riverside Bayview Homes, the Court explained: "It was the significant nexus between the wetlands and `navigable waters' that informed our reading of the CWA in Riverside Bayview Homes." Id. at 167, 121 S.Ct. 675.

Five years later, in Rapanos, the Supreme Court revisited the issue of the Corps' jurisdiction over adjacent wetlands. 547 U.S. at 715, 126 S.Ct. 2208. Although recognizing the continuing validity of Riverside Bayview Homes, a majority of the Court found troubling the Corps' assertion of jurisdiction over wetlands adjacent to tributaries far away from, and unimportant to, any traditional navigable water.8 See id. at 726, 730-33, 106 S.Ct. 455 (plurality opinion); id. at 776, 779-80, 106 S.Ct. 455 (Kennedy, J., concurring in the judgment). Accordingly, a fractured Court proposed two different ways to limit the reach of its earlier ruling so as not to allow jurisdiction over wetlands lying alongside "remote and insubstantial" ditches and drains. Id. at 778, 106 S.Ct. 455 (Kennedy, J., concurring in the judgment). The Rapanos plurality suggested that wetlands should only fall within CWA jurisdiction when they: (1) are adjacent to a "relatively permanent body of water connected to traditional interstate navigable waters"; and (2) have "a continuous surface connection with that water." Id. at 742, 126 S.Ct. 2208 (plurality opinion). Justice Kennedy, concurring, found this test too limiting. Instead, he borrowed language from SWANCC to establish an alternative new test for jurisdiction over adjacent wetlands. Id. at 779-80, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment). Under his formulation, when the Corps "seeks to regulate wetlands based on adjacency to nonnavigable tributaries," it must establish that a "significant nexus" exists "between the wetlands in question and navigable waters in the traditional sense." Id. at 779, 782, 126 S.Ct. 2208. The dissent, which drew four votes, found both of these tests too stringent. It thus suggested that in the future, jurisdiction should be established if either the plurality's or Justice Kennedy's test is met. Id. at 810, 126 S.Ct. 2208 (Stevens, J., dissenting).

The parties here agree that Justice Kennedy's "significant nexus" test governs and provides the formula for determining whether the Corps has jurisdiction over the Site Wetlands. We therefore do not address the issue of whether the plurality's "continuous surface connection" test provides an alternate ground upon which CWA jurisdiction can be established.9

Given that the significant nexus test undisputedly controls, it bears further elaboration. Justice Kennedy derived this test from a recognition that while Congress clearly intended to allow CWA jurisdiction over "at least some waters that are not navigable in the traditional sense," id. at 767, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment), some meaning had to be given to the term "navigable" as used in the statute, id. at 778-79, 126 S.Ct. 2208. To discern this meaning, he returned to Riverside Bayview Homes, where the Court upheld jurisdiction over some wetlands by explaining that "wetlands adjacent to lakes, rivers, streams, and other bodies of water may function as integral parts of the aquatic environment...." Id. at 779, 126 S.Ct. 2208 (quoting Riverside Bayview Homes, 474 U.S. at 135, 106 S.Ct. 455).

Drawing upon this purposive rationale for including certain wetlands within the term "navigable waters," Justice Kennedy explained that "the Corps' jurisdiction over wetlands depends upon the existence of a significant nexus between the wetlands in question and navigable waters in the traditional sense." Id. Wetlands possessing this significant nexus are those that "perform critical functions related to the integrity of other waters—functions such as pollutant trapping, flood control, and runoff storage." Id. Accordingly, Justice Kennedy set forth the following standards for evaluating the existence of a significant nexus:

Id. at 780, 126 S.Ct. 2208.

Justice Kennedy further explained that, in accordance with Riverside Bayview Homes, wetlands adjacent to navigable-in-fact waters necessarily satisfy this significant nexus test. Id. However, for wetlands adjacent to non-navigable tributaries, such as the Site Wetlands we consider here, the Corps must now "establish a significant nexus on a case-by-case basis." Id. at 782, 126 S.Ct. 2208. As possible indicia of the significance of such wetlands, Justice Kennedy noted that the Corps might consider documenting "the significance of the tributaries to which the wetlands are connected," a "measure of the significance of [the hydrological connection] for downstream water quality," and/or "the quantity and regularity of flow in the adjacent tributaries." Id. at 784, 786, 126 S.Ct. 2208.

B.

With this framework established, we turn to Precon's substantive challenges. We review the district court's decision to grant summary judgment de novo, S.C. Green Party v. S.C. State Election Comm'n, 612 F.3d 752, 755 (4th Cir.2010), including its findings on an administrative record, Ohio Valley Envtl. Coal. v. Aracoma Coal Co., 556 F.3d 177, 189 (4th Cir. 2009).

Precon's arguments require a careful examination of the Corps' application of the language Justice Kennedy has engrafted into the statutory requirements of the CWA. See Rapanos, 547 U.S. at 779, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment). We therefore treat compliance with Justice Kennedy's "significant nexus" test as a question of law, as we do any question of statutory interpretation, and review for compliance de novo. See, e.g., Stone v. Instrumentation Lab. Co., 591 F.3d 239, 242-43 (4th Cir.2009); cf. Kentuckians for Commonwealth Inc. v. Rivenburgh, 317 F.3d 425, 439 (4th Cir. 2003) ("When reviewing a particular agency action ... [t]he court is first required to decide whether the [agency] acted within the scope of [its] authority." (internal quotations omitted and alterations in original)). However, recognizing the Corps' expertise in administering the CWA, we give deference to its interpretation and application of Justice Kennedy's test where appropriate.10 See United States v. Mead Corp., 533 U.S. 218, 234, 121 S.Ct. 2164, 150 L.Ed.2d 292 (2001) ("[A]n agency's interpretation may merit some deference whatever its form, given the `specialized experience and broader investigations and information' available to the agency....") (quoting Skidmore v. Swift, 323 U.S. 134, 140, 65 S.Ct. 161, 89 L.Ed. 124 (1944)).

i.

Precon first challenges the Corps' decision to label 448 acres of surrounding wetlands as "similarly situated" wetlands for purposes of its significant nexus determination. On this point, recognizing the deference due the Corps' factual findings and interpretation of the phrase "similarly situated," we uphold the Corps' finding that 448 acres of wetlands were "similarly situated" to the Site Wetlands.

Justice Kennedy's significant nexus test clearly allows some aggregation of wetlands in determining whether a significant nexus exists. He explained that the significant nexus inquiry should focus on whether "wetlands, either alone or in combination with similarly situated lands in the region, significantly affect the chemical, physical, and biological integrity of other covered waters more readily understood as `navigable.'" Rapanos, 547 U.S. at 780, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment) (emphasis added). However, his concurrence provided no further explanation of what "similarly situated," or, for that matter, "region," should be taken to mean in this context.

To flesh out this concept, the Corps' Rapanos Guidance interprets "similarly situated" to mean "all wetlands adjacent to the same tributary." J.A. 486. A tributary, in turn, is defined as "the entire reach of the stream that is of the same order (i.e. from the point of confluence, where two lower order streams meet to form the tributary, downstream to the point such tributary enters a higher order stream)." Id.

Applying these definitions to the instant case, the Corps identified the relevant tributary as the 2,500-foot Ditch and the Saint Brides Ditch, collectively, down to the point where the Saint Brides Ditch converged with the Pleasant Grove Swamp. The Corps explained that it considered these ditches together because they were historically part of the same naturally defined wetland drainage feature before human-made ditches altered the area. The Corps then identified the 166 acres of wetlands located on the Edinburgh PUD and an additional 282 acres of wetlands outside of Precon's property as wetlands sitting "adjacent" to this "relevant reach." It noted that all 448 acres were part of a "physical, chemical and biological connection of wetlands and streams" that existed, "and had always existed," in the area. Id. at 267.

Precon acknowledges that the Rapanos Guidance's interpretation of "similarly situated lands in the region" is entitled to Skidmore deference. Under Skidmore, an agency's interpretation merits deference "to the extent that the interpretation has the power to persuade." U.S. Dep't of Labor v. N.C. Growers Ass'n, 377 F.3d 345, 353-54 (4th Cir.2004) (citing Skidmore, 323 U.S. at 140, 65 S.Ct. 161). Precon argues, however, that the Corps' determination that the 448 acres were "similarly situated" is unpersuasive because (1) adjacent and abutting wetlands cannot reasonably be considered to be "similarly situated," and (2) the Corps failed to follow its own guidance here.11

According to Precon, the primary flaw in the Corps' interpretation of "similarly situated" is its equal treatment of abutting and other adjacent wetlands. Specifically, Precon argues that the plurality and Justice Kennedy in Rapanos "expressly recognize[d] that there is a significant difference in the relationship between abutting and non-abutting wetlands and their nearest ditches." Appellant's Br. at 46.

Although the Rapanos plurality clearly found the abutting/adjacent distinction meaningful, see 547 U.S. at 740-42, 126 S.Ct. 2208, we find no evidence that Justice Kennedy, in permitting "similarly situated lands" to be included within the significant nexus analysis, intended to differentiate between abutting and other adjacent wetlands. To the contrary, his concurrence explicitly approved of the Corps' regulatory definition of "adjacent," which includes both those wetlands that directly abut waters of the United States and those separated from other waters "by man-made dikes or barriers, natural river berms, beach dunes and the like." 33 C.F.R. § 328.3(c). As Justice Kennedy explained, abutting wetlands are not necessarily any more important than other adjacent wetlands because "filling in wetlands separated from another water by a berm can mean that floodwater, impurities, or runoff that would have been stored or contained in the wetlands will instead flow out to major waterways." Rapanos, 547 U.S. at 775, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment). He thus concluded that "it may be the absence of an interchange of waters... that makes protection of the wetlands critical to the statutory scheme." Id.

Here, the Corps adopted this precise rationale in aggregating abutting and other adjacent wetlands. It explained that the berm separating the 4.8-acre Site Wetlands from the 2,500-foot Ditch did not disconnect these wetlands from surrounding ones, because it neither inhibited wildlife movement nor wetland functions. It also explained that the berm in fact provided the additional benefit of "allow[ing] floodwaters to be retained longer within the wetlands prior to being discharged downstream thus moderating and mitigating flood flows." J.A. 271. Given this reasonable explanation for its actions, we see no error in the Corps' decision to aggregate both abutting and adjacent wetlands in its significant nexus determination.

As for Precon's second argument, we acknowledge that it is difficult to determine whether the Corps precisely adhered to the Rapanos Guidance in identifying "similarly situated" wetlands. Specifically, it is not clear that the Guidance contemplates that multiple tributaries might appropriately be included within the "relevant reach." Although at oral argument the Corps took the position that its determination to aggregate these two ditches was merely an application of its Rapanos Guidance to the unique geography of the area, we are not convinced that the Guidance is so flexible on this point.

However, we conclude that we do not need to determine whether or not the Corps methodically adhered to its nonbinding guidance document in identifying the "similarly situated" wetlands here. Cf. J.A. 481 n.16 (footnote in the Rapanos Guidance explaining that the guidance "does not impose legally binding requirements... and may not apply to a particular situation depending on the circumstances"). Even if the Corps deviated from its guidance, it provided reasoned grounds for doing so.

The Corps explained that it decided to aggregate the wetlands surrounding both the 2,500-foot Ditch and the Saint Brides Ditch because the two ditches were, historically, part of the same naturally defined wetland drainage feature—a feature that was manipulated into discrete ditches in the late 1970s. We accept this finding as true, having no reason to believe it was "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. § 706(2)(A); see also Aracoma, 556 F.3d at 192 (explaining that we should be at our "most deferential" when reviewing findings of fact based on special expertise). And based on this finding, we are persuaded that the Corps acted reasonably in aggregating these two man-made ditches into a single "tributary." There is both logical and practical appeal to treating man-made ditches that would naturally be part of the same drainage feature together. Otherwise, a property owner could avoid CWA jurisdiction simply by digging a few well-placed drainage ditches on either side of the wetlands he wished to fill.

We find more questionable the Corps' decision, after determining that it would treat these two ditches together, to include adjacent wetlands stretching over three miles downstream as "similarly situated." However, we recognize that Justice Kennedy's instruction—that "similarly situated lands in the region" can be evaluated together—is a broad one, open for considerable interpretation and requiring some ecological expertise to administer. On the basis of this recognition, and with no appropriate limiting principle suggested by Precon as to which wetlands could properly have been considered "similarly situated" here, we uphold the Corps' finding that all 448 acres of non-contiguous wetlands adjacent to the 2,500-foot Ditch and the Saint Brides Ditch down to Pleasant Grove Swamp were "similarly situated." However, the Corps' record on this point gives us a bare minimum of persuasive reasoning to which we might defer. It only notes, somewhat conclusorily, that the Site Wetlands "continue to function as part of the entire" 448 acres. J.A. 271. We urge the Corps to consider ways to assemble more concrete evidence of similarity before again aggregating such a broad swath of wetlands.

For the foregoing reasons, we reject Precon's argument that the Corps impermissibly identified 448 acres of wetlands as "similarly situated lands in the region," and uphold the Corps' findings on this point.

ii.

We now turn to Precon's argument that the Corps did not adequately establish the existence of a significant nexus between the Site Wetlands—along with similarly situated wetlands—and the Northwest River. The Corps' factual findings on this point are not in dispute;12 rather, Precon challenges whether the Corps' administrative record, if accepted as accurate, suffices to meet Justice Kennedy's significant nexus test. Upon close examination of the record, we find that it contains insufficient information to allow us to assess the Corps' conclusion that these wetlands have a significant nexus with the Northwest River, a body of water situated miles away.

Precon's primary argument is that the Corps' record lacks any "measures" of the effects that these wetlands have on the Northwest River. And, it reasons, without such "measures," the wetlands' significance for the river's health cannot be established. The Corps responds that Justice Kennedy's significant nexus test does not require empirical or quantitative evidence of "significance," and that the evidence it provided more than sufficed to establish a significant nexus.

We have not yet had occasion to consider the evidentiary requirements of Justice Kennedy's significant nexus test. The Sixth Circuit has most directly addressed the issue. In United States v. Cundiff, 555 F.3d 200 (6th Cir.2009), it held that Justice Kennedy's significant nexus test does not require "`laboratory analysis' of soil samples, water samples, or ... other tests." Id. at 211. Instead, the Sixth Circuit found that the district court had not clearly erred by finding that a significant nexus was demonstrated through qualitative, rather than quantitative, physical evidence. This evidence included expert testimony that dredging and filling of the wetlands at issue "undermined the wetlands' ability to store water which, in turn, ... affected the frequency and extent of flooding, and increased the flood peaks in the Green River," and caused visible acid mine runoff previously stored in the wetlands to flow more directly to the Green River. Id. at 210-11.

We agree that the significant nexus test does not require laboratory tests or any particular quantitative measurements in order to establish significance. As Justice Kennedy explained, the significant nexus test is a flexible ecological inquiry into the relationship between the wetlands at issue and traditional navigable waters. See Rapanos, 547 U.S. at 779-80, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment). However, in announcing this test, he clearly intended for some evidence of both a nexus and its significance to be presented. Otherwise, it would be impossible to engage meaningfully in an examination of whether a wetland had "significant" effects or merely "speculative or insubstantial" effects on navigable waters. Id. at 780, 126 S.Ct. 2208. Justice Kennedy's opinion further provides specific examples of the types of evidence that might support a determination of significance. For instance, an adequate record might include documentation of "the significance of the tributaries to which the wetlands are connected," a "measure of the significance of [the hydrological connection] for downstream water quality," and/or "indication of the quantity and regularity of flow in the adjacent tributaries." Id. at 784, 786, 126 S.Ct. 2208.

The question is thus whether the Corps' record contained enough physical evidence—quantitative or qualitative—to allow us to uphold its determination that a significant nexus existed here. Relying on the final example described above, the Corps asserted at oral argument that its documentation of the flow of the adjacent tributaries sufficed, even standing alone, to establish that a significant nexus existed here. According to this theory, a measurement of these tributaries' flow adequately demonstrated that this area "help[ed] to slow flows/retain floodwaters, releasing them slowly so that downstream waters do not receive as much flow volume and velocity, all working to diminish downstream flooding and erosion," which led to the conclusion that a significant nexus existed. J.A. 265.

We cannot accept this conclusion for two reasons. First, as Precon points out, the Corps' administrative record does not appear to contain any measurements of actual flow. Nor was counsel able to point to such measurements at oral argument. Instead, the record reflects measures of the water storage capacity and the resultant potential flow rates of the Saint Brides Ditch and the 2,500-foot Ditch, without any indication of how often this capacity is reached or how much flow is typically in the ditches.13

Second, even if the record had sufficiently documented flow, we do not believe that recitation of the flow of an adjacent tributary alone, absent any additional information regarding its significance, would necessarily suffice to establish a significant nexus. The significant nexus inquiry emphasizes the comparative relationship between the wetlands at issue, their adjacent tributary, and traditional navigable waters. Cf. Rapanos, 547 U.S. at 780, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment) (drawing a critical distinction between wetlands with "significant" effects versus only "insubstantial" effects on navigable waters). We can therefore imagine, for example, that wetlands next to a tributary with minimal flow might be significant to a river one quarter mile away, whereas wetlands next to a tributary with much greater flow might have only insubstantial effects on a river located twenty miles away. Accordingly, in this case of wetlands approximately seven miles from any navigable water, we cannot say that recitation of the adjacent tributary's flow, standing alone, would necessarily have sufficed.

We acknowledge that the Corps' Significant Nexus Determination did contain other physical observations about the wetlands and adjacent tributaries.14 However, there is no documentation in the record that would allow us to review its assertion that the functions that these wetlands perform are "significant" for the Northwest River. In particular, although we know that the wetlands and their adjacent tributaries trap sediment and nitrogen and perform flood control functions, we do not even know if the Northwest River suffers from high levels of nitrogen or sedimentation, or if it is ever prone to flooding. This lack of evidence places the facts here in stark contrast to those in Cundiff, upon which the Corps relies. There, the Sixth Circuit noted that the district court credited expert testimony about the wetlands "in relation to" the navigable river. 555 F.3d at 210-11 (emphasis added). According to that testimony, the challenged actions had undermined the wetlands' ability to store water, which, in turn, had increased the flood peaks in the Green River. Id. Additional testimony established that acid mine runoff that had previously been stored in the wetlands flowed more directly into the river, causing "direct and significant impacts to navigation ... and to aquatic food webs" in the river. Id. at 211. There is no such testimony here.

Accordingly, we must conclude that this record does not support the Corps' determination that the nexus that exists between the 448 acres of similarly situated wetlands and the Northwest River is "significant." Particularly given the facts of this case, involving wetlands adjacent to two man-made ditches, flowing at varying and largely unknown rates toward a river five to ten miles away, we cannot accept, without any information on the river's condition, the Corps' conclusion that the nexus here is significant. Justice Kennedy created the significant nexus test specifically because he was disturbed by the assertion of jurisdiction over wetlands situated along a ditch "many miles from any navigable-in-fact water," carrying "only insubstantial flow toward it." Rapanos, 547 U.S. at 786, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment).

In support of our request for further information on these wetlands' significance, we observe that the geography of the wetlands at issue places them squarely in that category of wetlands over which jurisdiction is no longer assured. Carabell, one of the consolidated cases in Rapanos, involved wetlands similar to, but less remote than, the Site Wetlands. In Carabell, the Corps had asserted jurisdiction over 15.9 acres of forested wetlands, lying along a ditch—but separated from the ditch by a man-made berm—which eventually drained into Lake St. Claire approximately one mile downstream. Id. at 764, 126 S.Ct. 2208. Both the plurality and Justice Kennedy agreed that more evidence was needed about these wetlands' characteristics before jurisdiction could be established. Id. at 757, 126 S.Ct. 2208 (plurality opinion); id. at 786-87, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment). Given that the Site Wetlands are considerably more removed from traditional navigable waters than the wetlands at issue in Carabell, it follows that it would be even more important for the Corps to fully document the significance of their effects on navigable water. Indeed, even the Corps' own Rapanos Guidance cautions that "[a]s the distance from the tributary to the navigable water increases, it will become increasingly important to document whether the tributary and its adjacent wetlands have a significant nexus rather than a speculative or insubstantial nexus with a traditional navigable water." J.A. 487.

Recent Ninth and Sixth Circuit cases provide good examples of the types of evidence—either quantitative or qualitative—that could suffice to establish "significance." In Northern California River Watch v. City of Healdsburg, 496 F.3d 993 (9th Cir.2007), the Ninth Circuit held the significant nexus test satisfied in part because the district court found increased chloride levels in the relevant navigable water, from 5.9 parts per million to 18 parts per million, due to chlorine seepage from the wetlands in question into the navigable river. Id. at 1001. Alternatively, Cundiff provides an example of the type of qualitative evidence that can establish a significant nexus. As noted earlier, the Sixth Circuit's opinion in Cundiff rested on evidence that the wetlands' acid mine drainage storage capabilities and flood storage capabilities had "direct and significant" impacts on navigation in the Green River, via sediment accumulation, and that the diversion of water from the wetlands had "increased the flood peaks" in the Green River. 555 F.3d at 210-11. Thus, in contrast to the present case, both River Watch and Cundiff included some evidence not only of the functions of the relevant wetlands and their adjacent tributaries, but of the condition of the relevant navigable waters.

The Corps argues that we must afford deference to its significant nexus finding. We agree that its factual findings are entitled to deference under the APA, and should be reversed only if "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. § 706(2)(A); see also Marsh v. Or. Natural Res. Council, 490 U.S. 360, 376-77, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989). The Corps' factual findings, however, are not in dispute. The question is instead whether the Corps' findings were adequate to support the ultimate conclusion that a significant nexus exists. This legal determination is essentially now a matter of statutory construction, as Justice Kennedy established that a "significant nexus" is a statutory requirement for bringing wetlands adjacent to non-navigable tributaries within the CWA's definition of "navigable waters." See Rapanos, 547 U.S. at 779-80, 126 S.Ct. 2208 (Kennedy, J., concurring in the judgment). As we mentioned at the outset, on this question of statutory interpretation, absent the promulgation of new regulations, the Corps' conclusions are entitled at most to Skidmore deference. See Mead, 533 U.S. at 234-35, 121 S.Ct. 2164; cf. Rivenburgh, 317 F.3d at 439. Because the Corps' current administrative record contains no evidence of significance for us to review, we cannot find its conclusion that significance existed here persuasive. Cf. Shipbuilders Council of Am. v. U.S. Coast Guard, 578 F.3d 234, 245 (4th Cir. 2009) ("Some indicia of reliability and reasonableness must exist in order for us to defer to the agency's interpretation.").

For these reasons, we reverse the district court's holding that the Corps' administrative record adequately demonstrated that a significant nexus existed here, and remand to the Corps for reconsideration of its significant nexus determination. Cf. Motor Vehicle Mfrs. Ass'n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 52, 57, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (remanding for further agency consideration when the agency's view of the facts was accepted but the Court "appreciate[d] the limitations of th[e] record in supporting the agency's decision"); Cook v. Heckler, 783 F.2d 1168, 1174 (4th Cir.1986) (remanding for further consideration where it was "impossible to conclude that there [was] substantial evidence to support the Secretary's determination"). In doing so, we do not intend to place an unreasonable burden on the Corps. We ask only that in cases like this one, involving wetlands running alongside a ditch miles from any navigable water, the Corps pay particular attention to documenting why such wetlands significantly, rather than insubstantially, affect the integrity of navigable waters. Such documentation need not take the form of any particular measurements, but should include some comparative information that allows us to meaningfully review the significance of the wetlands' impacts on downstream water quality.

III.

For the foregoing reasons, the district court's grant of summary judgment is reversed and we remand to the district court with instructions to remand to the Corps for further consideration in light of this opinion.

REVERSED AND REMANDED