SKOPIL, Circuit Judge:
We are asked in these consolidated appeals to determine whether the Port Madison Indian Reservation in Washington includes adjacent tidelands. The Suquamish Indian tribe, joined by the United States, argues that the tidelands were reserved by treaty for the tribe’s exclusive use. The state of Washington and private landowners contend that the tidelands were never part of the reservation and therefore title to the tidelands passed to the state on admission to the union in 1889 pursuant to the “equal footing doctrine.” Non-Indians now hold record title to a substantial portion of the disputed tidelands by virtue of conveyances from the state.
The district court concluded that the treaty that established the reservation did not expressly include the disputed tidelands and that the evidence presented did not show that the United States intended that the reservation include the adjacent tidelands. The court therefore concluded that the United States retained ownership of the tidelands as public lands which thereafter passed to the state upon admission to the union. We agree with that conclusion and we affirm.
[192]*192FACTS AND PRIOR PROCEEDINGS
The Treaty of Point Elliott, signed January 22, 1855, created the Port Madison Indian Reservation on behalf of the Su-quamish, Duwamish and allied Indians. As enlarged by Order of the Secretary of the Interior in 1864, the reservation contained over 7000 acres of land bordering Puget Sound and abutting nearly eleven miles of tidelands. Although the location and size of the reservation have not changed since 1864, much of the land in the reservation was allotted to individual Indians during the 1880’s and eventually transferred to-non-Indians. Non-Indians own waterfront land within the reservation and possess corresponding deeds to the abutting tidelands conveyed by the state.
The Treaty of Point Elliott was intended and interpreted by the parties to extinguish all aboriginal title of the Suquamish Indian Tribe in the United States. In return for this cession of title, Article 2 of the treaty established the Port Madison Indian Reservation on behalf of the tribe:
There is, however, reserved for the present use and occupation of said tribes and bands the following tracts of land, viz: the amount of two sections, or twelve hundred and eighty acres, surrounding the small bight at the head of Port Madison, called by the Indians Noo-sohk-um; ... All which tracts shall be set apart, and so far as necessary surveyed and marked out for their exclusive use; nor shall any white man be permitted to reside upon the same without permission of said tribes or bands....
Surveys of the area were made in 1858 and 1859. Those surveys followed the regular practice of that time of showing land adjacent to a navigable body of water as a meandering line representing the high water mark. In the Secretarial Order of 1864, the proposed enlargement of the Port Madison Indian Reservation was diagrammed on a plat traced from the 1858 and 1859 surveys. Thus, the enlarged reservation was shown as bordering Puget Sound at the high water mark.
At the time of the treaty, the Suquamish Indians harvested shellfish from tidelands over a wide area of Puget Sound. The district court found that such shellfish formed a staple in their diet, although salmon provided most of the tribe’s food. The tribe’s right to gather food resources was provided by Article 5 of the treaty:
The right of taking fish at usual and accustomed grounds and stations is further secured to said Indians in common with all citizens of the territory, and of erecting temporary houses for the purpose of curing, together with the privilege of hunting and gathering roots and berries on open and unclaimed lands. Provided, however, that they shall not take shell-fish from any beds staked or cultivated by citizens.
The United States and the tribe brought these actions to resolve the adverse claims of ownership of the disputed tidelands. They sought a declaration of quiet title in favor of the United States as trustee for the tribe. After a bench trial, the district court held that “the plaintiffs have failed to overcome the presumption against inferring a conveyance of the case area tidelands by the United States.” The court concluded that “during the relevant period the United States retained ownership of the case area tidelands as public lands which then passed to the State of Washington upon its admission to the Union on November 11, 1889.”
DISCUSSION
I.
The Supreme Court has oft resolved disputed Indian claims to land under navigable waterways. In Choctaw Nation v. Oklahoma, 397 U.S. 620, 90 S.Ct. 1328, 25 L.Ed.2d 615 (1970), the Cherokees sought title to land underlying certain navigable portions of the Arkansas River within the outer boundaries of their reservation. The Court acknowledged that resolution of the issue required it to “pass upon the effect of treaties that were entered into nearly a century and half ago.” Id. at 622, 90 S.Ct. at 1330. The treaties were found to be ambiguous, but interpreting the treaties in favor of the Indians and “as they would [193]*193have understood them,” the Court held that the United States intended to and did convey title of the riverbed to the Indians. Id. at 631, 635, 90 S.Ct. at 1334, 1336. The Court distinquished a prior decision, United States v. Holt State Bank, 270 U.S. 49, 46 S.Ct. 197, 70 L.Ed. 465 (1926), where it had concluded that the United States did not intend to reserve beneficial title to the bed of a lake that was fully enclosed by the Red Lake Indian Reservation. Choctaw Nation, 397 U.S. at 634, 90 S.Ct. at 1336. In Holt, the Court noted that nothing in the treaties indicated a departure “from the established policy ... of treating such lands as held for the benefit of the future state.” Holt, 270 U.S. at 58-59, 46 S.Ct. at 200.
Prior to Holt, however, the Court had easily inferred the requisite intent in Alaska Pacific Fisheries v. United States, 248 U.S. 78, 39 S.Ct. 40, 63 L.Ed. 138 (1918). There, the Metlakahtla Indians, whose reservation consisted of a group of small islands, sought to prevent non-Indian fishing within the island’s tidewaters. Id. at 86, 39 S.Ct. at 41. The Court treated the issue as one of language interpretation — “what Congress intended by the words — the body of lands known as Annette islands.’ ” Id. at 87, 39 S.Ct. at 41. To answer that question, the Court reviewed, inter alia, the Congressional purpose in creating the reservation. It found that Congress sought to establish an Indian colony which would be self-sustaining. Id. at 89, 39 S.Ct. at 42. To achieve that purpose, the Court reasoned, Congress must have intended the tribe to have the exclusive use of adjacent fishing grounds because it was clear after reviewing the resources of the islands that the Metlakahtlans “could not sustain themselves from the use of the upland alone.” Id. Thus the Court, again construing doubtful expression in favor of the Indians, concluded that Congress intended that the reservation embrace the surrounding waters of the islands. Id.
Choctaw, Holt State Bank, and Alaska Pacific Fisheries were revisited by the Supreme Court in Montana v. United States, 450 U.S. 544, 101 S.Ct. 1245, 67 L.Ed.2d 493 (1981). There, the Crow Tribe of Montana, relying on its purported beneficial ownership of the Big Horn River, sought to exclude non-Indians from fishing within the reservation. Id. at 547, 101 S.Ct. at 1249. The Court held that the tribe failed to overcome the strong presumption against conveyance by the United States of land under navigable water. Id. at 553, 101 S.Ct. at 1252. While acknowledging that the creation of an Indian reservation can be an “ ‘appropriate public purpose’ ... justifying a congressional conveyance of a riverbed,” the Court reasoned that nothing at the time of the Crow treaties presented a “public exigency” which would indicate that Congress departed “from its policy of reserving ownership of beds under navigable waters for the future States.” Id. at 556, 101 S.Ct. at 1253-54 (citation omitted). The only factor cited in support of that reasoning was that the “Crows were a nomadic tribe dependent chiefly on buffalo, and fishing was not important to their diet or way of life.” Id.
After Montana was decided, we also examined the issue of beneficial tribal ownership of riverbeds within a reservation. In two cases decided the same day, we formulated a test to determine whether a grant of property to an Indian tribe includes title to land under navigable water. See Puyallup Indian Tribe v. Port of Tacoma, 717 F.2d 1251, 1258 (9th Cir.1983), cert. denied, 465 U.S. 1049, 104 S.Ct. 1324, 79 L.Ed.2d 720 (1984); Muckleshoot Indian Tribe v. Trans-Canada Enters, Ltd., 713 F.2d 455, 457 (9th Cir.1983) (per curiam), cert. denied, 465 U.S. 1049, 104 S.Ct. 1324, 79 L.Ed.2d 720 (1984). In those cases, we stated:
where a grant of real property to an Indian tribe includes within its boundaries a navigable water and the grant is made to a tribe dependent on the fishery resource in that water for survival, the grant must be construed to include the submerged lands if the Government was plainly aware of the vital importance of the submerged lands and the water resource to the tribe at the time of the grant.
[194]*194Puyallup, 717 F.2d at 1258 (footnote omitted); Muckleshoot, 713 F.2d at 457 (quoting Puyallup). Thus we created a three part test that requires the tribe to show that (1) the reservation grant includes the navigable waters within its borders; (2) the tribe is dependent on the fishery resource in that water for survival; and (3) the government was plainly aware of the vital importance of the water resources to the tribe at the time of the grant.
In formulating our test, we noted the tension between Montana’s application of the strong presumption against such conveyances by the United States, and Choctaw Nations's reliance on the canon of liberal construction in favor of the Indians. Puyallup, 717 F.2d at 1257. We concluded that we must “accord appropriate weight to both the principle of construction favoring Indians and the presumption that the United States will not ordinarily convey title to the bed of a navigable river.” Id. To achieve that goal, we reasoned that whenever it is shown that the government was aware of the importance of water resource to the tribe, the canon of favorable construction to the Indians “warrants the conclusion that the intention to convey title to the waters and lands under them to the Tribe is ‘otherwise made very plain’ within the meaning of HoltId. at 1258 (citation omitted). A tribe’s dependence upon the water’s resources for survival is thus a key factor in determining whether the United State may have intended to retain beneficial title on behalf of the tribe. See id. at 1259. Compare Confederated Salish and Kootenai Tribes of the Flathead Reservation, Montana v. Namen, 665 F.2d 951, 962 (9th Cir.) (tribe’s heavy dependence on fishing could be “public exigency” that required a departure from the federal policy of reserving ownership of land under navigable waters for future states), cert. denied, 459 U.S. 977, 103 S.Ct. 314, 74 L.Ed.2d 291 (1982) with United States v. Aranson, 696 F.2d 654, 666 (9th Cir.) (Indians were not so dependent on the river that Congress would have intended to depart from the equal footing doctrine and convey the riverbed), cert. denied, 464 U.S. 982, 104 S.Ct. 423, 78 L.Ed.2d 358 (1983).
Since our decisions in Puyallup and Muckleshoot, the Supreme Court has decided two cases concerning title to land beneath navigable waters. See Phillips Petroleum Co. v. Mississippi, 484 U.S. 469, 108 S.Ct. 791, 98 L.Ed.2d 877 (1988), and Utah Div. of State Lands v. United States, 482 U.S. 193, 107 S.Ct. 2318, 96 L.Ed.2d 162 (1987). We have reviewed these post-Montana decisions, neither of which concern beneficial title to Indians, and conclude that they do not alter the Puyallup analysis.
There is no dispute in this case that the district court applied the Puyallup standards. Our task therefore is to determine if the court correctly applied the standards and if so, whether its factual findings support its ultimate conclusion that the treaty did not preserve the tidelands for the exclusive use of the tribe. See United States v. Lummi Indian Tribe, 841 F.2d 317, 319 (9th Cir.1988) (historical facts are reviewed for clear error while ultimate treaty interpretation is a mixed question of law and fact reviewed de novo).
Preliminary to this inquiry, however, we must address the contention that the district court erroneously prevented the tribe and the United States from proving their case by limiting the admission of evidence on intent. The record shows that the court ruled prior to trial that evidence of matters occurring after Washington’s admission to the union in 1889 was too remote in time to be probative of the intent of the parties to the Treaty of Point Elliott or to the Secretary at the time of enlargement of the reservation. The tribe contends that the court’s pretrial ruling was erroneous because whenever the meaning of a treaty is at issue, a court is obligated to look to the practical construction placed upon the treaty by the parties. See, e.g., United States v. Top Sky, 547 F.d 486, 487 (9th Cir.1976) (treaty interpretation may include examining the practical construction given the treaty by the parties). The tribe contends that but for the court’s ruling, it would have submitted evidence that the tribe and the United States interpreted the treaty to guarantee the tribe’s exclusive use of the tidelands.
[195]*195We conclude that the trial court did not abuse its discretion in ruling that post-statehood evidence was generally not relevant and hence not admissible. See Kisor v. Johns-Manville Corp., 783 F.2d 1337, 1340 (9th Cir.1986) (evidentiary rulings are reviewed for abuse of discretion). In all cases where courts have reviewed tribal claims of beneficial title to land under navi-gible waters, the inquiry focused on the circumstances surrounding the creation of the reservation. The Supreme Court in Choctaw Nation, 397 U.S. at 622, 90 S.Ct. at 1330, noted that to settle the Cherokees’ claim to exclusive use of parts of the Arkansas River, the Court was required to assess the effect of the treaties “entered into nearly a century and half ago.” Similarly, in Puyallup, 717 F.2d at 1259, we limited our inquiry to “those facts relevant to a determination of whether the United States made plain its intention to convey to the Puyallup Indians that portion of the Puyallup River bed that runs through their reservation.” That inquiry was limited entirely to the events of the 1854 treaty and the 1857 executive order enlarging the reservation. We are thus satisfied that the court did not improperly limit the evidence before it, and we turn to the first of the Puyallup factors.
1. Are the Tidelands Within the Reservation?
We begin by questioning whether it was necessary for the district court under the first Puyallup standard to determine conclusively that the reservation did not include tidelands. In Montana, Puyallup, Muckleshoot and Ñamen, the respective tribes sought beneficial title to navigable waters that either ran through the reservation or were understood to be enclosed by the reservation boundary. Thus, the first Puyallup test—whether the navigable waters at issue are within the boundaries of the reservation—was not at issue in any of those cases. In contrast, the location of the seaward boundary is the very issue of this case. We agree with the the United States that the first Puyallup test should be understood to require only that the grant be capable of being interpreted to include the navigable waters within the reservation boundaries. See Puyallup, 717 F.2d at 1258 n. 7 (calling for flexible application of the test). Such a flexible application is supported by case law in which Indians sought exclusive use or beneficial title to tidelands. In Alaska Pacific Fisheries, 248 U.S. at 87, 39 S.Ct. at 41, the Supreme Court noted that Congress had reserved “the body of lands known as Annette islands” with no mention of whether the grant included tidelands and other submerged lands. The Court looked beyond the words of the grant to conclude that Congress intended to include the disputed lands surrounding the islands, based on the Indians’ reliance on those water resources for their subsistence. Id. at 89, 39 S.Ct. at 42. Similarly, in Skokomish Indian Tribe v. France, 320 F.2d 205, 206 (9th Cir.1963), cert. denied, 376 U.S. 943, 84 S.Ct. 797, 11 L.Ed.2d 767 (1964), we reviewed the Skokomish’s claim of title to a strip of tidelands adjoining their reservation. The treaty did not expressly describe the tidelands at issue. Id. We affirmed the trial court’s rejection of the tribe’s claim, based not on whether the reservation grant encompassed the tidelands but on the fact that “tidelands were not essential to the livelihood of the Indians at the time of the treaty and the executive, order providing for the reservation.” Id. at 212.
Here, the trial court concluded that the “Port Madison Indian Reservation, as enlarged by the Secretarial Order, did not include within, its boundaries any tidelands.” We agree that the tidelands were not expressly included within the description of the reservation boundaries. Nevertheless, we conclude that the treaty language, when construed favorably for the Indians, supports the tribe’s contention that the tidelands may arguably have been intended to be a part of the reservation. See United States v. Adair, 723 F.2d 1394, 1413 (9th Cir.1983) (“any ambiguity in a treaty must be resolved in favor of the Indians”), cert. denied, 467 U.S. 1252, 104 S.Ct. 3536, 82 L.Ed.2d 841 (1984).
First, we believe the trial court placed too great an importance on the treaty mak[196]*196ers’ use and understanding of certain terms in the treaty. The description in the treaty of “two sections” of land, for example, was construed by the court to mean a certain piece of land instead of an amount of land as the Indians would have understood. The court also applied a common law rule that whenever a sea or bay is named as a boundary, the line or ordinary high water mark is always intended. See United States v. Pacheco, 69 U.S. (2 Wall.) 587, 590, 17 L.Ed. 865 (1864) (relating common law rule). Undoubtedly, the makers of the treaty, specifically Governor Isaac I. Stevens, were aware of these technical meanings and rules of survey descriptions. But, as the trial court realized, at the relevant time “the Suquamish Indians would not have conceived of boundaries in the legal sense applied by non-Indians.”
Second, the court narrowly construed the treaty description granting the tribe land “surrounding the small bight at the head of Port Madison, called by the Indians Noo-sohk-um_” The court rejected the tribe’s argument that Noo-suhk-um was Miller Bay, finding instead that the term referred to a small spit located on Miller Bay. A construction more favorable to the tribe would have supported a finding that at least some of the substantial tidelands in Miller Bay were within the reservation.
Finally, we question whether the trial court gave sufficient weight to the tribe’s understanding that their reservation would include tidelands. See Swim v. Bergland, 696 F.2d 712, 716 (9th Cir.1983) (treaties “are to be construed according to the probable understanding of the original tribal signatories”). The court concluded that the tribe “would have conceived of tidelands adjacent to their reserve as containing resources for their exclusive use” and “as part of their domain.” The court nevertheless gave no express weight to that finding in making its determination that the tidelands were not part of the reservation.
Puyallup’s first factor should be applied flexibly to fit the facts and circumstances of each given case. We interpret that first factor in this case to require that the tribe show that the United States may have intended to include the tidelands within the boundaries of the reservation. We conclude that the tribe here met that modified standard by showing that the Treaty of 1855 and the Secretarial Order of 1864 are capable of being interpreted to include the possibility that tidelands at issue are “within” the reservation boundaries.
2. Was the Tribe Dependent on the Tidelands?
The trial court concluded that during the relevant time period, the “Suquamish Indians depended upon resources obtained from tidelands for their survival.” Nevertheless, the court found that the tribe’s primary staple food was salmon and that shellfish and other tideland resources were widely available and gathered throughout the aboriginal territory of the Suquamish Indians. While acknowledging that the adjacent tidelands were important to the tribe “during times of food shortages, epidemics, war, or other crises,” the court nonetheless concluded that the tribe was dependent on other food resource sites. Thus, the court found that “[ojnce the reservation was enlarged, the tidelands adjacent to the reserve encompassed a significant, although less than a majority, of their most important tideland resource procurement sites.”
The tribe contends that the court’s findings are sufficient to meet Puyallup’s second factor—whether the tribe is dependent on the fishery resource for its survival. Both the tribe and the United States argue that Puyallup should not be interpreted to require that a tribe rely exclusively on the disputed water resources. They argue that in all cases in which courts have concluded that Indian tribes have exclusive use of disputed water resources, those tribes also had the right to fish and gather foods off reservation. E.g., Muckleshoot, 713 F.2d at 456 (“treaties secured to the tribes on- and off-reservation fishing rights”).
Although a close question is presented, we conclude that Puyallup requires that the disputed water resource supply a significant portion of the tribe’s fishery needs. The findings made by the [197]*197court indicate that the tidelands at issue were important to the tribe, both as a source of food and as a way of access to other fishing and food gathering locations. It is clear, however, that the district court concluded that the Suquamish Indians did not normally rely on shellfish and other resources gathered from the disputed tidelands, but instead relied on salmon, shellfish, and other food resources from traditional hunting, fishing, and food gathering locations away from the reservation. The court reached this conclusion after listening to testimony, reviewing the various studies submitted by the parties, and resolving the conflicting opinions of expert witnesses. We can find no clear error in its conclusion. See, e.g., Lummi Indian Tribe, 841 F.2d at 319 (“Credibility of witnesses is a quintessentially factual determination which will not be disturbed in the absence of clear error.”). Thus, the tribe failed to meet the second Puyallup factor.
3. Was the Government Plainly Aware of the Importance of the Tidelands to the Suquamish Indian Tribe ?
The creation of an Indian reservation is an “appropriate public purpose” that may justify congressional conveyance of land under navigable water. Montana, 450 U.S. at 556, 101 S.Ct. at 1253; Puyallup, 717 F.2d at 1257. In the absence of an express grant of such land, however, courts have required that there be some “public exigency” to show that Congress was required “to depart from its policy of reserving ownership of beds under navigable waters for the future States.” Montana, 450 U.S. at 556, 101 S.Ct. at 1254. The district court here held that there existed no public exigency either at the time of the treaty or the secretarial enlargement that would have induced the treaty writers to include the tidelands in the grant of reservation land. The court found that unlike the Puyallups or Muckleshoots, the Suquamish did not resort to violence to assure that their respective reservations included a water resource. The court further reasoned that the treaty makers would not have been aware that inclusion of tidelands within the reservation boundaries was necessary, given the fishing and hunting rights provided by Article 5.
The tribe and the United States argue that the district court misconstrued and misapplied the public exigency requirement. Although the tribe admits that Indian hostilities played a key role in Puyallup and Muckleshoot, the tribe argues that such hostilities should not be necessary to create a public exigency. Unlike the Puyallups or Muckeshoots, the Suquamish Indians were not initially denied access to their water resources, so that potential violence was avoided. Both the tribe and the United States argue that a public exigency exists if a tribe is dependent upon the fishery resources of the lands at issue. Compare Montana, 450 U.S. at 556, 101 S.Ct. at 1253 (Crow Indians failed to show public exigency because fishing was not “important to their diet or way of life”), with Namen, 665 F.2d at 962 (public exigency existed because the Kootenai Indians showed that they “depended heavily on fishing”).
The trial court found, however, independent of the lack of Suquamish hostilities, there was “no evidence that the United States’ official or agents perceived that the case area tidelands contained resources of vital importance to the Suquamish Indian Tribe.” We find no clear error in that conclusion. Although it is true that the reservation was obviously designed to assure the Indians’ access to the water, there was insufficient proof that the United States perceived that the tribe depended upon those particular tidelands for survival. Rather, the tribe’s primary food resources were seen by the United States as off-reservation with access guaranteed to the tribe by Article 5 of the treaty. The district court correctly concluded that this is not a case like Puyallup or Muckleshoot involving the ability of the tribe to procure salmon, “the most important staple in the diets of all tribes located on Puget Sound.” Thus, the district court did not erroneously conclude that the tribe did not prove a sufficient dependence on the case area tidelands to warrant an inference that the government was plainly aware of the need [198]*198to restrict non-Indians’ access on those tidelands. See Aranson, 696 F.2d at 666 (court may “infer congressional intent to convey the bed beneath navigable waters if the Indians can prove they depended heavily on the particular body of water”).
II.
As an alternative argument, the tribe contends that Article 26 of Washington's Constitution precludes the state from asserting title to the disputed tidelands. Article 26 required Washington as a condition of statehood to disclaim all right and title to lands “owned or held” by Indian tribes. Wash. Const. art. XXVI. The tribe contends that the subject tidelands were “held” by the tribe in 1889 and accordingly, Washington was required to disclaim title. See United States v. O’Brien, 170 F. 508, 508-09 (C.C.W.D.Wash.1903, 1904) (holding that beaches and shores of reservation island were being used by an Indian tribe and therefore the state was obligated to disclaim all right and title).
The trial court rejected the Suquamish's argument, reasoning that the tribe must still show that it acquired a beneficial grant of the tidelands from the United States. We need not decide if the district court’s reasoning was correct because the court also found that the tribe failed to show that at the time of Washington’s statehood in 1889 tribal members continued to use the tidelands. That finding is not clearly erroneous.
III.
The United States suggests that we decide whether the tribe possesses the right of taking fish and shellfish at their usual and accustomed places on the tidelands involved in this case. We decline to do so. The extent of the Suquamish Indians’ off-reservation shellfishing rights have not yet been judicially determined. The tribe admits that “the scope of the right under Article V is not the issue in this case” and that non-Indians’ present-day exclusion of tribal members from the case area tidelands “is an issue to be resolved elsewhere.”
CONCLUSION
We conclude that the district court was not clearly wrong when it determined that the tribe and the United States failed to prove either (1) the tribe was dependent upon the fishery resource in the disputed tidelands for survival; or (2) the United States was so plainly aware of the importance of the tidelands to the tribe that it may be inferred that the United States intended to reserve the resource for the exclusive use of the tribe. Thus, Puyall-up’s, second and third factors have not been satisfied.
We reject the tribe’s alternative argument that Article 26 of the state’s Constitution provides relief for the tribe. We decline the invitation by the United States to adjudicate or remand for adjudication of the scope of the tribe’s reserved fishing rights.
AFFIRMED.