j. RaptorRes. 34(4):299-304 ¸ 2000 The Raptor ResearchFoundation, Inc.

ARE NORTHERN SAW-WHET NOMADIC?

JEFFREYS. MARKS MontanaCooperative Wildlife Research Unit, Universityof Montana,Missoula, MT 59812 U.S.A.

JOHN H. DOREMUS U.S.Bureau of Land Management, Boise District, 3948 Development Avenue, Boise, ID 83705U.S.A.

ABSTRACT.---Thefirst knownnesting of a Northern Saw-whetOwl (Aegoliusacadicus) in the SnakeRiver Birdsof PreyNational Conservation Area occurredin a nestbox in 1986,4 yr after nestboxes were constructedin the studyarea. Occupancy of nestboxes by Northern Saw-whetOwls varied substantially overthe next 13 yr (0-8 nestsper yr). The numberof mice countedon nocturnalsurveys fluctuated widelyduring this same period, and the annualnumber of NorthernSaw-whet nestsin the boxes waspositively correlated with an indexof mouseabundance. Only one of the 52 breedingadults that we bandedwas recaptured in a subsequentyear, and none of the 139 nestlingsproduced in the boxes wasreencountered. A male that we bandedat a nestin April 1990 wasfound dead in BritishColumbia in January1993, more than 900 km NNW of our studyarea. Data from the BandingLaboratory were insufficientto evaluatebreeding-site fidelity because few researchershave banded adult Northern Saw-whet Owls at nests. Northern Saw-whet Owls seem to exhibit some of the characteristics associated with nomadismin (e.g., high fecundityand low survival),but they differ from typicalnomadic speciesbecause they do not specializeon cyclicprey. We suggestthat Northern Saw-whetOwls are nomadicin someparts of their range,settling to breed in areasof high food abundancethat they encounter during the nonbreedingseason. KEYWOP, DS: NorthernSaw-whet Owl; acadicus; nomadism; breeding biology; Idaho.

Es Aegoliusacadicus una especienomada? RES0MEN.----E1primer registro de un Aegoliusacadicus en anidacitnen el/treade conservaciondel Snake Riverocurri6 en una cajade anidacitnen 1986,4 aftosdespu•s que lascajas fueron construidasen el firea de estudio.La ocupacitnde las cajaspor los buhosvari6 substancialmenteen los prtximos 13 aftos.(0-8 nidospor afio). E1nfimero de ratonescontabilizados de nochefluctfio ampliamente durante estemismo periodo y el nfimeroanual de nidosde buhosen cajasfue positivamentecorrelacionado con el indice de abundanciade ratones. Solo uno de los 52 adultos en reproduccitn anillado, fue recapturadoy ninguno de los139 pichones producidos en lascajas de anidacitnfue reencontrado.Un machoque fue anilladoen el nidoen abrilde 1990fue encontradomuerto en ColumbiaBriffmica en enero de 1993, a mas de 900 km NNW, del area de estudio. Los datos del laboratorio de anillacion fueroninsuficientes para evaluar la fidelidadal territoriode reproduccitndebido a que pocosinvesti- gadoreshah anilladoadultos de Aegoliusacadicus en losnidos. Aegolius acadicus parece mostrar algunas de las caracteristicasasociadas con el nomadismoen las aves(e.g., alta fecundidady baja sobrevivencia), pero difierende las especiesntmadas tlpicas debido a que no se especializanen una presaciclica. Sugerimosque los buhos son ntmadas en partede surango, estableci•ndose para reproducirse en fireas de alta abundanciade cornidaque encuentrandurante la estacitn no reproductiva. [Traduccitn de C•sar Mgrquez]

The raptorsnesting in the SnakeRiver Birdsof Burrowing Owl (Athenecunicularia), Long-eared Prey National ConservationArea (NCA) in south- Owl (Asiootus), and Short-eared Owl (A. flammeus). western Idaho have been under intensive study The first three speciesare permanent residentsin since the mid-1970s (USDI 1979). Through 1985, the NCA, the Burrowing Owl is a typical migrant, six speciesof owlswere knownto nestin the area: and the two speciesof Asioare year-roundresidents (Tyto alba), Western Screech-Owl(Otus in someyears and migrantsin others (J.S. Marks kennicottii), Great Horned Owl ( Bubovirginianus) , and J.H. Doremus pers. obs.).

299 300 MARItS AND DOREMUS VOL. 34, NO. 4

The first known nest of a Northern Saw-whet suggestthat Northern Saw-whetOwls are nomadic Owl (Aegoliusacadicus) in the NCA occurred in a in southwestern Idaho. If we are correct, then our nest box in 1986, 4 yr after boxeswere constructed study birds would constitute the only known ex- in the study area. In marked contrast to the six ample of nomadism in a strigid that does not spe- speciesof owlsthat occurregularly in the NCA, the cialize on cyclicprey. presence of nesting Northern Saw-whetOwls var- STUDY AR•A AND METHODS ied substantiallyover the next 13 yr (0-8 nestsper yr). This variation in numbers, combined with a We studied breeding Northern Saw-whetOwls in the NCA in southwestern Idaho from 1986-99. The NCA is nearly complete lack of recapturesof adults,led us a shrubsteppedesert dominated by big sagebrush(Arte- to speculate that Northern Saw-whetOwls are no- misia tridentata).Compared with typical breeding habitat madic in the NCA. tbr Northern Saw-whet Owls (i.e., coniferous forest), Many speciesof birds exhibit strong breeding- trees are scarce and are confined to riparian areas and site fidelity, remaining in or returning to the same human settlements. All of the owls that we studied bred in nest boxesplaced in willows (Salix spp.), Russianolives breeding placesyear after year (Andersson1980). (Elaeagnusa%mstifolia), and black locusts(Robinia pseu- Site fidelity may be adaptivebecause it allowsin- doacacia).We set out boxes in pairs (1-40 m apart) be- dividualsto learn the best placesto feed, nest, and ginning in 1981. Since 1986, when Northern Saw-whet avoid predators, which in turn may enhance the Owls first nested in one of our boxes, two or more boxes have been availableat 95-47 siteseach year along 95 lin- ability of territory holders to attract mates (Hinde ear km of the Snake River and its tributaries. 1956, Greenwood and Harvey 1976, Greenwood To assessthe availability of small , we con- 1980). Nomadism (i.e., lack of breeding-site fidel- ducted nocturnal surveysin the NCA each spring from ity) in birds is much lesscommon than site fidelity 1984-94 (see Marks et al. 1989). With the aid of a spot- and tends to be restricted to speciesthat feed on light, observersin a slowly moving vehicle counted all "mice" seen along 547-709 km of secondary roads. cyclic prey or for which environmentalvariation Basedon thesesurveys, we calculatedan index of mouse (e.g., periodic rains) resultsin large fluctuationsin numbersby dividing the total number of mice seen by the availability of suitable breeding habitat (An- the total length of roads surveyedeach year.No surveys dersson 1980, 1981). were conducted after 1994. The classicexamples of nomadismin birds come We captured breeding female owls in nest boxesdur- from boreal seedeaters such as finches and cross- ing the brood-rearingperiod and caughtmales at night in mist nets set in front of the boxes. We determined the bills that move large distances in response to sex of adults by the presence (females) or absence changing availability of beechmast and (males) of an incubation patch. Adults were weighed, seeds (Newton 1972). In owls, nomadism is best measured, and banded at first capture, and nestlings were banded about a week before they left the nest. known in specialistssuch as Boreal Owls (Ae- We used partial correlation analysisto assessthe rela- goliusfunereus), Long-eared Owls, and Short-eared tionship between the mouse index (log transformed) Owls (Wallin and Andersson1981, Village 1987, and the number of owl nests in the boxes and to see Korpimfiki and Norrdahl 1991), although docu- whether the number of Northern Saw-whet Owl nests was correlated with the number of Western Screech-Owl mentation of the sameindividuals breeding in vast- nests.All testswere one-tailed becausewe predicted that ly different locationsis rare. Severalfemale Boreal numbers of nesting owls of both specieswould be posi- Owls have been captured at nestsmore than 500 tivelycorrelated with the mouse index. Similarly,we pre- km apart in different years (Wallin and Andersson dicted that the number of nesting Northern Saw-whet 1981, Korpimiiki et al. 1987), but the extent of no- Owls would be negativelycorrelated with the number of nestingWestern Screech-Owls in the boxesbecause West- madism in this speciesvaries widely among popu- ern Screech-Owlsare permanent residentsin the NCA, lations, and individuals may remain on the same and their body massis two to three times that of North- home ranges for two or more years in succession ern Saw-whetOwls (J.S. Marks andJ.H. Doremus unpubl. (Korpimfiki et al. 1987, Haywardet al. 1993). In data). short, comparedwith "classic"nomadism exhibit- RESULTS ed by boreal seedeaters, the owl noted above seem to be "periodically" nomadic. The number of Northern Saw-whet Owl nests in The Northern Saw-whet Owl is common in for- our boxes varied considerablyamong years,rang- ested habitats across the northern United States ing from zero to eight (• = 2.7 ñ 3.09 [ñSD]) and southern Canada. Despite its abundance, rel- (Table 1). In contrast, Western Screech-Owls nest- atively little is known about its breeding biology ed in the boxes every year (range 4-13), and the (Cannings 1993). In this paper, we present data to number of nests per year was larger (• = 8.8 ñ DECEMBER 2000 NOMADISM IN NOR•rHEaN SAW-WHET OWLS 301

Table 1. Number of Northern Saw-whet Owl and West- 10 ern Screech-Owl nests in boxes in the Snake River Birds of Prey National ConservationArea, 1984-99. Pairs of • 8 nest boxes were availableat 25-47 siteseach year. The number of adult Northern Saw-whetOwls captured at nests each year is in parentheses.

NORTHERN WESTERN SAW-WHET SCREECH- OWL OWL

1984 0 7 ß ß

1985 0 5 ß ß ß ß ß ß 1986 1 5 1987 7 (11) 8 2 4 6 8 10 12 14 1988 0 4 NO. of screech-owl nests 1989 0 6 1990 8 (11) 9 Figure 1. Relationship between the number of North- 1991 6 (10) 11 ern Saw-whet Owl nests in boxes and the number of West- 1992 7 (5) 13 ern Screech-Owlnests in boxes,Snake River Birds of Prey 1993 1 13 National ConservationArea, 1984-99. Only 15 points are 1994 0 8 shown because two were identical. 1995 5 (6) 13 1996 3 (4) 11 1997 0 8 number of Western Screech-Owl nests in the boxes 1998 0 9 (partial r = -0.28, N = 11, P = 0.21; Fig. 2b). 1999 5 (5) 10 We caught 52 adult Northern Saw-whetOwls (29 females, 23 males) at the 43 nests that we moni- tored (Table 1). Only one returned to breed in a subsequentyear, a female that nested in boxes 360 2.93) and less variable than that of Northern Saw- m apart in 1990 and 1991. A male that bred suc- whet Owls (Table 1). Rather than the negative re- cessfullyin the studyarea in 1990 wasfound freshly lationship that we expected,the number of North- ern Saw-whetOwl nestsin the boxeswas positively Table 2. Resultsof nocturnal spotlight surveysfor small correlated with the number of Western Screech- mammalsin the Snake River Birds of Prey National Con- Owl nestsin the boxes each year (partial r = 0.58, servation Area, 1984-94 (no surveyswere conducted af- N = 16, P -- 0.038; Fig. 1). Indeed, in severalcases ter 1994). Northern Saw-whet Owls nested in boxes within oc-

cupied Western Screech-Owlterritories. Thus, the SURVEY presence of Western Screech-Owlsappeared to LENGTH have no negative effect on yearly fluctuations in YEAR MICEa (km) MICE/km the number of Northern Saw-whet Owl nests in the 1984 2 547 0.004 boxes. 1985 30 547 0.055 The number of mice counted during nocturnal 1986 4 547 0.007 surveysalso varied substantiallyfrom year to year 1987 293 612 0.479 (Table 2). The yearswith the highest mouse num- 1988 20 612 0.033 bers tended to coincide with the highest numbers 1989 43 709 0.061 1990 146 700 0.208 of nesting Northern Saw-whetOwls in the boxes 1991 24 604 0.040 (Tables 1, 2), and the number of Northern Saw- 1992 47 603 0.078 whet Owl nestswas positivelycorrelated with the 1993 15 660 0.023 mouseindex during the yearsthat we had data on 1994 21 659 0.032 small mammals (partial r = 0.69, N = 11, P = Total number counted on surveyseach year. "Mice" includes 0.013; Fig. 2a). In contrast,no significantcorrela- Perognathusparvus, Onychomysleucogaster, Peromyscus maniculatus, tion existed between the mouse index and the Reithrodontomysmegalotis, and Microtusmontanu& 302 MARItS AND DOREMUS VOL. 34, NO. 4

typicallyremains there for life). Western Screech- (a) Owl nests were not confined to boxes; some oc- curred in natural cavities in trees and cliffs, and one pair nestedin an old Black-billedMagpie (Pica hudsonia)nest (J.S. Marks and J.H. Doremus pers. obs., Marks 1983). We have never found a North- ern Saw-whetOwl nest anywhere in the NCA ex- cept in a nest box, nor have we heard males sing- ing from tree grovesthat did not contain a box. Thus, we suspectthat few Northern Saw-whetOwls ßß nest in sites other than our boxes.

DISCUSSION

0.6 0.0 0.1 0.2 0.3 0.4 0.5 The Northern Saw-whetOwls that we studiedap- peared to be nomadic.Turnover among breeding 14 adults was high, and no juveniles were known to (b) have returned to the studyarea to breed. The best • 12 evidence for nomadism would be the capture of marked individuals at widely separatedbreeding • lO sites in different years. Such evidence is absent from banding records, although we note that the

ß banding data are not ideal for assessingbreeding- site fidelity. ß Our review of the 1276 reencounters (through o 6 ß August 1999) of Northern Saw-whetOwls in the z ß ß database of the USGS Bird Banding Laboratory ß showed that less than 3% were of birds banded during the breeding season,and we found no re- 0.0 0.• 0.2 0.3 0.4 0.5 0.6 cords of birds banded at a nest in one year and No. of mice per km recaptured at a distantsite in a subsequentbreed- ing season.Indeed, asidefrom the instancewe doc- Figure 2. Relationship between the mouse index and umented in the NCA, the only known casesof the number of Northern Saw-whet Owl nests in boxes (a) and the number of Western Screech-Owl nests in boxes breeding-sitefidelity in Northern Saw-whetOwls have come from British Columbia, where 5 of 36 (b), Snake River Birds of Prey National Conservation Area, 1984-94. breeding adults (two females,three males) banded over an 8-yr period returned to the same or adja- cent territories in subsequent years (Cannings dead in British Columbia in January 1993, approx- 1993). imately920 km NNW of the NCA. No other adults Predatorybirds that exhibit nomadismtypically that we banded in the study area have been reen- feed on cyclicprey, and in theory they shouldpro- countered, nor have we reencountered any of the duce large clutch sizes,reproduce at one year of 139 nestlingsproduced in the boxes. Nesting suc- age, and have high juvenile survivaland low adult cesswas rather high, with 29 of 42 nests(69%; one survival (Andersson 1980). Northern Saw-whet nest was not monitored completely) producing Owls in the NCA have high fecundity (typical one or more young that survivedto leave the nest clutchescontain 6-7 eggs,which is at the high end (i = 3.3 young per nesting attempt and 4.8 per of the range of clutch sizesfor the continent; Can- successful nest). nings 1993), but they do not specializeon cyclic In contrast to the situation with Northern Saw- prey. Rather, their diet consistsof a mixture of whet Owls, we made hundreds of recaptures of house mice (Mus musculus),harvest mice (Reithro- adult Western Screech-Owls,which occupied their dontomysmegalotis), montane (Microtusmon- breeding territories year-round (indeed, once a tanus), and deer mice (Peromyscusmaniculatus) Western Screech-Owl settleson a breeding site, it (Marks and Doremus1988, Rains 1997). The abun- DECEMBER 2000 NOMADISM IN NORTIqERN S^W-WHET OWLS 303 dance of these prey speciesin the NCA varies un- birds had been captured outside of the NCA in predictably rather than in a cyclic pattern. Else- subsequentbreeding seasons,two alternative hy- where in the range of Northern Saw-whet Owls, potheses could explain the apparent lack of site vole populations can be cyclic. However, voles sel- fidelity that we observed.First, the NCA may be a dom comprise a major portion of the diet during "sink" in which Northern Saw-whet Owls have low the breeding season (Marks and Doremus 1988, annual survivalrelative to those nesting elsewhere. Swengel and Swengel 1992, Cannings 1993). We have no way to test this hypothesis,but the Northern Saw-whet Owls probably reproduce at presenceof one of our malesin Canada three years one year of age, but the age of first breeding is not after breeding in the NCA, and the apparently known for wild birds (Cannings1993). In addition, high reproductive successof Northern Saw-whet little is known about annual survivorshipof adults Owls in our studyarea (see Cannings 1993), argue or juveniles, although adult survivalappears to be againstthe notion that these birds constitutea sink low (ca. 50%; Cannings 1993). Thus, Northern population. Second, we may have failed to detect Saw-whet Owls exhibit some of the characteristics adults that returned to breed in the NCA because of nomadic species,but they differ fundamentally they used natural cavities.We have not conducted from typical nomads in that they do not specialize systematicsurveys of tree cavitiesthroughout the on cyclic prey. NCA, but as noted above, we have never heard Nomadism in owls has been documented most Northern Saw-whetOwls singing in areasthat did thoroughly in the (e.g., Wallin and An- not contain nest boxes. We suspectthat few owls dersson1981, L6fgren et al. 1986, Korpimfki et al. nest in natural sites relative to the number that use 1987). Male Boreal Owls in Fennoscandia tend to our boxes. Nonetheless,it is possiblethat we have remain on their territories year-round, whereas fe- missed some nests in natural sites and that site te- malesare more likely to dispersebetween breeding nacity by Northern Saw-whetOwls is more preva- sites,especially during lows in the vole cycle (L6f- lent in the NCA than we believe. gren et al. 1986, Korpimfki et al. 1987). Nomadism If we are correct in our assertion that Northern and site fidelity have been documented in both Saw-whet Owls are nomadic in southwestern Idaho, sexesof Boreal Owls in Idaho, but the tendency then these birds would constitute the first known toward nomadismis strongestin females (Hayward case of nomadism in a speciesof owl that does not et al. 1993:33-35). Thus, unlike the situation in the specializeon cyclic prey. Given the apparent geo- NCA, where both sexes of Northern Saw-whet Owls graphic variation in nomadism exhibited by Boreal almost never display site fidelity, nomadism in Bob Owls (Korpimfki et al. 1987), it is likely that the real Owls is confined mostly to females and occurs incidence of nomadism in Northern Saw-whet in some years but not others. Owlsvaries geographically. The shrubsteppedesert Aside from our study population, the only pre- of southwesternIdaho is not typical breeding hab- vious hints that Northern Saw-whet Owls are no- itat for Northern Saw-whet Owls. Indeed, without madic have come from studies of vocal activity in the presence of nest boxes, we suspect that few Colorado (Palmer 1987) and Wisconsin(Swengel Northern Saw-whetOwls would breed in our study and Swengel 1995). In the Colorado study,annual area. Northern Saw-whetOwls have probably win- changes in the number of singing birds corre- tered in the NCA for many years, but the avail- sponded with changesin the numbers of voles in ability of nest boxes has only recently made the the area. In Wisconsin,nightly surveysconducted area suitablefor nesting.We suggestthat Northern in late winter and early spring over a 10-yr period Saw-whetOwls are nomadic in some parts of their revealed a fairly strong 4-yr "cycle" in the amount range, settling to breed in the same areasin which of singing by adults. Swengeland Swengel (1995) they winter or migrate during years when food noted that the vole cyclewas a possibleexplanation availabilityis high. Long-term banding efforts at for the pattern in vocal activitythat they observed. nests,coupled with monitoring of prey availability, Neither study involved finding nestsand trapping will be necessaryto thoroughly addressthe ques- adults, however, so it is difficult to determine the tion of nomadism in Northern Saw-whet Owls. extent to which changesin vocal activitytranslate to actual changes in the presence of breeding ACKNOWLEDGMENTS birds. We thank L. Singer, C. Rains, M. McCoy, and V. Saab Lacking the proof that would be provided if our for field assistance;S. Knick for providing data on small 304 MAR• AND DOREMUS VOL. 34, NO. 4 mammalsfrom the early 1990s;M. Kochert and K. Steen- goliusfunereus(L.)) in relation to cyclicfood produc- hof for technical support; G. Hayward, C. Marti, and S. tion. Oecologia69:321-326. Swengelfor commentson the manuscript;and T. Martin MARKS,J.S. 1983. Unusual nest sitesof a Western Screech- for suggestingthe partial correlation analysis.We also Owl and an American Kestrel. Murrelet 64:96-97. thank the USGS Bird Banding Laboratoryfor banding -- ANDJ.H. DOREMUS.1988. Breeding-seasondiet of data, D. Orcutt (Idaho Department of Fish and Game) Northern Saw-whet Owls in southwestern Idaho. Wil- for allowing accessto the C.J. Strike Wildlife Manage- son Bull. 100:690-694. ment Area, and the Boise District Bureau of Land Man- agement for logisticalsupport. , --, ANDRJ. CANNINGS.1989. Polygynyin the Northern Saw-whet Owl. Auk 106:732-734. LITERATURE CITED NEWTON,I. 1972. Finches. Collins, London, U.K. PALME}•,D.A. 1987.Annual, seasonal, and nightlyvaria- ANDE}•SSON,M. 1980. Nomadism and site tenacity as al- tion in calling activityof Boreal and Northern Saw- ternative reproductive tacticsin birds. J. Anim. Ecol. whet owls.Pages 162-168 in R.W.Nero, RJ. Clark,RJ. 49:175-184. Knapton, and R.H. Hamre [EDS.], Biology and con- 1981. Reproductive tactics of the Long-tailed servationof northern forest owls: symposiumpro- Skua Stercorariuslongicaudus. Oikos 37:287-294. ceedings.USDA For. Serv. Gen. Tech. Rep. RM-142, CANNINGS,RJ. 1993. Northern Saw-whetOwl (Aegolius Fort Collins, CO U.S.A. acadicus). In A. Poole and F. Gill [EDS.], The birds of RAINS,C. 1997. Comparisonof food habitsof the North- North America, No. 42. Academyof Natural Sciences, ern Saw-whetOwl (Aegoliusacadicus) and the Western Philadelphia, PA, and American Ornithologists' Screech-Owl (Otus kennicottii)in southwesternIdaho Union, Washington, DC U.S.A. Pages 339-346 in J.R. Duncan, D.H. Johnson, and GREENWOOD,P.J. 1980. Mating systems,philopatry, and T.H. Nicholls [EDS.],Biology and conservationof owls dispersal in birds and mammals. Anim. Behav. 28: of the Northern Hemisphere. USDA For. Serv. Gen 1140-1162. Tech. Rep. NC-190, St. Paul, MN U.S.A. --AND P.H. HARVEY.1976. The adaptivesignificance SWENGEL,A.B. ANDS.R. SWENGEL. 1995. Possiblefour-year of variation in breeding area fidelity of the Blackbird cycle in amount of calling by Northern Saw-whet ( Turdusmerula L.). J. Anim. Ecol.45:887-898. Owls.Passenger Pigeon 57:149-155. HAYWARD,G.D., P.H. HAYWARD,AND E.O. GARTON. 1993. SWENGEL, S.R. AND A.B. SWENGEL. 1992. Diet of Northern Saw-whet Owls in southern Wisconsin. Condor94:707- Ecologyof Boreal Owls in the northern RockyMoun- 711. tains, U.S.A. Wildl. Monogm124:1-59. U.S. DEPARTMENT OF THE INTEPdO}•. 1979. Snake River HINDE, R.A. 1956. The biological significanceof the ter- ritories of birds. Ibis 98:340-369. Bird of Prey SpecialResearch Report. Bureau of Land Management, Boise, ID U.S.A. KORPIM•KI, E. AND K. NORRDAHL. 1991. Numerical and VILLAGE,A. 1987. Numbers,territory size,and turnover functional responsesof kestrels,Short-eared Owls and of Short-earedOwls Asioflammeus in relation to vole Long-earedOwls to vole densities.Ecology 72:814-826. abundance. Ornis Scand. 18:198-204. , M. LAGERSTROM,AND P. SAU}•OLA.1987. Field ev- WALLIN, K. AND M. ANDERSSON.1981. Adult nomadism in idencefor nomadismin Tengmalm'sOwl Aegoliusfu- Tengmalm'sOwl Aegoliusfunereus. Ornis Scan&12' nereus. Ornis Scan& 18:1-4 125-126. LOFGREN, O., B. H(SRNFELDT,AND B.-G. CAmpSSON.1986. Site tenacity and nomadismin Tengmalm's Owl (Ae- Received4 March 2000; accepted22 July 2000