Defendant Wynn Lewis Robinson appeals his sentence of 63 months, pursuant to his plea of guilty to manufacturing marijuana with intent to distribute, a violation of 21 U.S.C. § 841 (a)(1). Robinson’s appeal presents a number of challenges to the district court’s finding that Robinson, at the time of his arrest, was cultivating 156 marijuana plants. He argues that the district court erred in counting as “plants” forty-eight rootless marijuana cuttings, that it erred in counting individual stalks growing from intertwined root systems as separate plants, that its finding that there were 108 mature plants is not supported by a preponderance of the evidence, and that it should have held an evidentiary hearing concerning the possibility that some of those 108 plants were no longer (or never had been) in the government’s possession.

We reverse the district court’s determination that rootless cuttings are to be counted as plants for sentencing purposes, and affirm in all other respects. Although the length of Robinson’s sentence is not affected, we remand so that the record can be corrected to reflect a sentence based on 108 rather than 156 marijuana plants.

On December 11, 1991, state police and federal Drug Enforcement Administration (DEA) agents executed a search warrant at Robinson’s home and adjoining cucumber and tomato business, Crest Hydroponics. They discovered that in some of Crest’s greenhouses Robinson was growing a less edible, more lucrative crop: marijuana.

Deputy Sheriff Dan Jopes of the San Diego Sheriffs Office supervised the search. According to his testimony, agents found marijuana plants in three rooms. In the first room plants three to five feet tall were growing in rubber pots. Most of these pots contained more than one growing stalk; Jopes counted a total of forty-one stalks. In the second room plants four to seven feet tall were growing in pits dug into the ground. Again, most of the pits contained more than one stalk; Jopes counted sixty-seven stalks. In the third room agents found forty-eight small cubes which each contained one rootless marijuana cutting. After Jopes made his count, agents took the cuttings, pulled out some of the smaller plants, and cut down the rest near the bases of their stems. The cuttings and plants were subsequently held as evidence by the DEA.

An indictment was filed on December 18, 1991, charging Robinson (along with his wife and one of their alleged dealers) with conspiracy to manufacture marijuana with intent to distribute, and manufacturing marijuana with intent to distribute. At the stipulation of the parties, a series of evidentiary hear ings were held prior to the entry of Robinson’s plea, in order to determine the number of marijuana plants seized. On April 20, 1993, the district court found that Robinson’s sentence should be based on 108 mature plants and forty-eight cuttings (which, it held, counted as plants for sentencing purposes), for a total of 156 plants. On April 29, 1993, Robinson pled guilty to the charge of manufacturing marijuana with intent to distribute.

At a sentencing hearing held on August 10, 1993, the district court denied Robinson’s requests to redetermine the number of marijuana plants and to conduct an evidentiary inquiry into the alleged discrepancy between Jopes’ count and the number of plants currently in the DEA’s possession. The court then based Robinson’s sentence on 156 plants, as it had previously indicated it would. Under the sentencing guidelines, 156 plants are counted as the equivalent of 156 kilograms of marijuana. U.S.S.G. § 2Dl.l(c), note following Drug Quantity Table (stating that if more than 50 plants are involved, one plant equals one kilogram of marijuana).

The district court’s finding that Robinson had been cultivating more than 100 plants had a weighty effect on his sentence. Under 21 U.S.C. § 841 (b)(l)(B)(vii), persons convicted of manufacturing more than 100 kilograms of marijuana face a mandatory minimum sentence of five years. And under the guidelines, persons convicted of manufacturing 100-400 kilograms of marijuana receive an offense level of 26. U.S.S.G. § 2Dl.l(e)(9). In Robinson’s case, these provisions led to a guidelines range of 63-78 months. The district court sentenced him to 63 months.

II.

Robinson first challenges the district court’s determination that the forty-eight rootless cuttings seized from Robinson’s greenhouse counted as plants for sentencing purposes. What counts as a plant for sentencing purposes is a question of law, to be reviewed de novo. United States v. McConney, 728 F.2d 1195, 1200-01 (9th Cir.), cert. denied, 469 U.S. 824 , 105 S.Ct. 101 , 83 L.Ed.2d 46 (1984). Unfortunately, “[t]he guidelines do not define ‘marijuana plant’ and do not distinguish between cuttings and mature plants.” United States v. Carlisle, 907 F.2d 94, 96 (9th Cir.1990) (per curiam).

We have not previously considered whether rootless cuttings constitute plants for sentencing purposes. In Carlisle , however, we upheld the district court’s determination that certain cuttings with roots counted as plants. We reasoned,

Id. (emphasis added). Our holding there thus depended in part upon a finding that, for guidelines purposes, roots are an important characteristic of marijuana plants.

Other circuits have unanimously decided that marijuana cuttings must have root formations to be considered plants for sentencing purposes. United States v. Burke, 999 F.2d 596, 601 (1st Cir.1993) (“at the first sign of roots, a plant exists for sentencing purposes”); United States v. Edge, 989 F.2d 871, 877 (6th Cir.1993) (plant exists if there is “some readily observable evidence of root formation”); United States v. Curtis, 965 F.2d 610, 616 (8th Cir.1992). These courts have found that Congress intended to create a clear, easily-implemented standard rather than one requiring extensive litigation and expert botanical testimony. E.g., Edge, 989 F.2d at 878 (quoting United States v. Eves, 932 F.2d 856, 860 (10th Cir.), cert. denied, — U.S. -, 112 S.Ct. 236 , 116 L.Ed.2d 192 (1991)).

Today we join these circuits. Marijuana plants have three characteristic structures, readily apparent to the unaided layperson’s eye: roots, stems, and leaves. Until a cutting develops roots of its own, it is not a plant itself but a mere piece of some other plant. We therefore adopt the rule that cuttings are not “plants” unless there is “readily observable evidence of root formation.” Burke, 999 F.2d at 601 ; Edge, 989 F.2d at 879 . We think that requiring readily observable evidence of root formation is a common-sense approach that will prevent the costly and confusing “battle of botanical experts which occurred in this case.” Appellee’s Brief at 13.

Forty-seven of the cuttings seized from Robinson’s greenhouse had no root formations. The district court therefore erred in counting them as plants for sentencing purposes. With regard to the forty-eighth cutting, however, we must resolve a borderline issue under the new rule announced above. Specifically, that cutting had a structure which the government’s expert characterized as “root primitiae.” This structure was undisputedly no more than a swelling of the cambial cells underneath the cutting’s outer layer of plant tissue. It was therefore a structure from “which roots will come,” Edge, 989 F.2d at 879 , not a root structure itself. Since this cutting, like the others, has not yet produced readily observable roots or root hairs, the district court erred in treating it as a plant for sentencing purposes. See id. (rejecting the argument that a “swollen area” or “callus” from which roots would eventually come qualified the cutting as a plant).

III.

Robinson next challenges the district court’s holding that individual stalks protruding from common root systems should be treated as separate plants at sentencing. Again, this is a legal determination which we review de novo. McConney, 728 F.2d at 1200-01 . It, too, is a question not previously ruled upon by this circuit.

According to the testimony of Deputy Jopes and of the defense expert, Dr. Sanees, quite a few of Robinson’s mature marijuana plants were growing in close proximity to one another. Many of the pots or dirt pits in which the plants were growing contained more than one stalk. Robinson maintained before the district court that because these plants shared a common root system they should be counted as one plant for sentencing purposes. The district court initially agreed, but ultimately reversed course and held that each stem or stalk protruding from the dirt should be counted as a separate plant.

The record is clear on at least some of the botanical background. The experts agree that marijuana is a single stem plant, in that one seed, when germinated, will typically produce one stem. Moreover, when two or more marijuana plants grow in close proximity, their roots do not fuse but rather intertwine and intermingle. Over time, these roots may become so tangled that the two plants cannot be separated without damaging one or both sets of roots. Plants with such intertwined root systems share a common root environment in that they receive the same amount of water, fertilizer, and nutrients. The defense expert found that almost half of Robinson’s plants shared common root systems with one or two other plants.

Robinson argues that a common root system should be counted as one plant for sentencing purposes whether it has one, two, or more stalks protruding from it. He argues that this approach would be consistent with horticultural practice as well as common sense. As for horticultural practice, Robinson cites only to his own expert’s testimony:

Transcript of Apr. 16, 1993, at 134. This is unpersuasive; whether the multiple-stem plants are “sold as” one plant has no bearing on whether horticultural science really so characterizes them. Moreover, when tested for common sense Robinson’s argument fares no better. He would apparently have us consider two petunias growing with intertwined roots in a single pot to be “one plant” even though the flowers of one are red and those of the other, blue.

Horticultural metaphysics aside, Robinson’s remaining argument is functional: “[s]ince each ‘plant’ under the Sentencing Guidelines is treated as the equivalent of one kilogram of marijuana, the definition of what constitutes a plant should take into account the factors that affect the yield of a plant.” Appellant’s Brief at 23-24. He claims that “the primary factor in determining the yield of a marijuana plant is root growing space, not number of stems or trunks. ” Id. (emphasis supplied). He thus concludes that because two or three plants sharing a common root system will not produce as much marijuana as two or three plants growing separately, they should only be counted as a single plant.

There are several problems with Robinson’s reasoning. First, if “root growing space” is really the primary factor in determining what counts as a plant for sentencing purposes, Congress would no doubt have specified that plants growing in small pots are to be penalized less harshly than those growing in open fields; needless to say, it has not. Second, and more importantly, Robinson fails to recognize that the guidelines’ “one plant equals one kilogram” formula is a legislative fiction rather than an accurate statement of the actual yield of particular marijuana plants. See United States v. Jordan, 964 F.2d 944, 947 (9th Cir.) (guidelines’ “treatment of marihuana plants does not purport accurately to translate the amount of marihuana harvestable from a given plant... .[but rather to assign] a higher level of culpability for marijuana growers compared to those who merely possess the harvested product.”) (emphasis supplied; quotations omitted), cert. denied, — U.S. -, 113 S.Ct. 478 , 121 L.Ed.2d 384 (1992). The government rightly points to United States v. Barton, 995 F.2d 931 (9th Cir.), cert. denied, — U.S. -, 114 S.Ct. 413 , 126 L.Ed.2d 359 (1993), and United States v. Traynor, 990 F.2d 1153 (9th Cir.1993), to establish that the guidelines formulation is not dependent on yield. In those cases, we held that the sentencing guidelines did not violate the Due Process Clause in penalizing male and female plants alike, despite the fact that male plants have very little THC content — and thus a much lower yield of smoka-ble marijuana. The Traynor court said pointedly that the “direction of authority is not to consider a particular plant’s [yield] when applying the straightforward language of 21 U.S.C. § 841 (b)(1)(D).” 990 F.2d at 1160 ; see also United States v. Beaver, 984 F.2d 989, 991 (9th Cir.1993) (rejecting argument that weight or potential yield of marijuana plants is relevant under 21 U.S.C. § 841 (b)(1)).

Moreover, Robinson’s proposed rule would create perverse incentives. If any and all plants with intertwined root systems were to count as only one plant for sentencing purposes, growers would be encouraged to grow four, five, ten or more plants in large, root-bound clusters. While such arrangements might not produce quite as much marijuana as would the same number of plants growing independently, growers would undoubtedly view the loss in total yield to be more than compensated for by the dramatic decrease in the length of their expected sentence, were they to be apprehended.

We find the district court’s rule eminently sensible, and therefore adopt it. Each stalk protruding from the ground and supported by its own root system should be considered one plant, no matter how close to other plants it is and no matter how intertwined are their root systems. This rule best effectuates the intentions of Congress and the Sentencing Commission that marijuana growers be sentenced in accord with the number of individual plants they have cultivated. Moreover, like the rule for marijuana cuttings adopted above, it is a clear rule and one that is unlikely to lead to excessive botanical litigation. See Edge, 989 F.2d at 878 .

IV.

Apart from the forty-eight cuttings, the district court found that agents had seized 108 mature plants at Robinson’s property. Robinson challenges this finding in two ways. First, he argues that the government should have been bound by the testimony of its own expert, Dr. Johnson, who reported seeing fifty-four plants in the DEA’s evidence storage facility, five months after Robinson’s arrest. Second, he argues more generally that the district court clearly erred in finding, under the preponderance of the evidence standard, that 108 plants had been seized. Both arguments require some additional factual background.

Deputy Jopes credibly testified that in executing the search warrant on December 11, 1991, he counted forty-one mature plants in one room and sixty-seven in another, for a total of 108 plants. Dr. Sanees, the defense expert, testified that on January 23, 1992, he visited Robinson’s greenhouse and made an inspection. Some of the smaller plants had been pulled out of the ground, roots and all, and were not counted by Dr. Sanees. He testified that he saw twenty-seven “root systems” in one room and forty-eight root systems in the other, for a total of seventy-five root systems. He noted that thirty-six of these seventy-five root systems had two, three, or four trunks protruding from them. Thus, conservatively assuming that these thirty-six multiple root systems were only “doubles,” the total number of actual plants indicated by Dr. Sance’s observations would be seventy-five plus thirty-six,-or 111. Dr. Sanees admitted as much; indeed, he admitted that if one counted each stalk rather than each root system, the total number of mature plants was “closer to 130.” Transcript of Apr. 20, 1993, at 38-39.

In May, 1992, five months after the marijuana plants were seized, and about four months after Dr. Sances’s inspection, Drs. Sanees and Johnson examined the marijuana plants at the DEA evidence storage facility. They were shown two boxes which, they agreed, contained fifty-four stems. Dr. Sanees testified that he asked the DEA worker who had retrieved the boxes for them if there were any more boxes (since he expected there to be 108 plants, as Deputy Jopes had reported). She replied that there were no more boxes of evidence in this ease. Dr. Johnson, on the other hand, testified that “[w]e were told by the person in the examination room that the key to the vault where the plants were located was held by another person, and that person was absent that day due to illness.” Id. at 40.

The defense attempted to subpoena the “missing box,” but was unsuccessful. The government stood by its missing box theory until the sentencing hearing, at which point it conceded that there had never been a third box of mature plants. It offered no explanation for the missing evidence. Before us, the government once again advances the missing box theory.

The district court inquired into the discrepancy between Jopes’s and Sances’s original plant counts and the amount of marijuana actually in the DEA’s possession, but got no solid answers from counsel. The court then said, based on the testimony of Jopes and Sanees,

Transcript of Aug. 30, 1993, at 20-21. Because the court also considered the forty-eight cuttings to be plants, it based Robinson’s sentence on the cuttings along with the 108 mature plants, for a total of 156 plants.

Citing Fed.R.Evid. 607, Robinson argues that the government should not have been permitted to abandon Dr. Johnson’s testimony that he saw only fifty-four plants in the DEA’s possession. This argument is merit-less. The government accepted Dr. Johnson’s testimony as a true report of his observations at the DEA in May, 1992. It argued, however, that he had not seen all the marijuana seized on December 11, 1991. In this regard, the government relied on the testimony of Deputy Jopes, also a government witness, who testified that he had counted 108 plants, and on that of Dr. Sanees, a defense witness, who testified that he had seen evidence of more than 100 plants during his inspection in January, 1992.

The government rightly points out that if anybody should be bound by their expert’s testimony, it is Robinson. His expert, Dr. Sanees, actually visited the greenhouse and investigated the remaining roots and stumps, and found seventy-five root systems with at least 111 stems protruding. Based on Dr. Sances’s report, Robinson’s own initial position in the district court was that the indictment should be altered to charge only seventy-five plants.

We next consider whether a preponderance of the evidence supports the district court’s finding of 108 mature plants. “[T]he preponderance of the evidence standard is a meaningful one that requires the judge to be convinced ‘by a preponderance of the evidence that the fact in question exists.’ ” Restrepo, 946 F.2d at 661 [citation omitted). Here, the testimony of Deputy Jopes alone would have been sufficient to establish the existence of 108 plants. See United States v. Belden, 957 F.2d 671, 673-74 (9th Cir.) (conviction based on testimony of officers who made the initial plant counts; a number of the plants were subsequently mangled or destroyed), cert. denied, — U.S. -, 113 S.Ct. 234 , 121 L.Ed.2d 169 (1992); United States v. Cody, 7 F.3d 1523, 1527 (10th Cir.1993) (conviction based on officer’s recollection of number of plants); United States v. Scalia, 993 F.2d 984, 988 (1st Cir.1993) (same); United States v. Allen, 954 F.2d 1160, 1168-69 (6th Cir.1992). Jopes’s testimony did not stand alone, however: it was strongly corroborated by the testimony of Dr. Sanees, who investigated Robinson’s greenhouses one month after the arrest, and concluded that more than 100 plants had been growing there. See United States v. Lennick, 18 F.3d 814, 820 (9th Cir.1994) (permissible to estimate number of plants based on circumstantial evidence such as the amount of harvested marijuana, number of pots, and size of utility bills).

The sole question, then, is whether the testimony of Jopes and Sanees is sufficiently undermined by the fact that Johnson and Sanees, five months later, counted only fifty-four plants in the DEA’s possession. We think not. The missing plants are unquestionably troubling, as is the government’s inability consistently to explain their absence. However, when both the officer on the scene and the defense expert himself credibly testify to more than 100 plants, the logical inference is not that only fifty-four plants were seized, as Robinson would have us believe, but rather that only fifty-four have been preserved — with the balance being lost or destroyed. We thus follow the course we took in Belden , where we were presented with a similar “lost plants” claim and we declined to exclude evidence and reverse a conviction. Belden, 957 F.2d at 674 . The district court’s finding of 108 plants is supported by a preponderance of the evidence.

Robinson also argues that the district court erred in denying his motion for an evidentiary hearing concerning the discrepancy between Jopes’s and Sances’s testimony that there were more than 100 plants, and the DEA’s alleged possession of only fifty-four plants. We review for an abuse of discretion. United States v. Upshaw, 918 F.2d 789, 791 (9th Cir.1990), cert. denied, 499 U.S. 930 , 111 S.Ct. 1335 , 113 L.Ed.2d 266 (1991).

The district court denied Robinson’s request for an evidentiary hearing, basing its ruling on the following assessment of witness testimony and the arguments of counsel: “I’m satisfied [the missing plants] existed, and that they were seen repeatedly early on, but later on they aren’t there.” Transcript of Aug. 30, 1993, at 21. We first note that this finding was not clearly erroneous. See Hunt v. National Broadcasting Co., Inc., 872 F.2d 289, 292 (9th Cir.1989) (district court abuses its discretion if it rests decision on clearly erroneous finding of fact). Indeed, the district court’s interpretation seems to us the best way to harmonize the testimony of Jopes, Sanees, and Johnson. Secondly, the denial of an evidentiary hearing was not a clear error in judgment because an evidentiary hearing would have had little value. See United States v. Plainbull, 957 F.2d 724, 725 (9th Cir.1992) (reviewing court should not find abuse of discretion unless it has firm conviction that district court committed a clear error of judgment). On the one hand, if the plants had indeed vanished, an additional evidentiary hearing would have turned up no new evidence whatsoever. On the other hand, if the additional plants were discovered somewhere at the DEA facility, they would only support a conclusion which the district court had already properly reached — that more than 100 plants had been seized.

While the district court surely would have been within its discretion in inquiring into why the evidence was lost, it was not required to do so because that was not a dispositive issue in the case before it. See Transcript of Aug. 30,1993, at 21-22 (district court stated, “[t]his is not a case of trying the DEA for their administrative preservation policies. This is a case of growing marijuana. ..”). We conclude that the court did not abuse its discretion in denying Robinson’s request for an evidentiary hearing.

VI.

The district court erred in counting the 48 cuttings as marijuana plants. A preponderance of the evidence, however, supports the court’s finding that there were 108 countable mature plants, and it did not otherwise err in conducting Robinson’s sentencing. Because Robinson is still liable for more than 100 plants, his guidelines range will remain 63-78 months. And because the district court already sentenced him to the bottom of the guidelines range (63 months), reduction from 156 to 108 plants will have no effect on the length of his sentence. We therefore affirm the sentence, but remand the matter to the district court to correct the record to reflect that the sentence was based on the lower number of plants.

AFFIRMED; FOR CORRECTION OF RECORDS; REMANDED.

. The level decreases to 24 for 80-100 kilograms, and to 22 for 60-80 kilograms. Id. Had the district court adopted Robinson's favored method of counting plants, it would have had the option of giving him a sentence one to two years shorter. . The government now concedes that rootless cuttings should not be counted as plants at sentencing, but asks the panel not to reach this issue because even if the 48 cuttings are not counted, there are still more than 100 mature plants on which to base Robinson’s 63 month sentence. However, the ultimate issue in this case is whether a preponderance of the evidence supports the district court's finding that Robinson cultivated 156 plants. This necessarily involves inquiring whether the cuttings and the mature plants were correctly counted. Moreover, although the length of Robinson’s sentence would not change were he responsible for 108 instead of 156 plants, he might face other negative consequences in the future if his sentence is left unchanged. In any case, he has a right to have his record correctly reflect the number of plants for which he was sentenced. . Robinson implies that when plants share a common root system, they automatically only produce the yield of one plant. Surely, however, five plants growing in a ring, with a common root system, would produce more marijuana than one individual plant. Dr. Sance’s own testimony indicates as much. Tr. of Apr. 16, 1993, at 134 (growing multiple plants close together "produces a more bushy specimen”). . We adopt this rule as a rebuttable presumption. Thus, the rare plant that truly has two stems growing from a single (not intertwined) root system should not be counted as two plants. Nor should a single plant with multiple stalks be counted as multiple plants just because the point at which the stalks diverge happens to have been buried below the ground surface. . See United States v. Restrepo, 946 F.2d 654 (9th Cir.1991) (en banc) (preponderance standard applies to factual issues at sentencing), cert, denied, -U.S.-, 112 S.Ct. 1564 , 118 L.Ed.2d 211 (1992). . In Belden , moreover, we noted that absent bad faith, the "failure to preserve potentially useful evidence does not constitute a denial of due process of law.” Id. (citing Arizona v. Youngblood, 488 U.S. 51, 57-58 , 109 S.Ct. 333, 337 , 102 L.Ed.2d 281 (1988)).