David Hayashi appeals his conviction of taking a marine mammal in violation of 16 U.S.C. § 1372 (a)(2)(A). We hold that the Marine Mammal Protection Act (MMPA) and the regulations implementing the act do not make it a crime to take reasonable steps to deter porpoises from eating fish or bait off a fisherman’s line. Therefore, we conclude that insufficient evidence supported Haya-shi’s conviction, and we reverse.

I

On the morning of January 24,1991, Haya-shi, a part-time commercial fisherman, and his son were fishing for Ahi off the coast of Waianae, Hawaii. A group of four porpoises began to eat the tuna off Hayashi’s and his son’s lines. Hoping the impact of the bullets hitting the water would scare the porpoises away from their catch, Hayashi fired two rifle shots into the water behind the porpoises. The shots did not hit the porpoises. When the Hayashis reeled in their lines,'they discovered that a porpoise had in fact eaten a part of at least one of the tuna.

A state enforcement officer reported to the National Marine Fisheries Service (NMFS) that occupants of Hayashi’s vessel had fired at dolphins. In February 1991, NMFS agents interviewed Hayashi and his son, taking written statements from each. An April 22, 1991 information charged Hayashi with knowingly taking a marine mammal in violation of the MMPA, 16 U.S.C. § 1372 (a)(2)(A).

The parties consented to proceed before a magistrate judge. In July 1991, after denying Hayashi’s motion to dismiss the information for unconstitutional vagueness, the mag- ' istrate judge tried and convicted Hayashi on stipulated facts. The submitted facts consisted of the Hayashis’ statements, and an NMFS agent’s report and notes on the interviews of the father and son. Hayashi appealed to the district court, renewing his vagueness argument and raising a claim of insufficient evidence. In December 1991, without oral argument, the district court affirmed the conviction by written order. Hayashi appeals on grounds of unconstitutional vagueness and insufficiency of the evidence. We agree that his. conviction must be reversed. 1

II

The MMPA declares it,unlawful for any person to “take” a marine mammal in United States waters. See 16 U.S.C. § 1372 (a)(2)(A). “The term ‘take’ means to harass, hunt, capture, or kill, or attempt to harass, hunt, capture, or kill any marine mammal.” 16 U.S.C. § 1362 (13). The MMPA prescribes both civil and criminal penalties, but the latter apply only to persons who “knowingly” violate any provision of the act. See 16 U.S.C. § 1375 (b).

The government agrees that the only definition of “take” with possible application to. Hayashi is “to harass” or “attempt to harass.” The statute itself fails to define “harass.” Various agencies of the federal government have promulgated regulations implementing the MMPA. 2 The regulations applicable to porpoises, issued by the NMFS, do not define “harass” but further define “take” as including:

50 C.F.R. § 216.3 . 3 The “disturbing or molesting” example is the only regulatory definition potentially applicable to Hayashi’s act of firing a rifle into the water behind a group of porpoises to scare them away from his fishing lines. 4 We conclude that the regulation does not reach the conduct underlying Hayashi’s conviction.

A

Initially, we note that two substantial errors infected the proceedings before the magistrate judge and the district court. First, both parties, the magistrate judge, and the district court all employed the incorrect regulatory definition of the charged crime. Second, the district court’s affirmance rested, in part, upon the erroneous belief that negligent acts are criminally punishable under the MMPA.

In opposing Hayashi’s motion to dismiss, the government suggested using the definition of “harass” set forth in 50 C.F.R. § 17.3 , “an intentional or negligent act or omission which creates the likelihood of injury to wildlife by annoying it to such an extent as to significantly disrupt normal behavioral patterns which include ... breeding, feeding, or sheltering.” At oral argument, the government did submit for the magistrate judge’s consideration the regulatory definition in section 216.3. Nonetheless, in answering Hayashi’s appeal to the district court, the government again argued that section 17.3 was the appropriately-controlling regulatory definition. Hayashi acquiesced in and promoted this view. 5

As a result of the parties’ briefing, both the magistrate judge at trial and the district court on appeal employed the regulatory definition in 50 C.F.R. § 17.3 , which does not define the crime with which Hayashi was charged. 6 The regulation differs from the definition of “take” in 50 C.F.R. § 216.3 , particularly in the former’s specific reference to disruption of “feeding.” Section 17.3, issued by a different agency in a different federal department from the NMFS, implements an entirely separate statute, the Endangered Species Act. 7 In effect, the district court entirely ignored the appropriate regulatory definition set out in section 216.3. Ha-yashi’s conviction and its affirmance rest *1281 upon application of the wrong regulatory definition. 8

Consistent with the MMPA criminal penalty scheme, the information charged Hayashi with “knowingly” taking a marine mammal. Under the MMPA, no criminal penalty can attach for negligent conduct. However, in referring the district court to the regulatory definition in section 17.3, which includes negligent acts,, the parties failed to inform the court of this statutory requisite of “knowing” conduct for criminal liability to attach. Indeed, in briefing his appeal to the. district court, Hayashi argued that the MMPA specifies no mens rea element. In answering Hayashi’s brief, the government took no steps to draw the district court’s attention to 16 U.S.C. § 1375 (b), and its explicit “knowledge” requirement.

Apparently unaware of this clear statutory command, the' district court, relying upon 50 C.F.R. § 17.3 , concluded that a criminal violation of the MMPA could rest upon either negligent or intentional conduct. 9 In addressing Hayashi’s insufficiency claim, the district court held that, “[Hiring the rifle in waters containing porpoises was a negligent act that created a likelihood of injury to the porpoises.” Although it went on to suggest that the evidence also showed an intentional attempt to stop the porpoises from eating, the district court’s erroneous belief that negligence could support an MMPA criminal conviction tainted its consideration of Haya-shi’s appeal. Thus, Hayashi’s conviction and-its affirmance rested not only upon an incorrect regulatory definition, but also upon a misunderstanding of the required mens rea.

These errors affected the two most basic elements of every criminal proscription — the actus reus, or act itself, and the mens rea, or mental element required for criminal liability. The errors resulted not from a misunderstanding of obscure interpretive gloss, but from a basic misreading of clear statutory and regulatory commands. Although every lawyer involved was eomplicit in these errors, the responsibility for prosecuting the correct crime lies ultimately with the government. 10

Under our precedent, if this case had been tried to a jury with instructions. similarly incorrect as to both the definition of the crime and its mens rea, we would not hesitate to reverse. See, e.g., United States v. Washington, 819 F.2d 221, 226 (9th Cir.1987) (“ambiguous and equivocal” instructions on “basic issue” of level of intent is reversible error); United States v. Combs, 762 F.2d 1343, 1346 (9th Cir.1985) (instruction broadening possible basis for conviction “was so lacking in explanation of the offense charged and its elements that it was fundamentally erroneous and inadequate”). 11

*1282 Here, we need not determine whether a similar result should obtain in a case tried to a magistrate judge rather than to a jury. Hayashi was tried on stipulated facts; the magistrate judge was npt required to resolve any factual disputes or to weigh the credibility of witnesses. We have a clear factual record from which to review Hayashi’s conviction, and can do so without further perpetuating the fundamental errors described above. Therefore, cognizant of, but not hindered by, the elemental mistakes that pervaded Hayashi’s trial and district court appeal, we consider whether there was sufficient evidence to convict Hayashi of the “knowing taking” of a marine mammal in violation of the MMPA.

B

As noted above, the government contends that Hayashi committed a criminal “taking” by “harassing” the porpoises; “harass” is left undefined by the statute. See 16 U.S.C. § 1362 (13). If we look to the administrative regulations, the only form of “taking” specified in 50 C.F.R. § 216.3 that is potentially applicable to Hayashi’s act of firing into the water behind the porpoises to divert them from his fishing lines is the prohibition of any “intentional act which results in disturbing or molesting a marine mammal.” However, “disturb” and “molest” are not much more definitive than “harass.” All three are very general terms, requiring us to resort to their context to ascertain their meaning.

Following the “familiar principle of statutory construction that words grouped in a list should be given related meaning,” we look to .the other statutory and regulatory examples of “taking.” Third Nat’l Bank in Nashville v. Impac Ltd., 432 U.S. 312, 322 , 97 S.Ct. 2307, 2313 , 53 L.Ed.2d 368 (1977). The statute groups “harass” with “hunt,” “capture,” and “kill” as forms of prohibited “taking.” The latter three each involve direct, sustained, and significant intrusions upon the normal, life-sustaining activities of a marine mammal; killing is a direct and permanent intrusion, while hunting and capturing cause significant disruptions of a marine mammal’s natural state. Consistent with these other terms, “harassment,” to constitute a “taking” under the MMPA, must entail a similar level of direct and sustained intrusion.

Interpreting “harassment” under the MMPA to involve a sustained, direct, and significant intrusion also comports with a common understanding of the term “take,” of which “harass” is simply one form. To “take” a marine mammal strongly suggests a serious and sustained diversion of the mammal from its natural routine. Congressional concern in passing the MMPA about marine mammals “in danger of extinction or depletion as a result of man’s activities” supports this conception of “take.” MMPA § 2(1), Pub.L. No. 92-522, 86 Stat. 1027 (1972) (findings and declaration of policy). 12 Killing, capturing, and hunting fit the common understanding of the term. “Harassment” under the MMPA, to constitute a “taking,” must entail a similarly significant level of intrusiveness.

The NMFS regulatory definition of “take,” as applied to porpoises, also supports this interpretation. See 50 C.F.R. § 216.3 . The detention and the tagging of a porpoise are each direct and significant intrusions upon the mammal’s ordinary activities. The collection of dead mammals also involves an extended and direct intrusion upon the porpoise in the wild; “collection” suggests sustained activity to locate — or even to facilitate the -demise of — deceased or dying marine mammals. The basic purpose of the “collection” prohibition is undoubtedly to deter difficult-to-detect killings of mammals by prohibiting possession of dead mammal by-products. In the context of these other examples, “molesting or disturbing” in 50 C.F.R. § 216.3 must refer to sustained and severe disruptions of a porpoise’s natural behavior.

*1283 The case for our interpretation is even stronger if we refer to 50 C.F.R. § 17.3 , the regulatory definition of “harass” urged by the government and upon which the district court exclusively relied. As discussed above, section 17.3, issued - under the Endangered Species Act, is not the controlling regulatory definition and is an improper starting point for any MMPA prosecution; it is nonetheless analogous authority that is-of considerable assistance in interpreting 50 C.F.R. § 216 .-3. 13 Section 17.3 specifically requires significant disruption of “normal” behavioral patterns. Deterrence of abnormal marine mammal activity is not proscribed.

This emphasis upon protecting natural animal behavior comports with the MMPA emphasis upon marine mammals as essential components of the natural marine ecosystem. See MMPA § 2(2), Pub.L. 92-522, 86 Stat. 1027 (1972) (marine mammals “should not be permitted to diminish beyond the point at which they cease to be a significant functioning element in the ecosystem of which they are a part”). The concern that underlies the prohibition against disturbing mammals was for mammals as a part of nature, not for mammals acting in ways that endanger human life or property.

Interpreting the act and regulations otherwise, as prohibiting isolated interference with abnormal marine mammal activity, would lead to absurdity. Under such a broad interpretation, anyone who acted to prevent or in any way interfered with any marine mammal activity would face potential criminal prosecution. Nothing could legally be done to save a modern-day Jonah from the devouring whale, or to deter a rampaging polar bear from mauling a child. Neither could a porpoise intent on swimming into severely contaminated waters, or into the propellers of a motorized boat, be diverted by the selfless actions of a good Samaritan. These are but examples of what the unreasonably broad interpretation advocated by the government would lead us to. Our conclusion that only sustained and serious disruptions of normal mammal behavior fall under the term “harass” comports with a more reasonable understanding of the extent and scope of the MMPA. 14

Applying our interpretation to the act for which Hayashi was convicted, we conclude that there was insufficient evidence to find a criminal “taking” by “harassment.” The stipulated facts, consisting almost entirely of Hayashi’s and his son’s statements to the NMFS investigator, show that Hayashi did not fire at the porpoises, nor did he hit them. He simply fired two successive shots behind and outside the area of the porpoises to discourage them from eating bait and hooked tuna from his fishing lines — -an act that is not a part of the porpoise’s normal eating habits. Even if the shots succeeded in scaring away the porpoises — and the stipulated facts do not tell us whether the porpoises were aware of, or reacted to, the shots— any diversion from eating off the fisherman’s lines is not of the significance or sustained effect required for a “taking” under the MMPA. Hayashi’s conduct was not the kind of direct, sustained disruption of a porpoise’s customary pursuits required to find a criminal “taking.” Reasonable acts to deter porpoises from eating fish or bait off a fisherman’s line are not criminal under the MMPA. 15

*1284 Our conclusion that the MMPA and the regulations implementing the act do not reach Hayashi’s act finds further and more specific support in the most recent NMFS amendment of the regulatory definition of “take.” Effective, April 19, 1991, the NMFS added “feeding or attempting to feed” to section 216.3 as an example of “harassment” under the MMPA. See 56 Fed.Reg. 11,693 (1991) (final rule); 55 Fed.Reg. 35, 328 (1990) (proposed rule) (“feeding activities may alter the behavior of marine mammals in such a way that they constitute harassment”). One concern motivating the regulatory amendment was the belief that human feeding of marine mammals, primarily dolphins and porpoises, led to potentially dangerous modification of normal feeding behavior. See 56 Fed.Reg. at 11,696 (“feeding ... is harmful because- it disrupts their natural behavior and normal feeding patterns”).

In this regard, the NMFS specifically noted that “the- dead fish offered to the animals might condition them to seek- other dead fish such as those found on baited hooks or in fish nets.” Id. The Marine Mammal Commission, in commenting favorably upon the proposed rule, noted that feeding may condition animals to approach watercraft “where there is an increased likelihood that they will become entangled in fishing gear, be struck by' vessels, or be shot, poisoned, or fed foreign objects.” Id. at 11,695 .

Thus, the regulatory amendment proscribing “feeding” was intended, in the long run, to deter precisely the kind of porpoise behavior that Hayashi witnessed and attempted himself to deter, in the short run, by firing the rifle into the water. It would make little sense for the MMPA to prohibit reasonable private measures to discourage the type of animal behavior that the NMFS has specifically identified as detrimental ■ to marine mammals. The most recent NMFS modification of the regulatory definition of “take” supports our interpretation of the MMPA as not reaching Hayashi’s conduct.

Finally, we note the quandary potentially created for one in Hayashi’s position were we not to hold that the MMPA allows reasonable steps — those not resulting in sustained and severe disruption of the mammal’s life activities — to deter porpoises from feeding on bait and hooked fish. Allowing porpoises to feed on bait might well run afoul of the “feeding” prohibition in section 216.3 and bring the fisherman to the attention of the vigilant enforcers of our criminal -laws. 16 If reasonable steps to divert porpoises from a fishing line were prohibited, the fisherman would face an impossible conundrum. Guilty of “harassment” by “feeding” if he did nothing, and guilty of “harassment” by “disturbing” if he took steps to prevent the feeding, our hypothetical Hayashi would face possible criminal prosecution under the MMPA no matter which course he chose. We do not believe that Congress or the NMFS intended to create such a “fisherman’s dilemma.”

Ill

We hold that reasonable actions — those not resulting in severe, sustained disruption of the mammal’s normal routine — to deter porpoises from eating fish or bait off a fishing line are not rendered criminal by the MMPA or its regulations. Because the evidence shows that Hayashi’s action was reasonable, we conclude that insufficient evidence supported his conviction. ' We reverse.

REVERSED.

. Because we reverse on Hayashi’s insufficiency argument, we "do not reach his unconstitutional vagueness claim.

. Both parties agree that administrative regulations clarify the prohibitions in the MMPA.

. The final "feeding” clause was added to section 216.3 by an amendment effective April 19, 1991. See 56 Fed.Reg. 11,693 (1991) (final rule). The amendment was therefore not in effect, although it had been proposed, when Hayashi committed the charged acts. It was in effect, however, when the information charging Hayashi was filed, and when he was tried.

. Although this regulatory clause and others include negligent acts, as noted above, only "knowing” violations can support .criminal conviction under the MMPA.

. Ignoring its issuance in implementation of a different statute, Hayashi argued that the 17.3 definition controls because it is more specific to the term "harass.” However, the "disturbing or molesting” clause in 50 C.F.R. § 216.3 seems most directly applicable to "harass” rather than to the other statutory forms of "take” — hunt, capture, or kill. Therefore, although technically defining "take," the "disturbing or molesting" clause in section 216.3 is logically as specific to "harass” as section 17.3. Cf. 56 Fed.Reg. 11,694 (1991) (NMFS) (referring to "disturbing or molesting standard of the current regulation” with regard to definition of "harass” under MMPA).

. On briefing this appeal, neither party identified 50 C.F.R. § 216.3 as,the appropriate regulation. Instead, both relied upon section 17.3. In addition, the government cited, for the first time, 50 C.F.R. § 18.3 , which is an MMPA regulation issued by the Fish and Wildlife Service, a separate agency in a different department from the NMFS. -Although section 18.3 and 216.3 are nearly identical, section 18.3, by its explicit terms, does not apply to porpoises, but to polar bears, otters, walrus, and manatees. We eventually ordered supplemental briefing in order to permit the parties to address the proper regulation, section 216.3.

. This is not to suggest that section 17.3 may not be useful analogous authority. However, even if a regulation implementing a different statute might aid in interpreting those under another *1281 statute, the starting point still must be the regulation facially applicable to the statute in question.

. On supplemental briefing, the government argues that this mistake "was not significant because section 17.3 is very similar to section 216.3. Even so, the two sections are facially distinct. Moreover, no level of similarity excuses the government’s failure to bring its own facially-applicable regulation to the courts’ attention at some point prior to supplemental appellate briefing in response to an order specifically identifying section 216.3.

. Although the court relied upon section 17.3, it might well have made the same error even had it resorted to the proper regulation. 50 C.F.R. § 216.3 also includes "negligent or intentional” acts. Only reference to the statute makes clear that criminal "taking” must be "knowing."

. We are tempted to note the complicity of others not directly represented in these proceedings. Suffice it to say that as the catalog of federal crimes expands, we can'only expect basic errors stemming from lack of familiarity with individual legal proscriptions to proliferate fur-thér.

. Even when the defendant failed to raise the issue in the district court, we have reversed, holding that an instruction permitting the jury to convict for an uncharged crime is plain error. See United States v. Vowiell, 869 F.2d 1264, 1271 (9th Cir.1989) (instruction permitted conviction for harboring when charge was assisting escape). We have excused the failure to raise the issue on appeal as well as in district court. See United States v. Arrellano, 812 F.2d 1209, 1211 , modified 835 F.2d 235 (9th Cir.1987) (erroneous instruction as to required mental state). Finally, we have even reversed when defense counsel affirmatively collaborated in the error. See United States v. Solis, 841 F.2d 307, 309 (9th Cir.1988) (instruction on uncharged crime in addition to charged crime is plain error despite defense acquiescence and collaboration). Thus, Hayashi’s failure to identify the fundamental errors that infected his conviction would not prevent reversal on those grounds had this case been tried to a jury-

. .In describing the need for legislation, the House Report accompanying the MMPA refers to marine mammals that “have been shot, blown up, clubbed to death, run down by boats, poisoned, and exposed to a multitude of other indignities.” H.Rep. No. 707, 92d Cong., 1st Sess. (1971), reprinted in 1972 U.S.C.C.A.N. 4144, 4144. This list also suggests that the "indignities" addressed by the MMPA prohibition of "taking" are seriously intrusive acts.

. The NMFS has referred to section 17.3 in interpreting the meaning of "harass” under the MMPA. See 56 Fed.Reg. 11,695.

. Although we do not reach Hayashi's vagueness challenge, we note that in the absence of our appropriately restrictive construction of the MMPA and its regulations, “harass” would raise a serious issue of adequate notice to potential violators.

. What is reasonable will vary with the circumstances. We emphasize, however, that the reasonableness of deterrent steps rests on their impact on the marine mammal alone. For instance, in sentencing Hayashi, the magistrate judge noted that "it's dangerous to be shooting from a boat because there’s a lot of other fishermen around and you could ricochet and you’re just looking for trouble.” From a human safety perspective, Hayashi's act may have been unreasonable, but the MMPA does not address human safety. Conversely, physically clubbing the porpoise with a baseball bat would have been more reasonable from a human safety perspective, but the injury from such a clubbing would almost certainly result in a severe, sustained disruption of the porpoise’s life activities.

. Indeed, if it did not, then the dolphin feeding cruises targeted by the NMFS “feeding"- amendment could circumvent the new regulatory prohibition by putting the food they use to attract dolphins on hook and line. The same deleterious porpoise behavior .modification intended to be prevented by the “feeding” prohibition would undoubtedly follow from this new form of "fishing” by dolphin feeding cruises.